PROCEEDINGS
ICDER - 2014 INTERNATIONAL CONFERENCE ON DEVELOPMENTS IN ENGINEERING RESEARCH
Sponsored By INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT ((Registered Under Indian Trust Act, 1882)
Technical Program 8TH October, 2014 Hotel Ramee Guestline, Tirupati
Organized By INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
www.iaetsd.in
Copyright Š 2014 by IAETSD All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written consent of the publisher.
ISBN: 378 - 26 - 138420 - 9
http://www.iaetsd.in
Proceedings preparation, editing and printing are sponsored by INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT COMPANY
About IAETSD: The International Association of Engineering and Technology for Skill Development (IAETSD) is a Professional and non-profit conference organizing company devoted to promoting social, economic, and technical advancements around the world by conducting international academic conferences in various Engineering fields around the world. IAETSD organizes multidisciplinary conferences for academics and professionals in the fields of Engineering. In order to strengthen the skill development of the students IAETSD has established. IAETSD is a meeting place where Engineering students can share their views, ideas, can improve their technical knowledge, can develop their skills and for presenting and discussing recent trends in advanced technologies, new educational environments and innovative technology learning ideas. The intention of IAETSD is to expand the knowledge beyond the boundaries by joining the hands with students, researchers, academics and industrialists etc, to explore the technical knowledge all over the world, to publish proceedings. IAETSD offers opportunities to learning professionals for the exploration of problems from many disciplines of various Engineering fields to discover innovative solutions to implement innovative ideas. IAETSD aimed to promote upcoming trends in Engineering.
About ICDER: The aim objective of ICCTER is to present the latest research and results of scientists related to all engineering departments’ topics. This conference provides opportunities for the different areas delegates to exchange new ideas and application experiences face to face, to establish business or research relations and to find global partners for future collaboration. We hope that the conference results constituted significant contribution to the knowledge in these up to date scientific field. The organizing committee of conference is pleased to invite prospective authors to submit their original manuscripts to ICDER 2014. All full paper submissions will be peer reviewed and evaluated based on originality, technical and/or research content/depth, correctness, relevance to conference, contributions, and readability. The conference will be held every year to make it an ideal platform for people to share views and experiences in current trending technologies in the related areas.
Conference Advisory Committee:
Dr. P Paramasivam, NUS, Singapore Dr. Ganapathy Kumar, Nanometrics, USA Mr. Vikram Subramanian, Oracle Public cloud Dr. Michal Wozniak, Wroclaw University of Technology, Dr. Saqib Saeed, Bahria University, Mr. Elamurugan Vaiyapuri, tarkaSys, California Mr. N M Bhaskar, Micron Asia, Singapore Dr. Mohammed Yeasin, University of Memphis Dr. Ahmed Zohaa, Brunel university Kenneth Sundarraj, University of Malaysia Dr. Heba Ahmed Hassan, Dhofar University, Dr. Mohammed Atiquzzaman, University of Oklahoma, Dr. Sattar Aboud, Middle East University, Dr. S Lakshmi, Oman University
Conference Chairs and Review committee:
Dr. Shanti Swaroop, Professor IIT Madras Dr. G Bhuvaneshwari, Professor, IIT, Delhi Dr. Krishna Vasudevan, Professor, IIT Madras Dr.G.V.Uma, Professor, Anna University Dr. S Muttan, Professor, Anna University Dr. R P Kumudini Devi, Professor, Anna University Dr. M Ramalingam, Director (IRS) Dr. N K Ambujam, Director (CWR), Anna University Dr. Bhaskaran, Professor, NIT, Trichy Dr. Pabitra Mohan Khilar, Associate Prof, NIT, Rourkela Dr. V Ramalingam, Professor, Dr.P.Mallikka, Professor, NITTTR, Taramani Dr. E S M Suresh, Professor, NITTTR, Chennai Dr. Gomathi Nayagam, Director CWET, Chennai Prof. S Karthikeyan, VIT, Vellore Dr. H C Nagaraj, Principal, NIMET, Bengaluru Dr. K Sivakumar, Associate Director, CTS. Dr. Tarun Chandroyadulu, Research Associate, NAS
ICDER - 2014 CONTENTS 1
A Layered Security Approach through FEMTO CELL using Onion Routing in MANET
1
2
A Survey on Enroute Filtering Scheme in Wireless Sensor Networks
7
3
Estimation of Damping Torque for Small-Signal Stability of Single Machine Infinite Bus System
13
4
An Enhanced Feature Selection for High-Dimensional Knowledge
22
5
Improving the Location of Nodes in Wireless Ad Hoc and Sensor Networks Using Improvised LAL Approach
32
6
Survey on Wireless Sensor Networks Routing Protocols Based on Energy Efficiency
38
7
AN EFFECTIVE ALARMING MODEL FOR DANGER AND ACTIVITY MONITORING USING WEARABLE SENSORS FOR CHILDREN
45
8
An Efficient Way of Classifying and Clustering Documents Based on SMTP
49
9
An Efficient Way of Detecting a Numbers in Car License Plate Using Genetic Algorithms
54
10
RF Controlled Sailing Robot for Oceanic Missions
58
11
Solar Power Satellites
60
12
Autonomous pick and place rover for long distance surveillance using ultrasonic sensors
66
13
Intelligent Bus Alert System For Blind Passengers
71
14
Traffic Sign Recognition for Advanced Driver Assistance System Using PCA
78
15
Ethernet Based Intelligent Security System
84
16
Modified Artificial Potential Fields Algorithm for Mobile Robot Path Planning
93
17
Protecting Privacy Preserving For Cost Effective Adaptive Actions Using Memitic Algorithm
98
18
Use of Mobile Communication in Data-Intensive Wireless Networks
105
19
Confidential Multiparty Computation with Anonymous ID Assignment using Central Authority
111
20
DESIGN OF RF BASED VOICE-CONTROLLED MULTI-TERRAIN ROBOT TO TRAVEL ON UNEVEN SURFACES
117
21
Zigbee For Vehicular Communication Systems
122
22
An Efficient and Large data base using Subset Selection Algorithm for Multidimensional Data Extraction
129
23
A Review on ECG Arrhythmia Detection based on DD-DW Transformation
136
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
A Layered Security Approach through FEMTO CELL using Onion Routing in MANET T. Sathish1, M.Senthil Kumar2 1
Department of Computer Science and Engineering, M.E. Scholar, Anna University, Sree Sowdambika College of Engineering, Aruppukottai, TamilNadu, India. princeofpersia123555@gmail.com
2
Department of Computer Science and Engineering, Assistant Professor, Anna University, Sree Sowdambika College of Engineering, Aruppukottai, TamilNadu,India. rmsenthik@gmail.com
Abstract— In earlier process, making a secure routing only discussed and never discuss about how to transfer data in a secured manner. Even though we performed routing in a secured manner, there will be chances of data should be dropped or revealed by an illegal persons. We use an onion routing to make a highly secured routing, so this routing includes the mechanism of layered by layered approach from one node to another node’s. And we transfer the data in secured manner by sending dummy packets from source to destination, and these dummy packets are mold up by the mechanism of node characterization technique. And in earlier process they never looked for time and speed reduction. That is, in existing system, they use the concept of secured routing each and every time when data will be sent. So this will increase the speed reduction process which is at every moment of transaction we need to perform separate routing, to overcome this we proposed onion routing and node characterization. And we using femto cell device for strengthening the signal when there is no sufficient signal to work on the process. Keywords— MANETs, Routing Protocols, Secure Onion Routing, Layered Approach, Packet authentication, Femto cell, NS2.
I. INTRODUCTION 1.1 MANET The term MANET refers to Mobile Ad-hoc NETwork. MANET is a less infrastructure network; nodes are under mobility. It should be moved here and there and it won’t rely stable. Mobile Ad-hoc networks are most familiar to security issues due to the characteristics of such networks such as a wireless medium and dynamic topology. It is very harder to provide trusted and secure communications in enemy environments such as battle fields. On one hand, the malicious persons outside the network may have an idea to reveal the information about the communicating nodes, even when communications are encrypted. On the other hand nodes involved inside network will always be trusted, since a trusted node may be hold by illegal person and become malicious. In MANET, a set of interacting nodes should cooperatively implement routing functions to enable end-to-end communication along dynamic paths composed by multi-
ISBN NO : 378 - 26 - 13840 - 9
hop wireless links. Several multi-hop routing protocols have been proposed for MANET, and most popular ones include: Dynamic Source Routing (DSR), Optimized Link-State Routing (OLSR), Destination-Sequenced Distance-Vector (DSDV) and Ad Hoc On-Demand Distance Vector (AODV). Most of these protocols are do their function on assumption of trusted manner. But sometimes it could not been trusted it behaves like secured less, because when there is a presence of malicious node at that time it emerges the weakness of MANET to cause various kinds of attacks. 1.2 ADHOC Ad-hoc is a Latin word which refers to “all purpose”. For example, take two access points as access point 1 and access point 2. There will be twenty kilometer distance in between those two access point.
Access Point 1 20 k Access Point 2 If user A starts to download the data from access point 1, immediately user A tends to travel from access point 1 to access point 2. Now download is under half fulfilled. When user A which comes under out of coverage from access point 1 due to travelling. They face a problem about downloading. Due to these criteria, we go for intermediate nodes concept to prevent interrupts which occurred in data transfer. So, that time the concept of Ad-hoc is established i.e. intermediate node is formed. Now data could be transformed very successfully but there will be a problem occurred under security concept. Because now the job is hold under intermediate nodes, we does not known details about intermediate nodes i.e. either it is a legal or illegal. When it is illegal, data will not been under secured way. So the major cause of this paper is about, Done secure routing Along with that the data transfer also will be in an secured manner The remainder of this paper is organized as follows. The basic concepts and routing processes are analyzed in Section II. The protocol design is presented in Section III. The protocol evaluation is discussed in Section IV. Performance
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 1
Proceedings of International Conference on Developments in Engineering Research
analysis is evaluated in Section V and Section VII concludes this paper. The future scope is conferred in Section VII.
www.iaetsd.in
II. BACKGROUND AND RELATED WORK
persons do not have their identity number. So step by step, it travels from source node to destination node which comes across through different intermediate nodes. The nature cause of this mechanism is to set a route request message from source to destination node.
In existing system, they tell about how to perform a secure routing. They choose the main concept of secure routing as onion routing. Why they named as secured means, how an onion makes different layers when we start to cut it into small pieces, likewise here also they we forms different layers for checking each and every nodes as an one by one. It is like a layered i.e. what is the action handled by two nodes as a one layered. So we can name it’s for our understanding as onion routing is also known as layered approach.
2.2 ONION ROUTING Once a route request message is perfectly reached by destination node, the destination node again they start a work about route reply. The ROUTE REPLY is a main process for this mechanism, by the mechanism using in my project is about onion routing. In setting a trap, the route request message is starts from source node to destination node. But here in route reply it starts from destination node to source node.
Layer 1
Layer 3
Layer 2
Source
Intermediate Nodes
Destination
WORKING PROCESS OF ONION ROUTING It is a simple process here we are going to removing the identity number of each intermediate nodes from destination end points to source end points(reverse process).
Fig.2.1 Layered approach The main concept done in existing system is they did how to do secure routing for providing path. This will be discussed as follows. To accommodate secure rating these cases requires three kinds of techniques, they are: Setting a trap Routing by means of layered approach(onion routing) Signature process 2.1 SETTING A TRAP The general meaning of trap is to find out an unidentified or illegal person by without knowing an alert for the person whom is involved. The main concept involved in this technique is as follows. Let consider, there are six nodes from source to destination.
Source
Intermediate Nodes SID
DID
Destination
Encrypted Message
I1
I2
Fig.2.2 Process of Onion Routing In the above diagram route reply request is send from destination node and its moved to the intermediate node. Here how they will choose intermediate node means that the path which the route request came across. So its starts from the intermediate node first its goes for last intermediate node. In that it removes the top of the key which is present. Intermediate Node SID
Source
Destination
To set a “trap” separate id is given to each node. The id should also been included for source and destination node too. For Source Node,
SID
DID
Encrypted Message
DID
Encrypted Message
I1
I2
From this diagram, I2 will be removed. This is a top of the key. Once it is finished then it will be moved for another node. This process is known as layered approach and finally its came to source node and delivered the route reply message. 2.3 SIGNATURE PROCESS In signature process, we are going to generate group signature and along with that session key. The main process key is to finish their process with in their time limits which we set for that.
I1 identity “id” for Node 1 So, these processes are going on until data or message reaches the destination node. By setting an identity for each node ‘I’ can able to find the illegal persons, because illegal
ISBN NO : 378 - 26 - 13840 - 9
SIO
DIO
Encrypted Message
Source Sign
I1 Sign
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 2
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
For example, the session number=5 packets means, there should be five packets could be sent at a time. If there is exceeding of any packet count there will be some jobs did by an illegal person
When the first dummy packets are in Node ‘A’ that is under source node it holds the information of,
III. PROTOCOL DESIGN
0th node Here sequence number provides some identity number for first packet as for example: ‘1’. And random number also has been provided. After that padding will be stored as 0th position because it’s a first position of the node.
We invent a much effective method to make an protection under both packet dropper and modifier’s, in this method let’s take upon routing tree which is at sink it will be accomplished very first. At the time of sensor data are make an survey along the tree structure towards the sink, every packet sender adds upon small number of extra bits to it, which is also known as packet marker’s to the packet. And then runs up the node characterization algorithm for identifying node’s which hold packet dropper and modifier. This is the one of the way to identify the bad nodes. 3.1 NODE CHARACTERIZATION ALGORITHM This algorithm includes the following methods. They are, o Stepwise ranking method o Global wise method o Hybrid ranking method Using this algorithm, we are going to send dummy packet with sequence number, random number and padding. Sequence number is for the packet which we are going to sent. Random number for the node, here why we choose random number for node means, to find out the hacker which is either it was good node or bad node. Padding shows the position of the node. If it is a first node means, it shows the position as 0th node (zeroth node). SID
DID
Sequence Number
Random Number
Padding
Source Intermediate Nodes Destination Fig.3.1 Parameters of Node Characterization Algorithm For example, I need to transfer 50 kp packets means, I split as 10 kp as first packet and 20 kp as second packet and 20 kp as third packet and it is an automated one. From the above diagram, first packet of node includes, SID
DID
Sequence Number
Random Number
Padding
This information will be in the source node only and this is a general concept.
Source
ISBN NO : 378 - 26 - 13840 - 9
B
C
DID
Sequence Number
Random Number
Padding
Step 2: Similarly the same first packet will be moved from Node A to Node B. The Node B holds the information like, SID
DID
Sequence Number
Intermediate Random Number
Intermediate Padding
1st Node When same first packet will be moved from Node A to Node B, it should contains same sequence number, and also been with source random number id, along with that new intermediate node random number, and finally intermediate node padding position is 1st Node. A C Source
B
C
D Destination
(Now, we are in Node B) Step 3: The same first packets move from Node B to Node C. For example, we found that we have packet modification. So, this packet modification shows that the presence of changed packet information. That is an anonymous person could modify the data. For example, how can we find our data will be modified means? 1. There will be changed in sequence number example we use ‘1’ as an sequence number means these will be changed as 1.4 or 1.7 which what we given it won’t present here. 2. The random number should not been generated for illegal person. Step 4: If we are going to send second packet means, the sequence number will be changed as ‘2’ then its move from Node A , Node B, Node C. General Rule: o First packet ‘1’ sequence number o Second packet ‘2’ sequence number o Third packet ‘3’ sequence number IV. PROTOCOL EVALUATION
PROCESS WHICH ARE GOING Step 1: A
SID
D
The proposed system has the following features: Identifies both a kind of malicious activity like packet dropping and modifying. Achieving low communication and overheads.
Destination
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 3
Proceedings of International Conference on Developments in Engineering Research
Much more suitable for existing false packet filtering schemes that is this scheme will make an end point for malicious packets. We going to deployment large number of sensor nodes in an two dimensional area. The nature job of each sensor node’s generate the data regularly and intended to forward packets towards an sink and the sink is located inside the network. A. STEP WISE RANKING METHOD Yet now we find the one will be safer node. So, now we going to note which node will be good node or which will be bad node. So for finding, I will just make all nodes under tree formation. That is the nodes which are nearby. A
www.iaetsd.in
either bad or good. If we reject these nodes, their child node also deleted and it is a drawback of this method. B. GLOBAL WISE RANKING METHOD To overcome the drawback which are comes under stepwise ranking method we need to seen these node under over all view that is which one will be best. By selecting that best one, we are going to use the best one for our process that is for packet transaction. C. HYBRID RANKING METHOD These methods are combination of stepwise ranking method and global ranking method. Here the node which is having highest priority will choose as first and it is most likely bad node.
D. FEMTO CELL Femto cell is an device which act as wireless access point , its used to strength the signal and to provide the same signal range whatever its came from the core network and the range of femto cell process about 10 metres. These femto cell are particularly designed for small environment, so this G D E F theme got correlated with my project which is after selecting an particular route for my data transmission here I’ll have some less amount of node’s , so the data starts to travel across the node which I framed. The main problem is node H I J K L M N O coverage will be differs according to distance of each node from the core network’s and this leads to higher time consumption in data transfer. So when we use femto cell Fig.4.1 Tree Structure of Nodes Formation near to node which does not have fulfilled coverage means For example, in node A we have 10 packets, now we going we can able to successful in over-come these drawback. to check of these whether this remaining nodes are good node or bad node. Node A sent 10 packets to every nodes which are present in the tree (b,c,d,e,f,g,h,i,j,k,l,m,n,o). B
C
A B
D
H
9/10
I
C
10/10
E
J
F
7/10
20/10
G
K
Fig.4.2 Step Wise Ranking Method Good Node – 10/10 Bad Node – 20/10, 8/10, 7/10 Unknown Node (Either Bad or Good) – 9/10,11/10 By using this, we may find either which one will be good node or which one will be bad node. It is a step wise ranking method. We are going to see step by step which is from top layer of the tree. When we seen, Node B as 10/10 means it will be good node and it will be selected. Similarly, take Node ‘D’, it has 9/10 so it is under the scenario of unknown
ISBN NO : 378 - 26 - 13840 - 9
V. PERFORMANCE ANALYSIS 5.1 NETWORK CONFIGURATIONS TOPOLOGIES In our research the network area is about 1500m x 600m with 80 nodes normally and it should be equally distributed. Here the function of distributed coordination IEEE 802.11 is used as the MAC layer. And the capacity of channel is about
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 4
Proceedings of International Conference on Developments in Engineering Research
ATTACKING MODELS The general assumption about the intermediate nodes along the route may become malicious If there is any malicious node then the routing packets are randomly dropped. For example in previous routing like ANODR and AODV will suffer more packet loss. The AASR detect the malicious node by using the signature method and find out the details of attackers in routing table.
350 Data Packet Delay (ms)
3Mbps, transmission range is occurred was 250m.The mobility in node should be there by using random way point model. The total of 20 UDP based CBR sessions are used to create the network traffic.
www.iaetsd.in
300 250 200 150
AODV
100
SR ANODR
50 0 1
2
3
4
5
Mean Node Speed (m/s)
SIMULATION RESULTS
c.
End-to-end Delay
Effects of Mobility Scenario
Average Throughput Per Flow (kbps)
16 15.5 15 14.5 14 13.5 13
AODV SR 1
2
3
4
ANODR
5
Effects of Malicious Attacks The configuration of mobile network with an average as 4m=s, therefore in general the number of malicious nodes increases similarly the throughput decreases. The following figures a, b, c show the performance in the presence of different number of malicious nodes Average Throughput Per Flow (kbps)
To simulate the enemy environment, we are going to choose twenty percentage of total nodes, which is ten nodes as malicious nodes and then we can able to change the network mobility from one to eight m=s and record the performance results. From the results the average nodal speed increases, similarly the throughput also varies because nodes have an capability of move randomly. Due to performance variation secure routing always achieve highest throughput. The following figures a, b, c describe the performance of different mobility settings.
AODV SR ANODR 1
3
5
7
9
Number of Malicious Node
Mean Node Speed (m/s) a.
d.
Per Flow Throughput
22
Per Flow Throughput
14
20 18 16 14
AODV
12
SR
10
ANODR
8
Packet Loss Ratio (%)
Packet Loss Ratio (%)
17 16.5 16 15.5 15 14.5 14
12 10 8
AODV
6
SR
4
ANODR
2 1
2
3
4
5
Mean Node Speed (m/s)
c.
Packet Loss Ratio
ISBN NO : 378 - 26 - 13840 - 9
1
3
5
7
9
Number of Malicious Node
e. Packet Loss Ratio
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 5
Data Packet Delay (ms)
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
3.
220 200 180 160 140 120 100 80 60 40 20
4.
AODV SR ANODR 1
3
5
7
6.
9
Number of Malicious Node
f.
5.
7.
8.
End-to-end Delay
9.
VI. CONCLUSION There will be lot of MANET security routing concept can be obtained, but yet now there will be no intimation to secure packet authentication. If there is any malicious activity means those packet will be dropped so the information might be destroyed. Therefore this is a correct solution for an hackers or intruders they achieve their task which means they finish their main job. But I’ll find a solution by introducing the dummy packet concept with some secured mechanism, this will helpful for lightning a successful path of conveying original information from source to destination. And I’ll used femto cell to make more strengthened signal coverage for achieving more efficient process.
10.
11.
12.
13. 14.
VII. FUTURE SCOPE The nature cause of anonymous is come due to a confidential message or data which is sent through an intermediate node. So I first focused about how to make an malicious intermediate node into trustable intermediate node. I’ll going to accomplish this task by generating survey about nodes by finding out good node, bad node and unknown status about either good or bad node. It is an attainable solution by using the ranking method concepts.
15.
Hussein Al-Bahadili and Khalid Kaabnel, “Analyzing the Performanceof Probabilistic Algorithm in Noisy MANETs”, International Journal of Wireless and Mobile Networks, pp.82-94, Vol.2, No.3, August 2010. Shio Kumar Singh, M.P Singh and D.K Singh, “Routing Protocols in Wireless Sensor Networks - A Survey”, International Journal of Computer Science and Engineering Survey, Vol.1, No.2, November 2010. Hee Yong Youn, Chansu Yu, Ben Lee,”Routing Algorithms for Balanced Energy Consumption in Ad-hoc Networks”. Zaiba Ishrat, Pankaj Singh, “An Enhanced DSR Protocol Using Path Ranking Technique”, International Journal of Engineering Research and Applications, Vol.3, Issue 3, pp.1252-1256, MayJUNE 2013. Wenjia Li, Anupam Joshi and Tim Finin, “SMART: An SVMBased Misbehavior Detection and Trust Management Framework for Mobile Ad-hoc Networks”. Ruchi Rani, Manisha Dawra, “Performance Characterization of AODV Protocol in MANET”, International Journal of Advanced Research in Computer Engineering and Technology, Vol. 1, Issue 3, May 2012. Kannan Govindan, Member IEEE and Prasant Mohapatra, Fellow IEEE, “Trust Computations and Trust Dynamics in Mobile Adhoc Networks: A Survey”. Changbin Liu, Yun Mao, Mihai Opera, Prithwish Basu, Boon Thau Loo, “A Declarative Perspective on Adaptive MANET Routing”, August 2008. Mamatha.T, “Network Security for MANETs”, International Journal of Soft Computing and Engineering, Vol.2, Issue 2, MAY 2012. Muhammad Arshad Ali and Yasir Sarwar, “Security Issues Regarding MANET (Mobile Ad-hoc Networks): Challenges and Solutions”, Master Thesis, Computer Science, Thesis No: MCS2011-11, March 2011. DSF for MANETs (Distributed Services Framework for Mobile Ad-hoc Networks). Gagandeep, Aashima, Pawan Kumar, “Analysis of Different Security Attacks in MANETs in Protocol Stack A-Review”, International Journal of Engineering and Advanced Technology, Vol.1, Issue 5, June 2012. WWRF/WG4/Ad-hoc Networking-Subgroup WhitePpaper, Version 1.0, 17th June 2002.
VIII. REFERENCES 1. (Nevin) Lianwen Zhang and David Poole, “Stepwise-Decomposable Influence Diagrams”, Department of Computer Science, University of British Columbia, Vancouver, B.C, V6T/Z2, Canada. 2. N.Vanitha, G.Jenifa, “Detection of Packet Droppers in Wireless Sensor Networks Using Node Categorization Algorithm”, International Journal of Advanced Research in Computer Science and Software Engineering, Volume 3,Issues 3,pp.69-74, March 2013.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 6
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
A Survey on Enroute Filtering Scheme in Wireless Sensor Networks P.Pritto Paul, Asst Professor, Velammal Engg. College, Anna University, Chennai. p.prittopaul@gmail.com
Abstract— Wireless Sensor Networking is one of the most prominent technology that is used in almost all real time applications. WSN is used in the estimation of temperature in Cyber Physical Network System(CPNS) where the sensor nodes are deployed in hostile environment. In this environment the sensor nodes sense the data and forward the report to the base station. When the report is being forwarded to the base station the attacker may forge the data or may inject false data into the report by compromising the sensor nodes in the network this leads the base station to generate false decision. The solution to overcome the False Data Injection Attack is to implement the Enroute Filtering Scheme in WSN. The Enroute Filtering is used to check the correctness of the data before it is being forwarded to the base station. In this paper some of the most efficient Enroute Filtering Schemes for filtering false data have been discussed with their advantages and disadvantages. And also forwarding and filtering of data in Cluster based environment which provides high security than other filtering schemes have been discussed.
I. Introduction WIRELESS sensor networks are expected to interact with the physical world at an unprecedented level to enable various new applications. However, a large-scale sensor network may be deployed in a potentially adverse or even hostile environment and potential threats can range from accidental node failures to intentional tampering. Due to their relatively small sizes and unattended operations, sensor nodes have a high risk of being captured and compromised. False sensing reports can be injected through compromised nodes, which can lead to not only false alarms but also the depletion of limited energy resource in a battery powered network.
ISBN NO : 378 - 26 - 13840 - 9
K.Thejaswi, PG-Student, Velammal Engg. College, Anna University, Chennai. tejaswi.kessamsetti@gmail.com
The false data injection in a cyber physical network system can be overcome by the formation of clusters where the neighbor sensor node with nearly similar properties will be organized into the form of clusters. In the hierarchical network structure each cluster has a leader, which is also called the cluster head (CH). The sensor nodes periodically transmit their data to the CH nodes.CH nodes aggregate the data and transmit them to the base station (BS) either directly or through the intermediate communication with other CH nodes. The BS is the data processing unit for the data received from the sensor nodes.The Base Station is fixed at a place in a stationary manner which is far away from the all the sensor nodes .The function of each CH,is to perform common functions for all the nodes in the cluster, like aggregating the data before sending it to the BS. In some way, the CH is the sink for the cluster nodes, and the BS is the sink for the CHs. The advantages of cluster based environment is: 1) supporting network scalability and decreasing energy consumption through data aggregation 2) It can localize the route setup within the cluster and thus reduce the size of the routing table stored at the individual node. . The main parameters included in clustering are: Number of clusters, Nodes and CH mobility, Nodes types and roles, Cluster formation methodology, Cluster-head selection.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 7
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
In past different scheme have been proposed for filtering false data in wireless sensor networks where the data is transferred in a environment where the sensor nodes are scattered. For example inStatisticalEnroute Filtering Scheme[1],Interleaved Hop-by-Hop Schemes[2] have the limitation of node compromising where the false data can be injected in order to generate the false reports. In the paper we discuss about the compromise resilient enroute filtering scheme where the sensor nodes are organized into the form of clusters. And the data is transferred to Base station (sink) with the help of forwarding nodes which act as an intermediate between the cluster and the base station. The rest of the paper is organized as follows: 1) a brief survey on existing filtering schemes. overview of compromise resilient enroute filtering scheme in cluster based environment in WSN. 2) a survey on compromised resilient enroute filtering scheme in WSN.3) Then, a detailed literature survey on enroute filtering devised for WSNs is provided along with comments on their prominent and lacking feature.
To overcome the threshold limitation and to reduce the increasing number of compromised nodes which we have seen in SEF. We come up with LBRS b) Location-Based Resilient Security (LBRS)[2] approach which make use of two techniques: location-binding keys and location-based key assignment. In location based resilient scheme the location of the sensors and sink is stationary by which it can assign fixed key values for the sensor in order to provide security. Based on its location, a node stores one key for each of its local neighboring cells and a few randomly chosen remote cells. LBRS provides a solution to this security problem, but it depends on the stationary of the sink and the fixed routing model such that it cannot work with mobile sinks and various routing protocols. The disadvantage of LBRS is it relies on special data dissemination protocol to confirm a bean model. c) Grouping-Based Resilient Statistical Enroute Filtering (GRSEF)[4]scheme for filtering false data. The GRSEF does not depend on sink stationary. It improves the filtering efficiency by II.RELATED WORKS dividing the sensor nodes into certain number of groups(e.g.: T-groups) and assigns authentications We discuss about existing filtering schemes, to the groups. GRSEF employees a multi-axis which make use of MAC(message authentication division technique to overcome the threshold Codes) for transferring the data. limitation problem that we have seen in SEF[1] and IHA[4]. In GRSEF, the Redundancy is increased to a) Statistical Enroute Filtering (SEF)[1] is the achieve the robustness against this attacks but the most basic mechanism in which dense deployment disadvantage of GRSEF is it has no resilience to the of large sensor networks takes place. To prevent selective forwarding attack and report disruption any single compromised node from breaking down attack . the entire system, SEF sends only limited of All the early proposed scheme have the amount of security information assigned to each disadvantage of T-threshold limitations, this node, and depends on the collective decisions of schemes does not adopt to dynamic topology, takes multiple sensors for false report detection. As a long time for a network to become stable, sensor report is forwarded through multiple hops toward nodes are scattered in a wireless networks were the sink, each intermediate node verifies the there is no security for the transmission of data, correctness of the MACs carried in the report and Requires node localization and takes a long time to drops the report if an incorrect MAC is detected. be stable. Inorder to overcome the disadvantages The disadvantage of SEF is probability of detecting our survey has come up with formation of clusters incorrect MACs increases with the number of hops in wireless sensor network where the sensor nodes the report travels. SEF[1] and IHA[4] have the Tare grouped to form clusters. threshold limitations. That is, if the adversary compromises T nodes from different groups, they could inject false data to generate the false report.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 8
Proceedings of International Conference on Developments in Engineering Research
III. Preliminary a) The Basics of Enroute Filtering: The enroute filtering is technique used in wireless networks with which the intermediate nodes checks the correctness of the data that is being travelled along the route from source to the sink with the help of intermediate nodes present in the network. The intermediate node not only checks the correctness of the data but also can filter the false data effectively. The intermediate nodes after receiving the report checks whether it contain valid T-MAC. The report with less number of T-MAC will be dropped. If any false data which is not filtered by the intermediate nodes will be detected by the sink where it gets filtered. The sink acts as the final defense that catches false reports not filtered out by forwarding nodes. b) System model of Enroute Filtering:
Sensor node
Intermediate node should check for correct MAC
sink Compromised node(injection of false data)
c) Security Model Of Enroute Filtering:
We consider a large sensor network field where nodes are deployed. So after the network initialization phase the sensor nodes forms into groups and elect a cluster head based on different parameters like remaining energy etc. Whenever events of interest occurs in the terrain say if a tank moves, all the cluster members near to the event will sense the happening and report to their cluster heads. On receiving the reports cluster head ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
aggregates them and sends a single copy of the valid report to the base station through selected report forwarding nodes. The selections of report forwarding nodes are up to the underlying routing protocol’s work . And also the selection parameters are independent of the application. We assume that there are attackers present within the terrain are capable of monitoring the communication pattern between the sensor members and the cluster head to guess the message from the reports if intercepted. We assume that each cluster contains at most t-1 compromised nodes, which may collaborate with each other to generate false reports by sharing their secret key information. The potential attacks which we consider in our work DoS attacks. DoS attacks include selective forwarding and report disrupt d) Proposed System for Enroute Filtering: The proposed system for enroute filtering is based on cluster environment where the sensor nodes are organized into groups(clusters).In cluster based sensor nodes makes use of two keys authentication key and check key instead of MAC used in the existing systems. The sensor nodes within the cluster are assigned with authentication key and the forward nodes are assigned with check key in order to provide additional security. The security keys for the sensing node and the forwarding node is assigned by the sink. Different nodes present in different clusters are assigned with different authentication keys. In this way a compromised node present in one cluster will not effect the nodes present in the other cluster. Therefore this scheme achieves better resilience to the increased number of compromised nodes. The main advantage of this scheme is that it does not depend on static routes and node localization. This scheme mainly consists of two principles: Management of authentication information: this is used to assign the key values to the nodes present in the network. Management of data security: this is used to detect and filter the false data. The report that is being forwarded from one cluster should contain: the encrypted measurement, cluster key, sensor node ID, local ID of the node from where it has been generated, authentication information of the measurement generated by the node.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 9
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
e) System model of proposed system:
C1
CH
FS1 S1
S2
C2
CH
S2
S3
FS2 C3
SINK
S1
CH S3
FS1, FS2 - Forward Sensor Nodes C1, C2, C3 - Clusters CH - Cluster Head S1, S2, S3 - Sensor Nodes
f) Algorithms used in Enroute Filtering:
Type Of Algorithm
Algorithm Usage
1. Kar and Banerjee’s algorithm 2. Greedy algorithm
Sensor-deployment in a network
Distributed algorithm
Compute Support Weight(SW) between the sensor nodes in a network.
Veltri et al. algorithm
Distributed localized algorithm
Kanan et al. algorithm
Polynomial time algorithms
Clustering Algorithm
1. LEACH – C Algorithm 2. Efficient Cluster Head Selection Scheme For Data Aggregation [EECHSSDA] 3. Hybrid Energy- Efficient Distributed Clustering
Greedy base-station algorithm ISBN NO : 378 - 26 - 13840 - 9
1. For unidirectional antennas 2. For omnidirectional antennas
Advantages 1. Achieve Coverage 2. Achieve Connectivity 1. Construct Local Neighborhood Graph. 2. Construct Best Support Path 1. To find an approximate minimal exposure path. 2. Linear programming formulation for minimal- and maximal-exposure paths is obtained. 1. compute the maximum vulnerability of a sensor deployment to attack by intelligent adversary . 2. To compute optimal deployments with minimal vulnerability. Cluster Head Selection takes place based on following clustering algorithm.
Forwarding of data to the base station
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 10
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
IV. Literature Survey Schemes for filtering false data
Advantages
Disadvantages
Statistical Enroute Filtering
Dynamic topology shared key mechanism
Threshold problems No resilience to attacks.
Location-Based Resilient Security
Not applied to dynamic topology Require node localization
Require node localization Lower resilience to attacks.
Grouping-Based Resilient Statistical
Avoid threshold limitations Location-basedkey generation
Uses multi-axis division technique Avoid threshold limitations
V. Conclusion The clustering scheme achieves not only high en-routing filtering probability but also high reliability for filtering the injected false data with multi-reports without depending on static routes and node localization. Due to the simplicity and effectiveness, the cluster based scheme could be applied to other fast and distributed authentication scenarios in wireless network.
VI. References 1) F. Ye, H. Luo, S. Lu, and L. Zhang, “Statistical en-route filtering of injection false data in sensor networks,” IEEE Journal on data in sensor networks,” IEEE Journal on selected areas in communication,VOL.23, NO. 4, April 2005 2)” Toward Resilient Security in Wireless Sensor Networks” Hao Yang, Fan Ye, Yuan Yuan, Songwu Lu, William Arbaugh
ISBN NO : 378 - 26 - 13840 - 9
3) L. Yu and J. Li, “Grouping-based resilient statistical en-route filtering for sensor networks,” in Proc. of the 28th IEEE International Conference on Computer Communications (INFOCOM’09), 2009, pp. 1782–1790. 4) S. Zhu, S. Setia, S. Jajodia, and P. Ning, “An interleaved hop-byhop authentication scheme for filtering of injection false data in sensor networks,” ACM Transactions on Sensor Networks (TOSN), vol. 3, no
5) N.Parashuram, Y.Sanjay sai raj, A.Sagar, B.Uma “An Active En-route Filtering Scheme for Secured Data Dissemination in Wireless Sensor Networks” IJCSET |April 2012| Vol 2, Issue 4,1102-1
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 11
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
6) “Clustering in Wireless Sensor Networks” textbook Basilis Mamalis, Damianos Gavalas, Charalampos Konstantopoulos, and Grammati Pantziou. 7)“Algorithms For Wireless Sensor Networks” Sartaj Sahni and Xiaochun Xu Department of Computer and Information Science and Engineering, University of Florida, Gainesville, FL 32611 {sahni,xxu}@cise.ufl.edu September 7, 2004 8)” A Random Perturbation-Based Scheme for Pairwise Key Establishment in Sensor Networks” Wensheng Zhang and Minh Tran Dept. of Computer Science, Iowa State University, Ames, IA 50014, USA 9) “Filtering Schemes for Injected False Data in Wsn” IOSR Journal of Computer Engineering (IOSR-JCE) e-ISSN: 2278-0661, p- ISSN: 2278-8727Volume 13, Issue 6 (Jul. Aug. 2013), PP 29-31.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 12
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Estimation of Damping Torque for Small-Signal Stability of Single Machine Infinite Bus System M.Venkateswara Rao
A.Krishna veni
Associate professor
PG scholar
mvr.venki@gmail.com,
krishnaveni203@gmail.com
Department of Electrical and Electronics Engineering, GMRIT, Rajam, A.P. INDIA.
Abstract— This paper discusses the Estimation of damping torque coefficient for Small-signal stability of infinite bus system. This damping torque coefficient is used to identify the angle stability of a system. Initially a mat lab coding was utilized to generate the time domain responses of rotor angle, rotor speed and electromagnetic torque under various loading conditions. The particle swarm optimization (PSO) technique is then used for accurate estimation of damping torque coefficient. The mat lab coding results using PSO, under various loading conditions shows the effectiveness of the proposed control strategy.
Index
terms – Damping torque coefficient, Particle swarm optimization, Small-signal stability, and Synchronizing torque coefficient. I. INTRODUCTION The power system instability can be demonstrated in many different ways depending on the system configuration and working mode. Since power system works on synchronous generators, an essential condition for system operation is that all synchronous machines remain in synchronism [1-3]. The small signal stability
ISBN NO : 378 - 26 - 13840 - 9
is the ability of the power system to maintain synchronism when subjected to a small disturbance [1]. The operating condition of the power system changes with respect to time because of the dynamic nature of the system. The rotor angle stability can be analyzed from the Synchronizing torque coefficient KS and Damping torque coefficient KD . For stable operation of the system, both synchronizing and damping torque coefficients must be positive. The electromagnetic torque deviation is split into Synchronizing torque and Damping torques. The Synchronizing torque is responsible for restoring the rotor angle excursion and the Damping torque damps out the speed deviations [4, 5]. In general the synchronizing and damping torques are expressed in terms of Synchronizing torque coefficient KS and Damping torque coefficient KD . These KS , KD can be calculated frequently for stability assessments. Various computational techniques like Simulation Annealing (SA) algorithm, Evolutionary programming (EP), Genetic Algorithm (GA) and Differential Evolution (DE) are employed for optimization problem [7, 8]. These techniques need more parameters, high calculation time and not
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 13
Proceedings of International Conference on Developments in Engineering Research
easy to implement when compared to particle swarm optimization. Particle swarm optimization (PSO) was developed by Kennedy and Eberhart. This has appeared as a promising algorithm for handling the optimization problems [20]. PSO is a robust, non-linear and population based stochastic optimal technique which can generate high quality solutions within shorter calculation time. The single-machine power system modeling and small-signal stability studies are carried out using Eigen analysis based technique With PSO (particle swarm optimization) optimal strategy. The suggested control technique is based on estimation of damping torque coefficient KD of a synchronous machine from the time responses of the rotor angle r (t ) , rotor speed (t ) and electromagnetic torque Te (t ) . Thus PSO has been chosen to coordinate the operation in estimating Damping torque coefficient KD for stability analysis [17-19]. II. POWER SYSTEM MODEL A simplified block diagram model of the small signal performance is shown in fig1 [1]. In this work, the proposed method has been tested on a system comprising a single machine connected to infinite bus system through a transmission line. Normally, for small signal stability study a second-order model is considered for the synchronous generator. The single machine infinite bus system model is linearized at a particular operating point to obtain the linearized power system model. This model is represented with some variables, such as electrical torque, mechanical torque, and rotor speed and rotor angle.
www.iaetsd.in
Figure1. Block diagram model of small signal performance. In the classical generator model, the acceleration circuit dynamic equations are:
r 1 (Tm Te K D r ) t 2H 0 r t
(1)
(2)
Where Tm , Te are mechanical torque, electromagnetic torque and 0 = 2 f 0 . From the block diagram, the following statespace form is developed.
X AX BU K d r D 2H dt 0
K s 1 r 2 H 2 H Tm 0 0
The elements of the system matrix A are function of K D , H , X T and the initial operating conditions. The perturbation matrix B depends on the system parameters only. From the block diagram of figure1, we have
0 S
1 2 HS K S K D r Tm (3)
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 14
Proceedings of International Conference on Developments in Engineering Research
modes is that the elements of the eigenvectors are dependent on units and scaling associated with the state variables. To overcome this participation matrix which combines the right and left eigenvectors are used. The participation factor provides a measure of association between the state variables and the oscillatory modes.
And the characteristic equation is S2
KD K S S 0 0 2H 2H
(4)
Therefore, the damping ratio is
(5)
1 2
KD
IV. SMALL-SIGNALSTABILITY ASSESSMENT USING SYNCHRONISING AND DAMPING TORQUES.
K S 2 H 0
E ' E B (cos j sin ) It jX T
(6)
X T X d' X E
(7)
KS
E 'EB cos 0 XT
(8)
III. SMALL SIGNAL STABILITY ASSESSMENT USING MODAL MATRICES The power system experiences small disturbances by small changes in loads. Then the system will be driven to an infinite state X (t 0 ) X 0 at time t 0 =0. The system responds according to the state equations. The linearized state equations can be used to find the Eigen values i of the system matrix A , where i i j i are the distinct eigenvalues corresponding to a set of right and left eigenvectors. Here i is a damping factor and i is Damped angular frequency. The right and left Eigen vectors are orthogonal and are usually scaled to be orthogonal. Real eigenvalues indicates modes which are aperiodic and complex eigenvalues indicates modes which are oscillatory. The uses of right and left eigenvectors are for identifying the relationshipbetween the states and the
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
The electromagnetic torque ( Te ) deviation of a machine can be expressed as its speed ( ) and angle ( ) deviations, which are called damping and synchronizing torques. The synchronizing and damping torques are expressed in terms of its synchronizing torque coefficient ( KS ) and damping torque coefficient ( KD ). Then the electromagnetic torque deviation will be expressed as: Te (t ) K D 0 (t ) K S (t )
(9)
Where r (t ) = change in rotor speed (t )
= change in rotor angle
V. OVER VIEW OF PARTICLE SWARM OPTIMIZATION PSO is one of the evolutionary based optimization techniques [Fukuyama, 1999; Kennedy and Eberhart, 1995]. This method is introduced based on the research of bird and fish flock movements behavior. Due to its many advantages like its simplicity and easy implementation, the algorithm can be widely used in function optimization. PSO ALGORITHM: The particle swarm optimization consists of ‘n’ particles and the particles position stands for the potential solution in D-
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 15
Proceedings of International Conference on Developments in Engineering Research
dimensional space. Each particle can be shown by its current speed and position. Particles change its position according to it’s: 1. Current position 2. Current velocity 3. Distance between its current position and pbest 4. Distance between its current position and gbest Velocity of each particle can be modified based on the following equation. k k k k k vidk1 wvidk c1r1k (pbest id xid ) c2r2 (gbest id xid )
By using velocity equation, a certain velocity which gradually gets close local best and global best can be calculated. The current position of the particle can be modified by the following equation.
xidk1 xidk vidk1
For the linearized system model presented in figure1, the eigenvalues of the local system can be evaluated. The proposed method is aiming to search for the optimal damping torque coefficient, such that the damping ratio can be maximized.
1 2
KD K S 2 H 0
Where =damping ratio. For stable operating condition of the system the Damping ratio must be in [0.4, 0.7]. Hence the Corresponding values of KD will be [35.712, 63.125]. The control parameters can be tuned through the optimization algorithm. The proposed algorithm will be as follows. VII. IMPLEMENTATION OF PSO FOR OPTIMAL ESTIMATION OF DAMPING TORQUE Step1. Read the system input data, PSO parameters.
Where, pbest represents the D-dimension quantity of the individual “i” at its most optimist position at its “k” times and gbest represents the D-dimension quantity of the individual “i” at its most optimist position at its “k” times. The speed of the particle at its each direction is confined in between – vdmix and +vdmax. If vdmax is too big, solution is far from the best and if vdmax is too low, it means that the solution will be local optimism. C1, C2 represents speeding figures which lies between 0 to 2. r1, r2 represents random fiction, and 0-1 is a random number. VI. ESTIMATION OF LOWER AND UPPER LIMITS OF DAMPING TORQUE COEFFICIENT.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
Step2. Initialize population of particle ( KD ) with random velocities and positions. Step3. Evaluate fitness values using the objective function.
1 2
KD K S 2 H0
Step4. Each particle has its own best position called local best and the best position among all the particles is called global best. Step5. Update the velocity of particle using
vidk1 vidk c1r1k ( pbestidk xidk ) c2r2k (gbestidk xidk )
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 16
Proceedings of International Conference on Developments in Engineering Research
Check the updated velocity, within the limits or not vimin vi v imax
www.iaetsd.in
The following responses show the rotor characteristics under various loading conditions.
Step6. Update the position of particle using
xidk1 xidk vidk1 Step7. Evaluate fitness values of the new particles. Update the local best values as current fitness values if these values are better than previous values, update it. Then find new global best values. Step8. Repeat the procedure until the stopping criteria is reached.
Figure 2-a. Rotor speed response for P=0.9,Q=0.3.
VIII. RESULTS AND DISCUSSIONS In this work the optimal values of KD are obtained using PSO and the rotor speed, rotor angle and electromagnetic torque responses are generated and are compared with those obtained in [1]. In which the damping torque coefficient is chosen randomly [-10, 0, 10]. In addition the same responses are generated for different loading conditions using mat lab coding. It is observed that the steady state stability of the system is improved. These responses are shown in figure2-a to figure4-c. The rotor responses are obtained for various conditions: 1. Nominal operating condition (P=0.9, Q=0.3). 2. Light operating condition (20% of the nominal values). 3. Heavy operating condition (50% higher than the nominal operating condition).
Figure 2-b. Rotor Angle response for P=0.9, Q=0.3.
Figure 2-c. Torque response for P=0.9, Q=0.3.
The following responses show the rotor characteristics under various loading conditions.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 17
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Figure 4-a. Rotor speed response for P=0.72, Q=0.24 Figure 3-a. Rotor speed response for P=1.35, Q=0.45.
Figure 4-b. angle response for P=0.72, Q=0.24.
. Figure 3-b. Rotor Angle response for P=1.35, Q=0.45
Figure 3-c. Torque response for P=1.35, Q=0.45.
ISBN NO : 378 - 26 - 13840 - 9
Figure 4-c. Torque response for P=0.72, Q=0.24.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 18
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Table2.PSO parameters:
CONCLUSION In this project the steady state performance improvement is obtained by the accurate estimation of damping torque coefficient using particle swarm optimization technique. The matlab programming results using PSO under various loading conditions shows the effectiveness of the proposed technique. Compared to normal operating conditions under heavy and light operating conditions the peak over shoots are very high which are reduced using PSO technique. The effectiveness of the proposed controller is to provide good damping of low frequency oscillations. It can be concluded that the proposed PSO controller extends the power system stability limit by enhancing the system damping.
Population size Maximum number of generations Acceleration coefficients(C1,C2) Inertia weight Table3 (Loading conditions):
S.no
1 2
3
APPENDIX: Input data: 1. Generator parameters:
10 50 1.4 1
Cases
P(p.u)
Nominal operating condition 0.9 Heavy operating condition (50% higher 1.35 than the nominal load) Light operating condition(20% of the 0.72 nominal operating condition)
Table4. (Optimal parameter)
value
of
Q(p.u)
0.3 0.45
0.24
control
H 3.5, Td' 0 8, X d 1.81, X q 1.76, X d' 0.3, Ra 0.003, Ksd Ksq 0.85
KD
57.8
2. Transmission line parameters: Re 0, X e 0.65, X L 0.15.
REFERENCES
Table1. (Lower, Upper limits of control parameter): PARAMETER KD Lower limit 35.712 Upper limit 63.125
ISBN NO : 378 - 26 - 13840 - 9
[1] Kundur, P.: “power system stability and control”, McGraw-Hill, 1994. [2] Hsu Y.Y., Chen, C.L.:“Identification of optimum location for stabilizer Application using participation factors”. In: IEEE Proc. Gen., Trans. & Distr., Pt. C. VOL. 134. No.3.1987. P. 238-244. [3] Demello F.P., concordia. C,“Concepts of synchronostability as affected by excitation control”. In: IEEE Trans.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 19
Proceedings of International Conference on Developments in Engineering Research
[4]
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12]
Power Apparatus and Systems, Vol.PAS-88, No.4.April 1969, p.316329. Feilat, E.A., Younan, N., Grzybowski, S.: “Estimating the synchronizing and damping torque coefficients using Kalman filterin”. In: Electric power Systems Research, Vol.52, No.2, 1992.p.145=149. Hassan Ghasemi, Claudio canizares. “On-line Damping Torque estimation and Oscillatory Stability Margin Prediction ”. In IEEE Transactions on Power Systems Vol. 22, No.2, May 2007. Gurunath Gurrala, Indraneel Sen.: ’’Synchronizing and damping torque analysis of nonlinear voltage regulators”. In: In IEEE Transactions on Power Systems Vol. 26, No.3, and May 2011. Y.U, Y.N.: “Electrical Power System Dynamics,” Academic Press, 1983. N.A.M Kamari, I. Musirin, M.M. Othman, “Improving Power System Damping Using EP Based PI Controller”. PEOCO 2011, June 2011, PP.121-126. T.K. Rahman, Z.M. Yasin, W.N.W. Abdullah, “Artificial-Immune-Based for Solving Economic Dispatch in Power System”. PEcon 2004, NOV 2004, PP.31-35G. M.Hunjan, G.K. Venayagamoorthy, “Adaptive Power System Stabilizer Using Artificial Immune System”. ALIFE’07, April 2007, PP.440-447. F.P. Demello, C. Concordia, “Concept of Synchronous Machine as Affected By Excitation Control”. In: IEEE trans. Power system apparatus, PSA-87. 1968. PP.835-844. Y.L. Abdel-Magid, A.H. Mantawy “Robust Tuning Of Power System Stabilizers in Multimachine Power
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
[13] [14]
[15]
[16]
[17]
[18]
[19]
[20]
Systems”. In: IEEE Trans. Power System, Vol. 15, No. 2, May2000. Kaith, T.: “Linear Systems”. Prentica Hall, NJ, 1980. T.C. Hsia, “system identification: least square methods”. Lexington, MA: Lexington Books, D.C. Health and Company, 1977. M.J. Gibbard, “Co-Ordinated Design of Multimachine Power System Stabilizers Based On Damping Torque Concepts”. In: IEEE Proceedings, Pt. C, Vol. 135, No. 4, July 1988, and PP.1276-284. F. Glover, “Artificial Intelligence Heuristic Frameworks and Tabu Search”. Managerial and Decision Economics, Vol. 11, PP.365-375, 1990. Shaltout, A., Feilat, E.A.: “Damping and Synchronizing Torque Computation in Multimachine Power Systems”. In: IEEE Trans. On Power Systems, Vol, PWRS-7, No.1, February 1992, P. 280-286. Hiroshi Suzuki, Soichi Takeda, Yoshizo Obata: “An Efficient Eigenvalue Estimation Technique for Multimachine Power System Dynamic Stability Analysis”. In: Electrical Engineering in Japan, Vol. 100, No.5, 2007, PP.45-53. F.P. Demello and C. Concordia, “Concepts of Synchronous Machine Stability as Affected By Excitation Control”. In: IEEE Trans. Power Apparatus and Systems, Vol. PAS-88, No. 4, PP. 316-329, April 1969. J. Kennedy and R.C. Eberhart,. “Particle Swarm Optimization”. Proceeding of IEEE international conference of Neural Networks, Vol. 4, 1995, PP. 1942-1948.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 20
Proceedings of International Conference on Developments in Engineering Research
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 21
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
An Enhanced Feature Selection for High-Dimensional Knowledge 1
L.Anantha Naga Prasad 1 2 K.Muralidhar M.Tech, Computer Science and Engg, Anantha Lakshmi Institute of Technology&Sciences,JNTUA, Andhra Pradesh, India Assistant Prof, Department of CSE, Anantha Lakshmi Institute of Technology&Sciences, JNTUA, Andhra Pradesh, India 1 mail id: naga4all16@gmail.com 2 mail id: muralidhar.kurni@gmail.com
2
ď€ Abstract Irrelevant features, at the side of redundant features, strictly have an effect on the correctness of the knowledge machines. Thus, feature set selection have to be compelled to be able to determine and take away the maximum as much of the unrelated and redundant knowledge as feasible. With this intention choosing a subset of features with relation to the target notions, feature set selection is an efficient alternative way for reducing spatial property or dimensionality (ex: subset), removing unrelated data (Ex: irrelevant data), increasing learning accuracy, and generating Qualitative result. Feature selection involves classifying a set of the foremost relevant features that generates appropriate outcome as the original entire set of features. Several feature set selection techniques are planned and studied for machine learning applications. By this criterion, an Enhanced fast clustering-based feature selection algorithm, EFAST, is employed during this paper. The EFAST algorithmic rule works in 2 steps. In the starting step, features are classified into clusters by exploitation graph-theoretic clustering approaches. In the second step, the foremost relevant representative feature that is powerfully associated with target categories is chosen from every cluster to form a set of features. Features in dissimilar clusters are comparatively autonomous and the clustering-based strategy of EFAST includes a high chance of generating a set of valuable and autonomous features. Keywords-EFAST Algorithmic rule, Correlations, Feature set Selection, and Graph based clustering
1 INTRODUCTION The use of feature selection can develop accurateness, relevancy, applicability and be aware of a learning method. For this reason, several ways of automatic feature selection are developed. Some of these ways are based on the search of the features that enables the data set to be measured consistent. In an exceedingly search problem we usually tend to evaluate the search states, in the case of feature selection we measure the promising feature sets. Feature set selection is a very important subject when preparing classiďŹ ers in Machine Learning (ML) issues. Selection of Feature set is an efficient system for dimensionality reduction, elimination of inappropriate knowledge, rising learning accurateness, and improving result unambiguousness. Based on the minimum spanning tree methodology, we propose an EFAST algorithmic rule. The ISBN NO : 378 - 26 - 13840 - 9
algorithmic rule is a 2 step method, in that features are separated into clusters by way of using graph theoretic clustering means. Within the succeeding step, the frequently used representative feature that is robustly related to target categories is specific from every cluster to structure the ultimate subset of features. Features in distorted clusters are comparatively autonomous. The clustering-based theme of EFAST includes a high risk of designing a set of constructive and autonomous features. In our planned EFAST algorithmic rule, it needs the building of the minimum spanning tree (MST) from a subjective comprehensive graph. The separation of the MST into a forest by means of each tree signifying a cluster and the collection of representative features from the clusters. The planned feature set selection algorithmic rule EFAST was tested and the investigational results demonstrate that, evaluated with different varied forms of feature set selection algorithms, the projected algorithmic rule not solely decrease the amount of features, but also advances the performances of the famed varied forms of classifiers.
The results, on publically obtainable real-world high dimensional image, microarray, and text knowledge, established that EFAST not only produces smaller sets of features however improves the performances of the classifier. In our study, we tend to apply graph theoretic clustering schemes to features. In exacting, we tend to accept the MST based clustering algorithms, since they do not imagine that knowledge points are classified around centres or separated by a normal geometric curve and are widely used in training. Based on the MST method, we tend to suggest an Enhanced Fast clustering-bAsed feature Selection algoriThm (EFAST). Good feature set is one that contains features extremely correlative with the target, so far uncorrelated with one another. In FOR the SKILL aboveDEVELOPMENT planned INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY 22
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
terms this system is an efficient fast filter method features and low numbers of training instances of the which may categorize relevant features as redundancy algorithmic rule. among relevant features without pair wise correlation Relief-F could be a feature choice strategy study, and repeatedly chooses features which exploit that chooses cases randomly, and altered the weights their mutual information with the category to expect of the feature importance based on the closest provisionally to the reply of any feature formerly neighbor. By its qualities, Relief-F is one in all the elected. In contrast to from these algorithms, our foremost prosperous methods in feature choices. projected EFAST algorithmic rule utilizes clustering 2.2 Disadvantages of Existing System based methodology to select features. The simplification of the chosen features is restricted and hence the complexness is large. 2 RELATED WORK Accurateness is not guaranteed. 2.1 EXISTING SYSTEM In the past approach there are many algorithms that Ineffective at deleting redundant features illustrate a way to maintain the knowledge into the Performance associated problems database and how to retrieve it quicker, however the Security problems difficulty here is no one cares about the database So the attention of our new system is to boost the maintenance with ease manner and safe methodology. outturn for any basis to eliminate the knowledge A Distortion algorithmic rule, that creates a personal security lacks there in and build a more recent system space for every and each word from the already outstanding handler for handling data in an elected transactional database, those are put together economical manner. named as dataset, which is able to be acceptable for a 2.3 Proposed System collection of exacting words, however it'll be In this proposed system, The Enhanced fast problematic for the set of records. An inference clustering-based feature selection algorithmic rule algorithmic rule build propagation to the higher than (EFAST) works in 2 steps. In the first step, features downside, and cut back the issues occurred within the are classified into clusters by exploitation existing distortion algorithmic rule, however here graph-theoretic clustering ways. conjointly having the matter known as knowledge In the second step, the foremost relevant overflow, once the user get confused then they will representative feature that is powerfully associated never get the knowledge back. The embedded ways with target categories is chosen from every cluster to incorporate feature choice as a locality of the training form a set of features. method and are sometimes specific to given learning Features in dissimilar clusters are comparatively algorithms and as a result could also be improved than autonomous and therefore the clustering-based the opposite 3 teams. Typical machine learning strategy of EFAST includes a high probability of algorithms like decision trees or artificial neural generating a set of valuable and autonomous features. networks are samples of embedded ways. The Inside this paper we tend to generate correlations wrapper techniques use the analytical accuracy of a for high dimensional knowledge supported EFAST planned algorithmic rule to decide the goodness of the algorithmic rule in four steps. actual subsets, the accurateness of the learning 1. Removal of unrelated features: algorithms is usually high. But the simplification of If we choose a Dataset 'D' with m features F= {F1, the chosen feature is restricted and the process F2... Fn} and class C, mechanically features are problem is high. The filter ways are autonomous of obtainable with target relevant feature. The learning algorithms, with fine generality. Their simplification of the chosen features is restricted process complexness is low, however the accuracy of and the process complexness is huge. If Fi is the learning algorithms is not assured. The hybrid relevancy to the target C if there exists some si, fi, techniques are a mix of filter and wrapper ways by and c specified for probability p(Si=si , Fi=fi) >0, employing a filter methodology to diminish search p(C=c | Si =si, Fi=fi)≠p(C=c | Si = si) otherwise space which will be measured by the succeeding feature Fi is an unrelated feature. wrapper. They primarily target on grouping filter and 2. T-Relevance, F-correlation calculation: wrapper ways to attain the most effective potential If (Fi ∈ F) then target notion C is treated as performance with a selected learning algorithmic rule T-Relevance. If (Fi, Fj ∈ F ^ i≠j) is named with similar time complexness of the filter ways. F-correlation. T-Relevance among a feature and Hierarchical clustering has been implemented the target notion C, the correlation F-Correlation in word choice within the context of text between a combine of features, the feature classification. And it is noise-tolerant and strong to redundancy F-Redundancy and therefore the feature communications, additionally as being representative feature R-Feature of a feature ISBN NO :relevant 378 - 26 - 13840 9 INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT for binary or continuous knowledge solely.23 cluster will be outlined. However, it does not discriminate between redundant
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
system the above thought are taken into account for 3. MST construction by fuzzy logic : We adopt the minimum-spanning tree (MST) developing the projected system. The core part of the clustering way in competence view. during developing projected sector considers and totally this method we calculate a neighborhood survey all the necessary needs for creating the project. graph of occurrences, then take away any For every project Literature survey is the most vital edge in the graph that is a lot shorter/longer sector in code development procedure. Preceding to (by fuzzy logic) than its neighbors. developing the tools and the associated planning it is necessary to decide and survey the time facet, 4. Relevant feature calculation: When removing all the unnecessary spare resource constraint, man power, financial system, and edges, a Forest will be obtained. In that every tree company strength. Once these items are fulfilled and represents a cluster. Finally it contains feature set and totally reviewed, then the subsequent step is to make then calculates the accurate/relevant feature. a decision concerning the code specifications within the relevant system such as what kind of operating 2.4 Problem Definition Many algorithms that illustrate a way to maintain the system the project would require, and what are all the knowledge into the database and the way to retrieve it essential code are required to proceed with the quicker, however the matter is no one cares about the subsequent step such as developing the tools and the database maintenance with ease manner and safe associated operations. methodology. The systems like Distortion and 3. FUZZY BASED FEATURE SET SELECTION congestion algorithmic rule, which makes an ALGORITHMS individual space for every and each word from the EFAST Algorithmic rule is a classic algorithm for already elected transactional database, those are put frequent item set mining and association rule learning together known as dataset, which is able to be over transactional databases. This EFAST appropriate for a collection of exacting words, algorithmic rule internally contains an algorithmic rule however it will be troublesome for the cluster of known as Apriori, which progresss by discovering the records, once the user get confused then they will frequent individual things in the database and never get the data back. The wrapper ways use the enlarging them to larger and well-built item sets as analytical accuracy of a predetermined learning long as those item sets seem sufficiently frequently in algorithmic rule to verify the goodness of the chosen the database. The common item sets confirmed by subsets, the correctness of the learning algorithms is Apriori will be accustomed to determine association usually high. Their computational difficulty is low, rules that highlight general trends in the database. however the correctness of the learning algorithms is 3.1 Feature set selection algorithmic rule not assured. An EFAST algorithmic rule analysis has In machine learning, statistics feature selection called targeted on sorting out relevant features. A famed as variable selection or attribute selection or variable example is Relief which weighs every feature in line set selection. Is the procedure of choosing a set of with its ability to discriminate instances below relevant features to be used in model creation. The completely different targets supported distance-based central hypothesis employing a feature selection criteria task. Though, Relief is unproductive at technique is that the data contains several redundant removing redundant features as too analytical or irrelevant features. Redundant features are those however extremely correlative features are seemingly which supply no more information than the presently each to be extremely weighted. Relief-F expands specific features, and irrelevant Features offer no Relief, permitting this methodology to work with helpful information in any background. Feature strident and incomplete data sets and to cope with selection ways are a set of the additional general field multiclass issues, however still unable to of feature extraction. Feature extraction creates new acknowledge redundant features. features from functions of the novel features, whereas feature selection returns a set of the features. Feature 2.5 Literature Review Literature survey is the most vital step in code selection techniques are usually employed in domains development procedure. Before finding the tool it is wherever there are several features and relatively few essential to decide the time facet, financial system and samples (or knowledge points). company strength. Once these things are satisfied, Feature selection techniques supply 3 main profits then the subsequent step is to decide which operating once constructing analytical models: system and language can be used for developing the • Improved model interpretability, tool. • Shorter training times, Once the programmers commence building the tool • Improved generalisation by reducing over fitting. the programmers require set of external support. This Feature selection is also useful as a part of the data ISBN NO :maintenance 378 - 26 - 13840 - will 9 INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT be obtained from programmers, 24 analysis method, as shows which features are vital for from books or from websites. Before building the prediction, and the way these features are connected.
Proceedings of International Conference on Developments in Engineering Research
3.2 Definitions In learning Machines [11], [15] Suppose to be the complete set of features, ∈ be a feature, = −{ } and ′ ⊆ . Let ’ be a value-assignment of all features in ′ , a value-assignment of feature , and a value-assignment of the target concept . The definition will be formalized as follows. Definition: (Relevant feature) has relevancy to the target concept if and only if there exists some ′, and , such that, for probability ( ′ = ′ , = )>0, ( = ∣ ′ = ′ , = ) ≠ ( = ∣ ′ = ). or else, feature is an irrelevant feature. There are 2 sorts of relevant features due to different ′ : (i) if ′ = , we will recognize that is directly relevant to the target concept (ii) if ′ ⊊ , we could get that ( ∣ , )= ( ∣ ). Definition: (Markov blanket) The definitions of Markov blanket and redundant feature are introduced as follows, correspondingly. let ⊂ ( ∕∈ ), is assumed to be a Markov blanket for if and only if ( − −{ }, ∣ , )= ( − −{ }, ∣ ). Definition: (Redundant feature) let be a collection of features, a feature in is redundant if and only if it has a Markov Blanket within . Relevant features have strong correlation with target concept so are always essential for a finest set, while redundant features are not since their values are fully correlative with one another. Definition: symmetric uncertainty ( ) is derived by normalizing the entropies of feature values or target categories. SU is the evaluation of correlation between either two features or a feature and the target concept. In existing system we tend to calculate SU as U( , )=2× ( ∣ ) / ( )+ ( ) Where ( )=− Σ ∈ ( )log2 ( ) n( ∣ )= ( )− ( ∣ ) n(Y∣X)= ( )− ( ∣ ) ( ∣ )=− Σ ∈ ( ) Σ ∈ ( ∣ )log2 ( ∣ ). In our paper we are proposing SU as U( , )=2× Ratio( ∣ ) / ( )+ ( ) Where ( )=− Σ ∈ ( )log2 ( )
Intrinsic Information is the entropy of distribution of instances into branches.
www.iaetsd.in
Definition: (F-Correlation) The correlation between any pair of features and ( , ∈ ∧ ≠ ) is named as the F-Correlation of and , and denoted by ( , ). Definition: (F-Redundancy) Let = { 1, 2... ... <∣ ∣} be a cluster of features. if ∃ ∃ , ( , ) ≥ ( , ) ∧ ( , ) > ( , ) is often corrected for every ∈ ( ≠ ), then are Redundant features with reverence to the given . Definition: (R-Feature) A feature ∈ ={ 1, 2, ..., } ( < | |) is a representative feature of the cluster ( i.e. is a R-Feature ) if and only if, = argmax ∈ ( , ). By this we will say a) Irrelevant features have no/weak correlation with Target concept b) Redundant features are assembled in a cluster and a representative feature will be taken out of the Cluster. 3.3 EFAST Algorithm by Gain Ratio Algorithm: EFAST Inputs: D(f1,f2,…fm, C) – given data set Output: S – Feature Selection // irrelevant feature removal For i=1 to m do T-Relevance=SU(Fi, C) //SU is calculated based on Gain Ratio If T-Relevance >Ɵ then // here Ɵ is the Threshold value S= S U {Fi } // MST Construction G= NULL For each pair of features { fi, fj } ⊂ S do F-Correlation = SU(fi, fj) Add fi and/or fj to G with the F-Correlation as its weight. Minspantree = Prim(G); //using Prim algorithm to construct MST // clustering and Enhanced Feature Selection Forest = Minspantree For each edge Eij ∈ Forest do If SU(fi, fj) < SU(fi ,C) ^ SU(fi, fj) < SU(fi ,C) then Forest= Forest - Eij S= For each tree Ti ∈ Forest do FjR = argmax Fk Ti SU(Fk, C) S= S U { FjR } Return s
( ∣ )=− Σ ∈ ( ) Σ ∈ ( ∣ )log2 ( ∣ ). Definition: (T-Relevance) The relevance between the feature ∈ and the target concept is consigned as The T-Relevance of and , and denoted by U( , ). If U ( , ) is > a predetermined threshold ISBN NO : 378 - 26say - 13840 INTERNATIONAL OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT , we that- 9 is a strong T-Relevance feature.ASSOCIATION 25
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
3.4 About Gain Ratio Information is measured in terms of bits Given a probability distribution, the info requisite to predict an event is the distribution’s entropy Entropy offers the knowledge needed in bits (this will involve fractions of bits!) We can calculate entropy as entropy( p1 , p 2 , , p n ) p1logp1 p 2 logp 2 p n logp n
Gain ratio: a modification of the information gain that reduces its bias on high-branch attributes Gain ratio ought to be Large when knowledge is equally spread Small when all knowledge belong to 1 branch Gain ratio takes range and size of branches under consideration when selecting an attribute It corrects the information gain by taking the intrinsic information of a split under consideration (i.e. what quantity data can we have to be compelled to tell that branch and instance belongs to) We can calculate Gain Ratio as
Intrinsic information: entropy of distribution of instances into branches Gain ratio (Quinlan’86) normalizes info gain by:
Fig. 1: Framework of the Fuzzy Based The outline of Frame work is characterized as
Data Sets
Irrelevant Feature Removal
We Use Gain Ratio for Effectiveness Ex: gainratio (“Attribute”) =gain (“Attribute”) Intrinsicinfo (“Attribute”) Ex: gainratio (“Id”) =0.94/3.8=0.24
4 Frame Work Our projected feature set selection framework structure involves irrelevant feature removal and redundant feature elimination by using fuzzy logic. It offers internal logical schema to form clusters with the assistance of EFAST Algorithmic rule. Smart feature subsets contain features extremely correlative with the class, yet unrelated with each other. [20] Frame work Analysis, it involves (i) building the minimum spanning tree (MST) from a weighted complete graph (ii) the partitioning of the MST into a forest with every tree representing a cluster and (iii) the selection of representative features from the clusters. ISBN NO : 378 - 26 - 13840 - 9
MST Construction
Clustering of MST
Enhanced Feature Selection
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT Fig. 2: EFAST Architecture 26
Proceedings of International Conference on Developments in Engineering Research
5 Algorithmic Rule Analyses The planned EFAST algorithmic rule rationally consists of 3 steps: (i) Removing irrelevant features, (ii) Constructing a MST from relative ones, (iii) Separating the MST and selecting Representative features. For a data set with features = { 1, 2... } and class , we tend to compute the T-Relevance ( , ) value for every feature (1 ≤ ≤ ). In the first step. The features whose ( , ) values are larger than a predefined threshold contain the target-relevant feature subset ′ = { ′1, ′2... ′ } ( ≤ ). In the second step, we 1st calculate the F-Correlation ( ′ , ′ ) value for every pair of features ′ and ′ ( ′ , ′ ∈ ′∧ ≠ ). Then, viewing features ′ and ′ as vertices and ( ′, ′ ) ( ≠ ) as the weight of the edge between vertices ′ and ′ , a weighted complete graph = ( , ) is built wherever = { ′ | ′i∈ ′ ∧ ∈ [1, ]} and = {( ′ , ′ ) ∈ ( ′ , ′ ∈ ′ ∧ , ∈[1, ] ∧ ≠ }. In the third step, we 1st eliminate the edges = {( ′ , ′ ) ∈ ( ′ , ′ ∈ ′ ∧ , ∈ [1, ] ∧ ≠ }, whose weights are smaller than both of the T-Relevance ( ′ , ) and ( ′ , ), from the MST. Each removal ends up in 2 disconnected trees 1 and 2. The complete graph reflects the correlations among all the target-relevant features. If graph has vertices and ( −1)/2 edges. For this we build a MST, which connects all vertices such that the sum of the weights of the edges is the minimum, using the famous Prim algorithmic rule [14].The weight of edge ( ′ , ′ ) is F-Correlation ( ′ , ′ ). If ( ′ , ′ ∈ ( )), ( ′, ′) ≥ ( ′, ) ∨ ( ′ , ′ ) ≥ ( ′ , ) this property assurances the features in ( ) are redundant. Suppose the MST shown in Fig.3 is generated from a complete graph . So as to cluster the features, we 1st pass through all the six edges, and then decide to remove the edge ( 0, 4) as its weight ( 0, 4) = 0.3 is smaller than both ( 0, ) = 0.5 and ( 4, ) = 0.7. This makes the MST is clustered into 2 clusters denoted as ( 1) and (T2).
www.iaetsd.in
between
F0,F2
and
F0,F3
and
F2,F3.
Fig.3: Clustering Step
After removing all the redundant edges, a forest is achieved. Every tree Tj ∈ Forest denotes a cluster that is signified as V(Tj). As examined above, the features in every cluster are redundant, thus for every cluster V (Tj) we tend to like a representative feature FjR whose T-Relevance SU (FjR,C )is the high. All FjR (j = 1...|Forest|) comprise the final feature subset ∪ FjR. 5.1 Time complexity analysis. The computation of SU values for T-Relevance and F-Correlation, that has linear complexness in provisos of the number of instances in a given data set. The first part of the algorithmic rule includes a linear time complexness O (m) as a result of the number of features m. If k (1 ≤ k ≤ m) features are elected as relevant ones in the 1st half, when k = 1, only 1 feature is chosen. Therefore, there is no need to continue the rest parts of the algorithmic rule, and therefore complexness is O(k). If 1 < k ≤ m, the second part of the algorithm first of all constructs a complete graph from relevant features and therefore complexness is O(k2), and then MST complexness is O(k2). The third part partitions the MST and chooses the representative features with the complexness of O(k).
Thus when 1 < k ≤ m, the complexity of the algorithmic rule is O(m+k2). This means when k ≤√m, EFAST has linear complexness O(m), where as the worst complexness is O(m2) when k = m. However, k is drastically set to be lower bound of √m*lg m in the implementation of EFAST. Therefore the From Fig.3 we identify that ( 0, 1) > ( 1, ), complexness is O (m * lg2m), which is usually less ( 1, 2) > ( 1, ) ∧ ( 1, 2) > ( 2, ), than O (m2) since lg2m < m. it makes EFAST has an F3) >SU(F3,ASSOCIATION C). enhanced ISBN NO :SU(F1, 378 - 26 - F3) 13840>- 9SU(F1, C) ∧ SU(F1, INTERNATIONAL OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT runtime performance with high dimensional 27 We also recognized that there is no edge exists knowledge.
Proceedings of International Conference on Developments in Engineering Research
6 knowledge Source To evaluate the performance and effectiveness of EFAST algorithmic rule we are using publicly existing data sets. The numbers of features of the data sets vary from 35 to 49152 with a mean of 7874. The dimensionality of the 53.3% data sets exceeds 5000, of which 26.6% data sets have more than 10000 features. The data sets cover a collection of application domains like images, text and microarray data categorization. Table 1: sample benchmark data sets __________________________________________ Data ID Data Name F I T Domain __________________________________________ 1 chess 37 3196 2 Text 2 mfeat-fourier 77 2000 10 Image, Face 3 coil2000 86 9822 2 Text 4 elephant 232 1391 2 Microarray, Bio 5 tr12.wc 5805 313 8 Text 6 leukemia1 7130 34 2 Microarray, Bio 7 PIX10P 10001 100 10 Image, Face 8 ORL10P 10305 100 10 Image, Face __________________________________________
www.iaetsd.in
7.2 Loading Data Set
7.3 Calculating Entropy, Gain And GainRatio
7 Results and Analysis 7.1 Main Form
7.4 calculating Attributes
T-Relevance
and
Relevant
Now we have to upload any dataset i.e. Here we are uploading “chess” data set.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 28
Proceedings of International Conference on Developments in Engineering Research
7.5 calculating F correlation
www.iaetsd.in
7.8 MST using InformationGain
7.9 clustering Inf.Gain Based MST
7.6 Generating MST
7.10 MST using Gain Ratio
7.11 clustering Gain Ratio Based MST
7.7 Relevant Feature Calculation
By Observing 7.8 to 7.11 we can say that Gain Ratio makes effective clustering than Information Gain as proposed in the existing system. We can represent the same thing in graphical form also. In 7.12 we are showing the graphical chart illustration of information gain vs. gain ratio. From the above discussions and experimental results we can conclude that Gain Ratio gives Enhanced Feature Selection than Information Gain.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 29
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
7.12 Graphical Representation of Inf.Gain Vs Gain Ratio Analysis on Chess Data Set Construction of MST using Inf.Gain
Construction of MST Clustering using Inf.Gain
Construction of MST using Gain Ratio
Construction of MST Clustering using Gain Ratio
Construction of MST Clustering using Inf.Gain
Construction of MST Clustering using Gain Ratio
ISBN NO : 378 - 26 - 13840 - 9
8 Conclusions In this paper, we have presented a completely unique relevant clustering-based EFAST algorithmic rule for high dimensional knowledge. The algorithmic rule involves a) removing unrelated features, b) building a MST from relative ones, and c) divisioning the MST and selecting representative features. a cluster consists of features. Each cluster is referred as a single feature and thus dimensionality is severely reduced. The proposed algorithm gets the most fractions of chosen features, the enhanced runtime, and the finest classification accuracy. For the future work, we plan to find out different types of correlation calculations, relevance measures and revise some formal properties of feature space. ACKNOWLEDGEMENTS The authors would be fond of to the editors and the anonymous commentators for their intuitive and helpful observations and suggestions that resulted in significant developments to the current work.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 30
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
[15] Das S., Filters, wrappers and a boosting-based REFERENCES [1] Almuallim H. and Dietterich T.G., Algorithms for hybrid for feature Selection, In Proceedings of the Identifying Relevant Features, In Proceedings of the Eighteenth International Conference on 9th Canadian Conference on AI, pp 38-45, 1992. MachineLearning, pp 74-81, 2001. [2] Almuallim H. and Dietterich T.G., Learning [16] Dash M. and Liu H., Consistency-based search boolean concepts in the presence of many irrelevant in feature selection. Artificial Intelligence, 151(1-2), features, Artificial Intelligence, 69(1-2), pp 279-305, pp 155-176, 2003. 1994. [17] Demsar J., Statistical comparison of classifiers [3] Arauzo-Azofra A., Benitez J.M. and Castro J.L., over multiple data sets, J. Mach. Learn. Res., 7, pp A feature set measure based on relief, In Proceedings 1-30, 2006. of the fifth international conference on Recent [18] Dhillon I.S., Mallela S. and Kumar R., A divisive Advances in Soft Computing, pp 104-109, 2004. information theoretic feature clustering algorithm for [4] Baker L.D. and McCallum A.K., Distributional text classification, J. Mach. Learn. Res., 3, pp clustering of words for text classification, In 1265-1287, 2003. Proceedings of the 21st Annual international ACM [19] Dougherty, E. R., Small sample issues for SIGIR Conference on Research and Development in microarray-based classification. Comparative and information Retrieval, pp 96-103, 1998. Functional Genomics, 2(1), pp 28-34, 2001. [5] Battiti R., Using mutual information for selecting [20] Fayyad U. and Irani K., Multi-interval features in supervised neural net learning, IEEE discretization of continuous-valued attributes for Transactions on Neural Networks, 5(4), pp 537-550, classification learning, In Proceedings of the 1994. Thirteenth International Joint Conference on [6] Bell D.A. and Wang, H., A formalism for Artificial Intelligence, pp 1022-1027, 1993. relevance and its application in feature subset selection, Machine Learning, 41(2), pp 175-195, 2000. [7] Biesiada J. and Duch W., Features election for high-dimensionaldataĹ&#x201A;a Pear-son redundancy based filter, AdvancesinSoftComputing, 45, pp 242C249, 008. [8] Butterworth R., Piatetsky-Shapiro G. and Simovici D.A., On Feature Se-lection through L.Anantha Naga Prasad M.Tech, Clustering, In Proceedings of the Fifth IEEE Computer Science and Engg, Anantha Lakshmi Institute of international Conference on Data Mining, pp Technology&Sciences, JNTUA, Andhra Pradesh, India. 581-584, 2005. naga4all16@gmail.com. His current research interests [9] Cardie, C., Using decision trees to improve include data mining/machine learning, information case-based learning, In Proceedings of Tenth retrieval, computer networks, and software International Conference on Machine Learning, pp engineering. 25-32, 1993. [10] Chanda P., Cho Y., Zhang A. and Ramanathan M., Mining of Attribute Interactions Using Information Theoretic Metrics, In Proceedings of IEEE international Conference on Data Mining Workshops, pp 350-355, 2009. [11] Chikhi S. and Benhammada S., ReliefMSS: a variation on a feature ranking Relief algorithm. Int. J. Bus. Intell. Data Min. 4(3/4), pp 375-390, 2009. [12] Cohen W., Fast Effective Rule Induction, In K.Muralidhar Assistant Professor, Proc. 12th international Conf. Machine Learning Department of CSE, Anantha Lakshmi Institute of (ICMLâ&#x20AC;&#x2122;95), pp 115-123, 1995. Technology&Sciences, JNTUA, Andhra Pradesh, India. [13] Dash M. and Liu H., Feature Selection for muralidhar.kurni@gmail.com. His current research Classification, Intelligent Data Analysis, 1(3), pp Interests include data mining/machine learning, 131-156, 1997. computer networks, and software engineering. [14] Dash M., Liu H. and Motoda H., Consistency based feature Selection, In Proceedings of the Fourth ISBN NO :Pacific 378 - 26 - Asia 13840 -Conference 9 INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT on Knowledge Discovery 31 and Data Mining, pp 98-109, 2000.
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Improving the Location of Nodes in Wireless Ad Hoc and Sensor Networks Using Improvised LAL Approach P. Saravanan1, Er S. Muthukumar2 , Dr.K.Rubasoundar 3 1
Department of Computer Science and Engineering, M.E. Scholar, Anna University, Sree Sowdambika College of Engineering, Aruppukpttai, TamilNadu, India. saravananselva007@gmail.com 2 Department of Computer Science and Engineering, Professor/Head of the Department, Anna University, Sree Sowdambika College of Engineering, Aruppukottai, TamilNadu,India. muthukumar_ssce@yahoo.com 3 Department of Computer Science and Engineering, Professor/Head of the Department, Anna University, PSR College of Engineering, Sivakasi, TamilNadu,India. rubasoundar@yahoo.com
Abstract: This Paper aims to present a localization of many nodes in wireless networks. Localization is an enabling technique for many sensor network applications. In the deployment of network includes many nodes, due to hardware or deployment constrains, the networks almost not entirely localizable, there is possible for any of the nodes placed as a non-localizable nodes in the wide range of network area. From that, the server doesn’t know where the destination nodes have been deployed. It is difficult to transfer the data from source to sink whether the deployed nodes is not in the range of particular network. So, I proposed a Improvised LAL approach that triggers a single round adjustment and that carries and beware of node localizability, that makes easy for us to make the non-localizable nodes in the network into localizable nodes. Index terms: Localization, localizability, approach, wireless and sensor networks.
Improvised
LAL
I. INTRODUCTION Beyond the established technologies such as mobile phones and WLAN, new approaches to wireless communication are emerging; one of them are so called ad hoc and sensor networks. Ad hoc and sensor networks are formed by autonomous nodes communicating via radio without any additional backbone infrastructure. Localization is the main problem in wireless ad-hoc and sensors networks in which each and every node determines its own location in network region.“If you board the wrong train, it is no use running along the corridor in the other direction”- Dietrich Bonhoeffer . As he says, wireless technology has the capability to reach any location on the earth. Defining the ad-hoc network in terms of network as an autonomous system of mobile hosts(MH’s),connected by wireless links, using that, the mobile hosts that connect with the base station. To locate non-localizable nodes, the traditional approach mainly focuses on how to tune network settings according to these nodes. At first they attempt to deploy additional nodes or beacons in application fields. Beacons are act as a backbone for our network. Due to this increment in the ISBN NO : 378 - 26 - 13840 - 9
deployment of node density and creates abundant internodes distance constraints thus, enhancing the localizability. But this attempt lacks to provide feasibility, since the additional nodes and beacons should be placed in the region of nonlocalizable nodes, in the network. The controlled motion of beacons provides thorough information for the localization of nodes, but it has a limitation on adjustment delay and controlling overheads. One approach is to augment that is to make the greater in transmitting power of nodes stage-by-stage until all nodes become localizable, which causes multiple rounds of configuration dissemination and data collection in a network. A straight-forward single-round configuration solution is maximizing the ranging capability of the network regions. The drawback of power maximization is that it introduces many unnecessary distance measurements, which are obtained with costs. In this paper, I propose an Improvised localizability-aided localization (LAL) that arrogates sufficient condition of node localizability to identify localizable and non-localizable nodes. II. RELATED WORK The localization methods can be categorised as a rangefree and range-based methods. Range-free localization methods, merely use neighbourhood information (such as node connectivity and hop count) to determine node locations. Range-based approaches assume nodes are able to measure internodes distances, from that we can derive the accurate locations of the nodes. There are number of rangebased algorithms are used to find out the accurate location of the nodes, these algorithm adopt distance ranging techniques, such as radio signal strength (RSS) and time difference of arrival (TDoA).The RSS maps received signal strength from the distance between two nodes like a signal extinction model, while TDoA measures the signal propagation time for distance calculation. One can see that naive approaches for localization are not adequate for all scenarios. While it may be possible to manually configure each node with its position
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 32
Proceedings of International Conference on Developments in Engineering Research
in small and static networks, this approach is impracticable in envisioned large-scale networks of thousands of geographically distributed nodes. From the past, a few works and techniques are done on localization in non-localizable networks. In previous technique such as range-free localization the cop count between two nodes are proportional to their distance. In range-based localization technique that measure the accurate distance between two nodes using specific ranging hardware. A technique is a GPS-equipped mobile node to localize fixed nodes by measuring the distance between the mobile node and fixed ones, which use a mobile node with known location information. Wu- proposes a similar approach with the assumption that each node can move around and measure the distances to its neighbors and the relative distances between successive positions along its route. However, the availability of mobile nodes is much more costly and not scalable for this approach. In contrast to this mobility based approaches, Anderson-propose a graph manipulation method to assure the network localizability. A distributed range free localization scheme is used in recent years for localization in nonlocalizable networks. In DRLS a node called anchors, get their own location information via GPS or some other mechanism. The other nodes called normal nodes that do not have its own location information a few algorithm has been used to estimate the nodes location information effectively. III. FINE-GRAINED APPROACH 3.1
Formation of Network Topology Here a network has been formed with the help of dynamic topology due to a mobility of the nodes in the network. Setting up and organizing such a virtual infrastructure is an important challenge. The inherent tradeoff between energy-efficiency and rapidity of event dissemination is characteristic for wireless sensor networks consisting of battery driven devices. Localization in wireless ad hoc and sensor networks is the main problem for every node to determines its own location, a wireless ad hoc or sensor network cannot be ridiculously dense because the mechanism of topology control is usually used to reduce the collision and interference, ignoring the localizability of thenetwork.Thetopology of these networks often plays a crucial in the speed with which certain tasks can be accomplished using these networks.
Figure 1 Formation of network
ISBN NO : 378 - 26 - 13840 - 9
3.2
www.iaetsd.in
Localization Method
In localization method, first attempt to deploy a network with corresponding topology that described in previous section how to deploy a network with multiple nodes. Then we are going to measure the deployed nodes location whether it is, within the range of network or not, there is a main problem to find out the locations of each and every node in the network. From the fig 1, the localizable nodes are shown within the network region and nonlocalizable nodes are indicated as red dots, are out of region of our network. If there is an additional nodes, which is going to be deployed in the network that increases in node density and creates abundant internodes distances constraints, this enhancing the node localizability. 3.3
Measurements on Network Using Distance Graph The network is said to be a unique and localizable, then it must have a unique set of rigidity and distance graph and also the set of anchors such of beacon nodes. Whether a formed network cannot be localizable given its distance graph, then it is called as a localizability problem. Some of the method used to deploy a distance graph using mathematical formation. The distance graph is formed from a collection of points in the Euclidean space. The set of measurements for the network can be modelled by a distance graph G, let G = (V,E),where V denotes the set of vertices and E denotes the set of edges. For (i,j) € V ;and (i,j) € E if the distance between i and j can be measured or both of them are in known locations, are exploited by a graph theory 3.4
Construction of Localizable Graph It is important to construct a localizable graph through an incremental construction. A common mathematical formation can also be used to construct the localizable graph. From the graph theory, Define G2 as (V,E U E2), where (i;j)€E2 if i≠j and ᴟk€2 V such that (i,k) and (j,k)€E. Similarly, define G3 as (V,EUE2UE3), where (i,j)€E3 if i≠j and ᴟk€V such that (i,k)€E and (j,k)€E2. IV. DESIGNING OF PROPOSED SYSTEM 4.1
Improvised Lal Design And Implementation
Design phase of Improvised LAL has been begun with being aware of node localizability, Improvised LAL can effectively conduct the adjustment of network ranges, but in traditional approach that is only an indistinctive network adjustment can be made out and could only make indistinctive augmentations. Basically, Improvised LAL consists of three major modules and is workflows are shown in Fig. 1. In improvised LAL approach, first the network has been deployed by using the network topology is described in previous sections. After the network formation, the proposed techniques are explored in the network and the nodes in the network are act depend upon the described techniques. After the explosive of all techniques has been finished the nodes are try to send the packet and location information. There is a large number of nodes has been deployed in the network, so the density of the network much more increased and each
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 33
Proceedings of International Conference on Developments in Engineering Research
node doesnâ&#x20AC;&#x2122;t know the location information of other nodes. An Improvised LAL approach used here to know the location information of all other nodes. Processing and work flows of Improvised LAL approach are explained below, Module 1: Localizability testing. In Localizability testing, after the network is deployed in an application field, due to some hardware or environmental factors that is for unpredictable issues in the design phase, it may not ready for localization. So, node localizability testing is conducted in an Improvised LAL, which identifies localizable and nonlocalizable node in a network for further adjustment. Module 2: Analysis of Network structure. In an analysis of the structure of the network must support fine-grained approach, to measure the accurate location information about the entire node. So, we have to decompose a constructed distance graph into two-connected components. These components are managed as a tree structure and the one of these components containing beacons in the root of the distance graph. Adjustments are made out along with the tree edges from the root to the leaves. Beacons are mainly used to find out the Localization and improve the accuracy of the nodes in the network.
www.iaetsd.in
rigidity, which is being tested in polynomial time by network flow algorithms and the pebble game algorithm, respectively. That is, a node localizability testing can be conducted. In Module 2, a created distance graph is decomposed into twoconnected components using depth-first search. And these components are managed as a tree structure and the adjustments can be made in these tree edges. In module 3, node adjustments are made out along the paths of the component tree starting at the root. I present an Algorithm (Basic _Localization _Algorithm) to explain the module 3. From that, the node will know the location information of all other nodes in the network. In the network topology, the nodes are formed as a cluster and a cluster has a cluster head, a cluster head have location information about all of the other nodes in the cluster. If a node placed in cluster A want to send a packet to a other node that placed in cluster B, then it will send a request to cluster A, a cluster head A have the location information about all of these node and where it is deployed then redirect the request to cluster head B, the B send the packet to the corresponding node that formed under that particular cluster 4.2
Add_Heuristic Approach
Figure 2 Decomposition of a Graph
Module 3: Distinctive adjustment. In Improvised LAL treats nodes differently regarding to their localization and places in the component tree. Through vertex augmentation, Improvised LAL converts all non-localizable in a single round. The network adjustments made out by Improvised LAL are localizable and can be localized by the existing localization and localizability techniques.
Figure 3 Workflows between traditional and Improvised LAL approach,
Module 1 can be done by applying Theorem 2(shown in below). For a given specific node, its localizability that depends upon the property of disjoint paths and redundant ISBN NO : 378 - 26 - 13840 - 9
Figure 4 to find out all non-localizable neighbors of localizable
In Add_heuristic approach, find out all non-localizable neighbors of localizable vertices. From the above figure 4, the edges have been added for each non-localizable node to convert these nodes as a localizable node. The edges have been added by heuristic algorithm to connect these nonlocalizable nodes for the purpose of delivering the data from localizable nodes from non-localized nodes in the deployed network. Algorithm for Add_heuristic is described in a later section. 4.3 Geographical Routing Approach Geographical routing is an emerging technique to find out the direction of any nodes in the certain region of any network. A 2D graph has been deployed and it has x-axis and y-axis using that a nodes direction has been found out. Basically a network has been deployed and measured by a distance graph. If a sender node try to deliver a packet to a receiver node but it does not know the location information of the receiver nodes, so using this approach the location that is the direction of the node has been found out. Here a beacon signals are act as a backbone of our network The fig 5 shows that, a sender collects the nodes location information based on the direction of the nodes, and finally they select a shortest
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 34
Proceedings of International Conference on Developments in Engineering Research
path from the all selected paths, from that shortest path they sent the data packet to the corresponding receiver.
Figure 5 Geographical Routing
4.4 Ease of Use Description of an Algorithm A Basic _Localization _Algorithm, applied here for the purpose of localize all the nodes in the network without any incompatibility. In that algorithm, edges are added by vertex augmentation of all non-localizable node vertices in GA, that is Graph A and GA is localizable from the Theorem 3(shown in below). This Algorithm repeats all of the steps, until all components get handled. After applying of an Algorithm, the entire network is localizable; all of node in the network gets localized. Popular localization algorithms can then be used seamlessly to localize all nodes in the network. The repeated edges has been reduced by analyze the graph properties of these components.
www.iaetsd.in
between the vertices, which connect two neighbor vertices on different vertex-disjoint paths to the non-localizable vertex, is enough to make the vertex localizable according to Theorem 2. From the analysis from the above graph, If a nonlocalizable vertex has more localizable neighbours within two-hop distance, that is, the distance between these neighbours are connected with only two edges, then, these localizable vertices makes it non-localizable as a localizable. Some of the decomposed components in a distance graph are not localizable due to the lack of beacons. If these nodes are adjusted to be localizable in a decomposed component, the component is instantly localizable without extra manipulation. V. THEOREMS Theorem 1. A graph with n≥vertices is globally rigid in two dimensions if and only if it is three-connected and redundantly rigid. A graph G=(V,E) is called k-connected (for k€IN) if | | > k and G-X is connected for every set X ⊆ V with | |< k. Theorem 2. In a distance graph G=(V,E) with a set B⊆V of k≥3 vertices at known locations, a vertex is localizable if it is included in the redundantly rigid component inside which there are three vertex-disjoint paths to three distinct vertices in B. Theorem 3. Suppose G=(V,E) is a two-connected graph with a set B of k≥3 beacons and B∋V . Let VN denote the set of non-localizable vertices, and EN denote the set of edges (i,j),i€VN and (i,j)€E2. Then, G=(V ,E U EN ) is localizable. VI. DISCUSSION The network topology is a by product of some basic services in ad hoc and wireless sensor networks,
Figure 6 proofs of theorem 3, red dots and black holes
Algorithm 2.Add_Heurictic _Algorithm. If GA is used as an input of Add_Heuristic, the algorithm first finds all non-localizable neighbors of localizable vertices, and adds edges for each non-localizable one according to the following analysis made out from the distance graph. Then add edges to all non-localizable vertices from the localizable vertices. After that, the adjustments are made out continuously to localize these non-localizable nodes by adding edges from non-localizable vertices to localizable vertices. 4.5 Analysis from the Graph From the above graph GA some non-localizable vertices have at least one localizable neighbour vertex. Adding two edges ISBN NO : 378 - 26 - 13840 - 9
Figure 7 GreenOrbs system in a campus
Its collection induces none or little additional overhead. Inspired by this fact, this paper naturally adopts the centralized scheme, which mainly comes from the localizability testing part. The fig 7 shows the wireless sensor nodes that scattered the signals to get the location information of other nodes, which is intended to send the data packets from source to sink and vice versa. The rigidity and beacons are used as backbone of our network to transform the data’s from source to destination. A rigidity, they find the co-ordinates and transforms these relative coordinates to a global co-ordinates,
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 35
Proceedings of International Conference on Developments in Engineering Research
Table 1 Experimental results on Greenorbs LAL_Basic
Add_heuristic
Indistinctive adjustment
No of edges
340
309
736
No of added edges No of power 1 nodes No of power 2 nodes No of power 3 nodes
55
24
451
61
73
0
39
27
100
0
0
0
Assume that noisy results (the outliers with large errors) are sifted by these approaches, and only used accurate ranging results in Improvised LAL design. VII. EVALUATION 7.1
Experiment To examine the correctness and effectiveness of Improvised LAL approach, I implement it on the ongoing wireless sensor network system, that is Green Orbs system and the data trace collected from the system Green Orbs. From the data collection the important factor is to reduce the energy consumption of the deployed nodes. Using, cycling theorem, the transmission power is also well controlled under the highest level to provide the enough connectivity for data collection or other services.
www.iaetsd.in
nodes that have original communication range as power1 nodes, the ones with doubled and tripled ranges are denoted as power2 and power3 nodes from doubled and tripled like wise From those following figures, the comparison made between the improvised LAL and traditional approach, to find out the location of the nodes. 7.2 Simulation Simulation is the process or operation that has a limitation among the real time implementation of original process. In this paper, I have to stimulate and further examined the scalability and efficiency of Improvised LAL under different network instances and varied network parameters.The transmission power requirements of LAL_Basic and AddHeuristic algorithms in deployed nodes are shown in fig.8. Plots the number of nodes at different power levels. As shown in fig.8a and 8b, results of Add_Heuristic and LAL_Basic, have much more power 1 nodes than the traditional approach except R denotes original communication range less than 0.66 and much less than power
Figure 8c High power levels at Improvised LAL intend to add more edges in the graph
Figure 8a Number of nodes plotted at different power levels
From the fig.9, there are 285 edges are needed to connect the edges between the nodes. A comparison can be made out between the edges and nodes with different transmitting power output from the resultant topology, I just denote the
3 nodes than traditional approach except when R greater than 0.99 that shows in fig.8b. The results in fig.8c also explain about the Improvised LAL that can have a ability to serve much extra edges than traditional approach because higher power levels tend to induce more edges than a lower one. Abbreviations used Traditional approachď&#x192; IND (Indistinctive Adjustment Approach) Improvised Traditional approachď&#x192; Improvised IND(Improved Indistinctive Adjustment Approach) Because Traditional approach treats the entire network indistinctly and obviously needs more adjustments than LAL, an improved approach (Improved IND) is implemented instead. From the above all description it suggests that LAL_Basic and Add_Heuristic are more fine grained than improved IND.
Figure 8b Number of nodes at different power levels in traditional approach
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 36
Proceedings of International Conference on Developments in Engineering Research
Figure
9
Testing
Improvised
LAL
and
Improvised
traditional
VIII. CONCLUSION The analysis of the limitation and power requirements of existing approach on localization in nonlocalizable networks, and propose a Improvised Localizability-aided Localization approach named Improvised LAL. Improvised LAL treats the network as whole and localize all the nodes in the network, while if it is in non-localizable state. That makes the adjustment corresponding to node localizability results in the first module, other than traditional approach, that includes the nodes in localizability testing and made a indistinctive adjustment. From that Improvised LAL approach, a nodes need to be augment with their ranging capability to connect and added new edges are needed to be measured. It also has some good characteristics for the purpose of implementation aspects in the real world application and I, implement the Improvised LAL and demonstrate its effectiveness through working system experiment and extensive simulations. IX. FUTURE SCOPE I, proposed the Adaptive Position Update strategy to address these problems. The APU scheme employs two mutually exclusive rules. The MP rule uses mobility prediction to estimate the accuracy of the location estimate and adapts the beacon update interval accordingly, instead of using periodic beaconing. The ODL rule allows nodes along the data forwarding path to maintain an accurate view of the local topology by exchanging beacons in response to data packets that are overheard from new neighbors, Then, We mathematically analyzed the beacon overhead and local topology accuracy of APU and validated the analytical model with the simulation results. X. [1]
[2]
REFERENCES
ISBN NO : 378 - 26 - 13840 - 9
approach [3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
Vaidyanathan Ramadurai,Mihail L. Sichitiu, ”Simulation-based Analysis of a Localization Algorithm for Wireless Ad-Hoc Sensor Networks”,Department of Electrical and Computer Engineering North Carolina State University Raleigh,NC27695,20/03,http://www4.ncsu.edu/~mlsichit/Research/Pub lications/opnet.pdf. Vaidyanathan Ramadurai,Mihail L. Sichitiu, ”Localization in Wireless Sensor Networks: A Probabilistic Approach” ,Department of Electrical and Computer Engineering North Carolina State University Raleigh, NC 27695,2003 http://www4.ncsu.edu/~mlsichit/Research/Publications/probabilisticLo calizationICWN.pdf.
www.iaetsd.in
[12]
[13]
on
network
instances
consists
of
400
nodes
Dragos ̧ Niculescu and Badri Nath, ”Ad Hoc Positioning System (APS) Using AOA”, DATAMAN Lab Rutgers University, IEEE Infocom 2003, http://infocom2003.ieee-infocom.org/papers/42_04.pdf. Sunil Jardosh, Prabhat Ranjan, “Intra-Cluster Topology Creation in Wireless Sensor Networks”, DA-IICT,Gandhinagar,India IEEE 2007, http://delab.csd.auth.gr/~papajim/presentations/files/04475768.pdf Jang-Ping Sheu, Pei-Chun Chen, Chih-Shun Hsu, “A Distributed Localization Scheme for Wireless Sensor Networks with Improved Grid-Scan and Vector-Based Refinement”, Department of Computer Science, National Tsing Hua University, Department of Computer Science and Information Engineering, National Central University, IEEE transactions on mobile computing, September 2008, http:www.researchgate.net/...A_Distributed_Localization_Scheme_for_ Wireless sensors networks. Wint Yi Poe,Jens B. Schmitt, “Node Deployment in Large Wireless Sensor Networks:Coverage, Energy Consumption, and Worst-Case Delay” disco | Distributed Computer Systems Lab University of Kaiserslautern, Germany,AINTEC’09, November 18–20, 2009, Bangkok, Thailand.https://disco.informatik.unikl.de/discofiles/publicationsfiles/AI NTEC09-1.pdf. Zhisu Zhu, Yinyu Ye, “Rigidity: Towards Accurate and Efficient Localization of Wireless Networks “Stanford University Stanford, Stanford, IEEE publications, infocom, 2010proceedingsconference,http://ieeexplore.ieee.org/xpl/freeabs_all.js p?arnumber=5462057&abstractAccess=no&userType=inst. Amitangshu Pal, ”Localization Algorithms in Wireless Sensor Networks: Current Approaches and Future Challenges”. Department of Electrical and Computer Engineering, The University of North Carolina, Network protocolandalgorithms2010,http://www.macrothink.org/journal/index.p hp/npa/article/viewFile/279/276. Zheng Yang and Yunhao Liu, ”Understanding Node Localizability of Wireless Ad-hoc Networks”, Hong Kong University of Science and Technology, IEEE Infocom proceedings2010,http://www.cse.ust.hk/~liu/localizability.pdf. Zheng Yang,Chenshu Wu,Tao Chen,Yiyang Zhao,Wei Gong,Yunhao Liu, “Detecting Outlier Measurements Based on Graph Rigidity for Wireless Sensor Network Localization” School of Software and TNList, Tsinghua University, Beijing 100084, ChinaNational University of Defense Technology, Changsha 410073, China,IEEE transactions on vehicular technology, vol. 62, no. 1, january 2013. Kuei-Li Huang·Li-Hsing Yen·Jui-Tang Wang·Chao-Nan Wu·ChienChao Tseng , “A Backbone-Aware Topology Formation (BATF) Scheme for ZigBee Wireless Sensor Networks”, National Chiao-Tung University, Hsin-Chu, Taiwan, Industrial Technology Research Institute, Hsin-Chu, Taiwan, ROC, Wireless Pers Commun (2013), http://download.springer.com/static/pdf/185/pdf. Jonathan Bachrach and Christopher Taylor, “Localization in Sensor Networks”, Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139 Quanjun Chen, Salil S. Kanhere, Mahbub Hassan, “Adaptive Position Update for Geographic Routing in Mobile Ad-hoc Networks”, The University of New South Wales,sydney,Australia.http://www.cse.unsw.edu.au/~mahbub/PDF_Pu blications/TMC_adaptive.pdf
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 37
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Survey on Wireless Sensor Networks Routing Protocols Based on Energy Efficiency Ms. S.NIVEDITHA1, Ms.M.USHA2 1 2
M.E., Department of Computer Science and Engineering, Velammal Engineering College, Chennai.
Assistant Professor, Department of Computer Science and Engineering, Velammal Engineering College, Chennai. unique.nive@gmail.com, umahalingam@gmail.com
Abstract-Energy consumption and balancing has always been a hot research topic in wireless sensor networks, in this survey, some of the hierarchical routing protocols(LEACH, HEED,PEGASIS, TBC, TREEPSI) with high energy efficient were examined based on some strategies such as data aggregation and clustering, routing and dynamic node allocation methods in the clustering environment. The routing protocols are compared with each other based on these strategies and based on the network life time definitions. Simulation results show which protocol has better prolonging lifetime than other traditional protocols by reducing the energy consumption and load balancing.
energy efficiency and prolong network lifetime in large scale WSN environment. The hierarchical routing protocol imply cluster based organization of the sensor node in order to perform data fusion and aggregation which leads to energy saving. In hierarchical network structure each cluster has a leader, which is called the cluster head (CH) which perform the data aggregation and data transmission to the base station by direct or intermediate communication with other CH nodes.
Key words: Load balancing, Data aggregation, clustering routing protocol.
I.
INTRODUCTION
Wireless sensor networks periodically collects the information of the interested unmanned monitoring area independently through deploying a large number of wireless sensor nodes and send the data to the remote base station.Wirelesssensornodesare deployedrandomlyand denselyina targetregion,especiallywherethephysicalenvironmentissohars hthatthesensor counterpartscannotbe deployed.Afterdeployment,thenetworkcannotworkproperlyun lessthereissufficientbatterypower. Wireless sensor nodes have so many important applications, including military application, disaster prediction and environment monitoring. The military application is performed in the rough environments where the sensor parts cannot deploy due to the energy consumption. Clusteringis very effectivetechnique.Itcan greatly contribute to overall system scalability, lifetime, and energy efficiencyinwirelesssensornetworks(WSNs).Grouping sensor nodes into cluster to satisfy the scalability and achieve high
ISBN NO : 378 - 26 - 13840 - 9
Energyconsumption ofanodeisduetouseful, wastefuloperations.The usefuloperationsincludestransmittingorreceivingdatamessag es,andprocessingrequests.Thewastefulconsumption includesconstructingroutingtree,overhearing, retransmittin gb e ca u s e ofharshenvironment, dealingwithredundantover- head messages,andidlelisteningto the media.Each node sends information directly to the base station to full fill the basic task of the sensor nodes. If the base station is located far away from the target area, the sensor nodes will die quickly due to much energy consumption. The distance is also very important factor and the direct transmission leads to the unbalanced energy consumption.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 38
Proceedings of International Conference on Developments in Engineering Research
In general WSN produce large amount of data so if data fusion could be used, the through put could be reduced. Data fusion reduces the redundant data by combining the redundant data to reduce the transmission by suppression, min, maxand average functions. Substantial energy can be saved through the data aggregation this technique has used to achieve energy efficiency and traffic optimization in a number of routing protocols. The data fusion or data aggregation assume that the length of message transmitted by each relay node should be constant. That is each transmits the same volume of data.
www.iaetsd.in
WSN is classified into three categories based on the structure: Flat, hierarchical, and location based routing.
The distance between the nodes to BS leads to two extreme cases 1.
2.
The length of message transmitted by the parent node not only depends on the own length also depends on the length of the message received from its child nodes. The relay node should transmit the length of the message which is the sum of its own sensed data and the received data
Network lifetime has two different definitions: [1] 1. 2.
The time from the start of the operation to death of the first node. The time from the start of the operation to death of the last node
Energy aware protocols and data gathering protocols offering high scalability. A common solution for the balancing energy consumption among all the network nodes, is to periodically re-elect new CH in each cluster. The remainder of the paper is presented as follow:Section IIdiscuss the literature survey. Finally, Section III concludes the paper. II. LITERATURE SURVEY A routingprotocol specifies the routers communicate with each other, and selects the routes between any two nodes on a computer network. Routing algorithms determine the specific choice of route. Each router has knowledge of networks attached to it directly. A routing protocol shares this information first among immediate neighbors, and then throughout the network. Routing strategy is generally based on the following criteria 1.
2. 3.
2.1Hierarchical protocols: Hierarchical routing is a method of routing in networks that is based on hierarchical addressing. It imposes the structure on the network to achieve energy efficiency, stability and scalability. In hierarchical protocols nodes are organized in clusters inwhich a node has some responsible takes the role of cluster head which is responsible for coordinating and message transmitting activities. Clustering reduces the energy consumption to prolongs the life time, can balance the load and increase the scalability. 2.1.1.
LEACH: [2][3]
LEACHisthe most energy efficientclusteringalgorithm forWSNs that forms nodes in clusterbased environment. Based onthe receivedsignal strengthand uses these localclusterheadsasrouters tothe BS.
Achieve minimum cost forwarding, while design of the most data forwarding protocol is based on the optimality. Reducing the minimum number of performed operations Scale to large network size with some constrains
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 39
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
by choosing the most appropriate cluster head (closest cluster head). In the steady-state phase the cluster heads receive data from the nodes, and transfer the fused data to the base station. And the communication is performed via the TDMA Advantages of LEACH: Data transferto theBSconsumesmoreenergy,all thesensor nodeswithina clustertaketurns withthe transmission by rotatingthe cluster heads. This leadstobalancedenergyconsumption ofallnodes,and hence a longerlifetime ofthe network The basic operations of LEACH was performed in two phases:
Setup phase: Cluster formation. Cluster head was selected and informed to all other nodes. Preparation of transmission schedule. Steady phase: Data fusion. Transmission to destination node
Inthe setup phase, each node n chooses a random number between 0 and 1. If the number is less than a threshold T (n), the node becomes a cluster head for the current round. The threshold is set as:
Disadvantages of LEACH:
T (n) = ___P______ if n belongs to G
1-P *(r mod1/P) Otherwise
0
P = percentage of cluster heads, r = current round
G = set of nodes G is not cluster head in the last 1/P rounds. After the cluster heads are selected, the other nodes organize the local clusters
ISBN NO : 378 - 26 - 13840 - 9
It performs most of the communication inside the clusters, and provides scalability. The data fusion leads to a limit on the traffic generated in the network. Single-hop routing from node to cluster head, hence saving energy. It increases network lifetime in three ways: Distributing the role of CH to the other nodes. Aggregating the data by the CHs. TDMA, which, assigned by the CH to its members, puts most of the sensor in sleep mode, in event-based applications. Itincreases the network lifetime and achieve a more than 7-fold reduction in energy dissipation compared to direct communication It is dynamic clustering and well-suited for applications where constant monitoring is needed and data collection occurs periodically to a centralized location.All thesensor nodeswithina clustertaketurns withthe transmission by rotatingthe cluster heads. This leadstobalancedenergy consumption ofallnodes,and hence a longerlifetime ofthe network.
Due to the selection of cluster heads. Because itdon’t consider the energy of the nodes for CH selection. It significantly relies on cluster heads and face robustness issues. Additional overheads due to cluster head changes leads to the energy inefficiency for dynamic clustering in large networks. It has no inter cluster communication, and this needs high transmission power. It selects the CH in random manner it results that the low power node can become a cluster head, it will deplete its energy quickly. And the local cluster head only by distance between themselves and it results the formation of minimum or maximum cluster.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 40
Proceedings of International Conference on Developments in Engineering Research
assigned to cluster heads among all n nodes, Cprob. Cprob is only used to limit the initial cluster head announcements, and has no direct impact on the final clusters.
Single hop method.
2.1.2.
www.iaetsd.in
HEED: [4]
HEED (Hybrid energy efficient distributed clustering) periodically selects cluster heads according based on the node residual energy and other parameter like proximity to node degree. HEED terminates in O (1) iterations, it incurs low message overhead and achieve uniform cluster head distribution across the network.
Before a node starts executing HEED, itsets its probability of becoming a cluster head, CHprob, as follows:
HEED was designed to select different cluster heads in a field according to the amount of energy that is distributed in relation to a neighboring node
Emax = reference maximum energy which is typically identical for all nodes. The CHprob value of a node, however, is not allowed to fall below a certain threshold pmin. Every node then doubles its CHprob and moved to next step.
The basic operations of HEED protocol was performed by the following steps:
Defining parameters formation. Protocol operation.
used
in
the
cluster
Cluster formation: Cluster head selection is based on the residual energy of each node. The parameters are measuring the residual energy and estimating the intracluster communication cost. The first parameter select initial set of cluster heads, the second parameter performing the cluster power level which increase initial reuse and reserve power for intercluster communication.
CHprob = Cprob * Eresidual / Emax Eresidual = estimated current residual energy in the node.
In HEED energy consumption it extends the lifetime of all the nodes, which adds to the stability of the neighbor set. Nodes automatically update their neighbor sets by periodically sending and receiving heartbeat messages. If a node elected as a cluster head, it sends an announcement message cluster_head_msg(Node ID, selection status, cost), where the selection status is set to tentative_CH if CHprob<1, or final_CH, if it’s CHprob = 1. A node considers itself “covered” if it is heard from either a tentative_CH. If a node completes HEED execution without selecting a cluster head that it considers itself uncovered, and announces itself to be a cluster head. Advantages:
Protocol operation: At each node, the clustering process requires a number of iterations, which was referred as Niter. Every step takes time tc to receive messages from any neighbor within the cluster range. An initial percentage was
ISBN NO : 378 - 26 - 13840 - 9
HEED distribution of energy extends the lifetime of the nodes within the network thus stabilizing the neighboring node.
Does not require special node capabilities, such as location-awareness
Does not make assumptions about node distribution
Operates correctly even when nodes are not synchronized.
The advantages of HEED are that nodes only require local (neighborhood) information to form the clusters
The algorithm guarantees that every sensors is part of just one cluster, and the cluster heads are welldistributed.
Disadvantages:
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 41
Proceedings of International Conference on Developments in Engineering Research
The random selection of the cluster heads, may cause higher communication overhead for:
The ordinary member nodes in communicating with their corresponding cluster head
Cluster heads in establishing the communication among them, orBetween a cluster head and a base station
The periodic cluster head rotation or election needs extra energy to rebuild clusters
In HEED Multi level hierarchy cannot be achieved. Recursive applications cannot be applied. 2.1.3.
PEGASIS: [5]
PEGASIS (Power-Efficient Gathering in Sensor Information Systems), is an optimal. In PEGASIS, each nodes communicates only with the close neighbor and transmits the information to the base station, it reduces the amount of energy spent per round. It distributes the energy load evenly among the sensor nodes in the network. The sensor nodes are randomly placed in the play field. The nodes are organized to form a chain, which was accomplished by the sensor nodes using a greedy algorithm starting from some node. On the other hand the BS can compute this chain and broadcast it to all the sensor nodes. Construction of the chain starts with the farthest node from the BS. The nodes farther from the BS have close neighbors, as in the greedy algorithm the neighbor distances will increase gradually. When a node dies, the chain will be reconstructed and the threshold can be changed to determine to elect leaders. In it (i mod N) th node is chosen to be a leader.
www.iaetsd.in
Data gathering in each round each node receives data from neighbor, fuses with its own data, and transmits to the other node in chain.The leader was selected in each round of communication in a random manner and random position on the chain, because of the death of nodes at random locations. Token passing approach initiated by the leader to start the data transmission from the ends of the chain. It performs data fusion at every node except the end nodes in the chain. Each node will fuse its neighbor’s data with its own to generate a single packet of the same length and then transmit that to its other neighbor (if it has two neighbors). Advantages of PEGASIS are 1. 2. 3. 4.
Local gathering distance is less than LEACH The amount of data for the leader to receive Limiting the number of transmissions and receives among all nodes. It eliminates the overhead of dynamic cluster formation, minimizing the distance non leader-nodes must transmit.
Disadvantages: Whenevera node dies the whole chain has to be reconstructed and the scalability is also an important issue in PEGASIS. 2.1.4.
Tree Based Clustering: [6]
Tree-Based Clustering (TBC) is also have cluster based environment and each cluster has a cluster-head (CH).
Tree configuration:The cluster construct a routing tree
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 42
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
where the cluster-head acts as a root. All sensor in a routing tree sends data to a single collection point know as root.
In GSTEB (General self-organized tree based energy balancing routing protocol) the basic tree construction and communication operation is similar to the TCB.
All the nodes arelocation-aware. The distance between the node and the root is for determining its level in the cluster. The nearest node is elected asitsparentnode.
In each round BS assigns a root node and broadcasts its ID and its coordinates to all sensor nodes. And data aggregation takes place.GSTEB changes the root and reconstruct the routing tree with short delay and low energy consumption by this betterbalancedloadis achieved.
Data transmission: Data transmission performed simultaneously. Data is gathered and each node fuses the received data and transmits it to its parent. InTBC each node holds the information of its neighbors. Fast data collection is main advantage in TBC 2.1.4.1. TREEPSI:[7] In TREEPSI sensor nodes are deployed randomly node i (ith node) is placed at a random location. Compute the path by BS and broadcast the path information to other nodes. After the construction of the tree it collects information from the field. TREEPSI is a tree based multi-hop routing protocol to construct a hierarchical path of the nodes. In this nodes will fuse the received data with their own data and forward the resultant data to their parent. It will repeat this process till all the data are received by the root node. The data are collected at the root and at last root takes responsibility to transmit the datatotheBS.Afterthedeathofnode, anewtreepathisconstructed. Sothe overhead per communication round is less as compared to the energy spent in the data collection phase.
The operation of GSTEB is divided into Initial Phase 1. 2. 3. 4.
Tree ConstructingPhase Self-OrganizedDataCollecting Transmitting Phase Information Exchanging Phase.
EL(i)=[residual
energy(i)/constant
reflect the
minimum energy unit] GSTEBoutperformsLEACH, PEGASIS, TREEPSI [9]andTBC.Because GSTEBisaselforganizedprotocol,itonlyconsumesasmallamountof energy ineach round tochangethetopographyforthepurposeofbalancingt heenergyconsumption.Alltheleafnodescantransmit datainthesameTDMAtimeslotsothatthetransmitting delay isshort.Whenlifetimeisdefinedasthetimefromthestar tofthe networkoperationto thedeath of thefirstnode in the network.
Two schemes: 1.
2.
After root initiate data gathering process by sending a small control packet to the children nodes using standard tree traversal algorithm.it will consume some negligible amount of energy but with little more delay. All the leaf nodes send sensed information to their parent. The parent nodes fuse the received data with their own data and forward the resultant data to their parent until data are received by the root node. It has less delay, but needs to introduce some multiplexing scheme to avoid collision.
This path is used for several rounds until node i dies. After the death of node i, a new tree path is constructed with node i+1 as the root. The simulation results shows that TREEPSI outperforms all the existing protocols in terms of energy efficiency. 2.1.5.
ISBN NO : 378 - 26 - 13840 - 9
GSTEB: [1]
Advantages of GSTEB:
Load balancing Self-organizing More efficient than other routing protocols.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 43
Proceedings of International Conference on Developments in Engineering Research
GSTEB prolongs the life time by balancing the load.
www.iaetsd.in
GSTEB
389
677
REFERENCES: [1] Zhao Han, Jie Wu, Liefeng Liu, Jie Zhang and kaiyunTian “A General self-organized tree based energy balance routing protocol for wireless sensor networks”., vol.61 , 0018-9499, 2014
III. Conclusions: In this survey paper, routing protocols are compared based on the network life time and data aggregation and the load balancing criteria. Then the new GSTEB protocol’s simulations show that GSTEB outperforms LEACH, PEGASIS, TREEPSI and TBC. Because GSTEB is a selforganized protocol, it only consumes a small energy and it balancing the energy consumption and it have short transmission. GSTEB prolongs the lifetime by 100% to 300% compared with PEGASIS [1]. GSTEBprolongsthelifetimeofthenetworkbymorethan100% compared with HEED. TABLE 1:
LEACH
ROUND TAKEN TO A NODE DIE 118
ROUND TAKEN TO ALL NODES DEAD 248
PEGASIS
246
568
TREEPSI
267
611
TBC
328
629
ISBN NO : 378 - 26 - 13840 - 9
[3] “An Integrative Comparison of Energy Efficient Routing Protocols in Wireless Sensor Network” ,Ali Norouzi1, Abdul Halim Zaim2 1Department of Computer Engineering, Istanbul University (Avcilar), Istanbul, Turkey 2Department of Computer Engineering, Istanbul Commerce University (Eminonu), Istanbul, Turkey Email: norouzi@cscrs.itu.edu.tr, azaim@iticu.edu.tr Received December 8, 2011; revised January 11, 2012; accepted January 30, 2012 [4] “HEED: A Hybrid, Energy-Efficient, Distributed Clustering Approach for Ad Hoc Sensor Networks”, OssamaYounis, Student Member, IEEE, and Sonia Fahmy, Member, IEEE, IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 3, NO. 4, OCTOBER-DECEMBER 2004 [5] “PEGASIS: Power efficient gathering in sensor information system” Stephanie Lindsey and Cauligi S. Raghavendra,”IEEE AC paper #242 , updated Sept. 29,2001. 0-7803-7231- 2002 IEEE.
NETWORK LIFE TIME
PROTOCOL
[2] Ankit Solanki,prof.niten b. patel “LEACH-SCH: An innovative routing protocol for wireless sensor network” 4th icccnt-2013
[6]O.durmazincel, a.ghosh,b.krishnamachari and k.chintalapudi “Fast data collection in tree based wireless sensor networks” ieeecs,comsoc. [7] “TREEPSI: TRee based Energy Efficient Protocol for Sensor Information”SiddharthaSankarSatapathy, NityanandaSarmaTezpur University, Napaam-784028, 2006 IEEE
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 44
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
AN EFFECTIVE ALARMING MODEL FOR DANGER AND ACTIVITY MONITORING USING WEARABLE SENSORS FOR CHILDREN P.Vidyadhar1, PG Student, Department of ECE, ASCET, Gudur, Andhra Pradesh, India. K. Dhanunjaya2, Head of the Department, Department of ECE, ASCET, Gudur, Andhra Pradesh, India. Abstract—This paper presents a child activity recognition approach using a single 3-axis accelerometer and a ultra-sonic sensor which is belted around the waist of the baby so as to prevent him from dangers and getting injured. The accelerometer data is collected and monitored in a computer using IEEE 802.15.4 protocol. In addition to the activity recognition child body temperature can also be monitored at regular time intervals. A fire sensor is also embedded in the proposal so as prevent the baby from fire accidents and a SMS using GSM will be sent to their parents if they are going to be involved in any fire accidents. Child activities are classified into 8 daily activities according to our consideration which are moving left, right, front, back, standing still, climbing up, climbing down, and stopping. The accuracy obtained for every activity is around 90% in any situation using single MEMS ADC sensor and ULTRA-SONIC sensor. Index Terms—Accelerometer, activity classification, activity recognition, baby care, child care.
concluded that positions depend on the activities being performed by the subject and these activity recognition can be made optimized using accelerometers.The dominant role in the designing the system is that it must be designed in a small space and low weight which can be bearable by the baby. As we are optimizing the sensor activity so we have to compromise in the accuracy. It is difficult to configure an optimal system. It depends on mainly on two important factors one is sensor activity and the positions of the baby. In our study, to decrease the incomfortness the waist belt is kept diaper for the children below three years of age, during physical activity and to measure body motions such as moving left, right forward backward, climbing up and climbing down.In our proposal mainly we have designed a wearable sensor device and a monitoring application to collect information about the activities and to recognize baby activities baby is doing. We classified baby activities into 8 daily activities which are moving left, right, front, back, standing still, climbing up, climbing down, and stopping. As multiple sensors are embedded in a wearable devicewhich are more accurate for collecting differenttypes of sensing informationbut would be very inconvenient for users.
I. INTRODUCTION At presentare getting into some accidents due to lack of personal monitoring in this busy world, usually child start walking between 9 and 16 months, there will bechance of falling fromhigher heights or stairs. As the child learns to climb, they will be at risk of falling from stairs, chairs and beds. Children frequent come across injury due to these TABLE I accidents unknowingly.Medical research show that these SENSOR AND TYPES OFTHEC OMPONANTS USED accidents are one of the most common cause of injuries that TYPE SENSOR VALUE FEATURE require medical care, and in some situations non fatal injuries Location also leads to hospitalization. The main areas these accidents Space RFID Room identification (kitchen, dining occur are at homes because of lack of parental care. Thus, a room, bed) new effective alarming model for danger and activity Object Object Name RFID ID monitoring using triaxial accelerometerfor children is required (electric socket) to prevent child from accidents at homes. These accidents have Activity Activity 3-axis [-2g, +2g] (moving left, great effect on growth and development of the child.Accident accelerometer right) prevention measures are to be taken effectively. The Height challenging thing is the classification activities in terms of Ultra-sonic Height from the [30kPa,129kPa] sensor ground safety and damage occurring to the child. There are many proposals for the recognition of activity but the challenging Ambient Temperature LM35 [20◦C,100◦C] task lies with the accurately recognition of activity the child is. temperature Detection of fire A smart sensor network is used in this proposal for rescuing the Fire Fire sensor near the baby baby from injuries and some small cracks on the body. Alarming the SMS GSM Multisensor link has been provided for elderly people and baby is in danger children at home. This approach will give the activity data in a simple and recognizable manner. According to theproposal the For this reason, we present only one single unit of sensor human activity is recognized by fusing two highly accurate nodes, which collects multiple types of information. The nature sensors one which is attached to one of the foot and another of information interaction involved in sensor fusion can be sensor to the waist of a baby subject, respectively. Due to the classified as competitive, complementary, and cooperative use of multiple sensors robustness of the classification systems fusion. In competitive fusion, each sensor provides equivalent has been improved drastically and increases thepartialness of information about the process being monitored. In high-level decision making. On the other hand, the ultra-sonic complementary fusion, sensors do not depend on each other sensor could fail to detect the activities like head motion, body directly, as each sensor captures different aspects of the tilt, and hand motion. In addition to that and for the purpose of physical process. The measured information is merged to form minimizing the number of sensors worn, it is important to know a more complete picture of the phenomenon. Cooperative the capability of a certain position to classify a set of activities. fusion of the two sensors enables recognition of the activity that Recently, Attallaet al. Investigated the effects of sensor could not be detected by each single sensor. Due to the position and feature selection on activity classification compounding effect, the accuracy and reliability of cooperative tasksusing accelerometers. Accelerometers are the mostbroadly fusion is sensitive to inaccuracies in all simple sensor used sensors to recognizeambulation activities such as walking components used. In this paper, we select the cooperative and running the advantage of the accelerometer is inexpensive, fusion model to combine information from sensors to capture require relatively low power, and are greatest applications in data with improved reliability, precision, fault tolerance, and most The study ofINTERNATIONAL the optimal sensor ISBN NO : 378of - 26mobile - 13840 -phones. 9 ASSOCIATION OF ENGINEERING & TECHNOLOGY DEVELOPMENT reasoning power to a degree that is beyondFOR the SKILL capacity of each 45
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
sensor. The main contributions of this paper over the earlier previous work are 1) To extend the method to work with arbitrary every day activities not just walking by improving the feature selection and recognition procedure. 2) To perform evaluation on a large (50 h) dataset recorded from real life activities. 3) To have studied ten divers subjects: 16, 17, 20, 25, 27 months-old baby boys and 21, 23, 24, 26, 29 months-old girls. 4) To employ a barometric pressure sensor for improving upon the previous algorithms. The proposed method classified daily physical activity of children by a diaper worn device consisting of a single triaxial accelerometer and a ultra-sonic sensor. We demonstrate our improvements in comparison to the accuracy results of only a single-wearable device and multiple feature sets to find an optimized classification method. II. METHODS A. Sensor Device In order to recognize daily activities, we adopt multiple sensors, as shown in Table I, as follows: 1) A 3-axis accelerometer measures the movement; 2) An absolute ultra-sonic sensormeasures absolute pressure enabling a measurement of a distance between the ground and the wearable sensor device; 3) A radio-frequency identification (RFID) (Skye Module M1mini) is selected to read/write tags and smart labels, which has compatibility with most industry standard 13.56 MHzâ&#x20AC;&#x2122;s. We used the MEMS ADC that is a 3-axis accelerometer for applications requiring high performance with low power consumption. It consists of three signal-processing channels where it is low-pass filtered and communicates with the processing layer based on SPI bus that is a full duplex synchronous 4-wire serial interface. We also used the as a ultrasonic sensor that measures absolute distance to measure distance between the ground and the sensor. The distance and temperature output data are calibrated and compensated internally. The sensor Communicates with the processing layer through an SPI bus. The Skye Module M1-mini has a read/write distance that is typically greater than or equal to two inches for an ISO15693 RFID inlay. The sensor allows us to recognize objects and space that may cause dangerous situations. Finally, we developed the prototype wearable sensor device (size of 65mm Ă&#x2014; 25 mm) including the dual-core processor and sensors as shown in Fig. 1.
Fig.1. Prototype of the wearable sensor device and the RFID reader. .
At the receiver section for monitoring purpose we will be having a computer attached with a zigbee module so as soon as the data is transmitted from the baby that will be displayed in the hyper terminal. For every change in the activity the corresponding activity name will be displayed on the screen.
Fig.2. Prototype of the receiver end for monitoring. .
Here in this proposal RFID tags are provided for each room and any dangerous objects like power sockets e.t.c, so when the active cards are detected by the RFID reader in the baby unit then an alarming message will be displayed on the computer hyper terminal so after seeing this a man can go and avoid the baby from that danger preventing injuries and some fedial injuries on the body. A message will also be sent to the chosen person by using the gsm module present with the baby. Here in this proposal we are giving a RFID tag to each and every room in the house so as to monitor the baby position in the house which makes it easy for the person to go for the rescue.
III. BLOCK DIAGRAM In the baby side we will be having LPC2148, LM35, Fire sensor, RFID reader, MEMS sensor and Ultra-sonic sensor a 3axis accelerometer measures the movement. An absolute ultrasonic sensormeasures absolute pressure enabling a measurement of a distance between the ground and the wearable sensor device. An absolute ultra-sonic sensormeasures absolute pressure enabling a measurement of a distance between the ground and the wearable sensor device. We also used the as a ultra-sonic sensor that measures absolute distance to measure distance between the ground and the sensor. The temperature sensor which measure the body temperature at regular intervals of time. The fire sensor which helps baby to be involved in any fire accidents. When the baby was in danger a sms will sent to their parents so they can come for his rescue. These all devices can be embedded on single chip and they can be attached on a belt so that it will not be uncomfortable to the baby. ISBN NO : 378 - 26 - 13840 - 9
Fig.3. Prototype of the waist belt
The above figure shows the prototype of the waist belt we are using in the proposal, this belt is properly placed on to the baby so that it will not be uncomfortable for the baby.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 46
Proceedings of International Conference on Developments in Engineering Research
D. Activity Recognition
www.iaetsd.in
transmitted to the receiver section and displayed in the hyper terminal for monitoring purpose with the help of zigbee modulus. Climbing down:If the baby is climbing down the position of the the plane of the mems sensor will be perpendicular to x-axis respectively. The mems sensor will sense all the positions accurately. Here in this activity the ultra-sonic sensor has a task a to perform. The ultra-sonic sensor will move up down for a fixed distance indicating that the baby is climbing down. Then the data is transmitted i.e. nothing but some milli volts voltage will be generated and it is compared according to the program. Then the data will be transmitted to the receiver section and displayed in the hyper terminal for monitoring purpose with the help of zigbee modulus. Sitting down:If the baby is Sitting down the position of the the plane of the mems sensor will be perpendicular to negative xaxis respectively. The mems sensor will sense all the positions accurately. Here in this activity the ultra-sonic sensor has task a to perform. Then the data is transmitted i.e. nothing but some milli volts voltage will be generated and it is compared according to the program. Then the data will be transmitted to the receiver section and displayed in the hyper terminal for monitoring purpose with the help of zigbee modulus.
After the preparation of the all the mechanical possibilities are ready and all the circuitry is complete then we will proceed to the working of accelerometer and ultra-sonic sensor to recognize the activities. A classification procedure separates the child’s activities from all other primitive features. The activity recognition algorithm should be able to recognize the accelerometer signal pattern corresponding to every activity. Standing still: If the baby is standing still the position of the the plane of the mems sensor will be perpendicular to x-axis respectively. The mems sensor will sense all the positions accurately. Here in this activity the ultra-sonic sensor has no task to perform. Then the data is transmitted i.e. nothing but some milli volts voltage will be generated and it is compared according to the program. Then the data will be transmitted to the receiver section and displayed in the hyper terminal for monitoring purpose with the help of zigbee modulus. Moving left:If the baby is moving let the position of the the plane of the mems sensor will be perpendicular to negative yaxis respectively. The mems sensor will sense all the positions accurately. Here in this activity the ultra-sonic sensor has no task to perform. Then the data is transmitted i.e. nothing but some milli volts voltage will be generated and it is compared TABLE II according to the program. Then the data will be transmitted to SENSOR POSSITION ACCOURDING TO ACTIVITIES the receiver section and displayed in the hyper terminal for monitoring purpose with the help of zigbee modulus. ULTRAMoving right:If the baby is moving right the position of the the ACCELEROMETER ACTIVITY SONIC plane of the mems sensor will be perpendicular to y-axis POSITION POSITION respectively. The mems sensor will sense all the positions Standing still Perpendicular to x-axis accurately. Here in this activity the ultra-sonic sensor has no Perpendicular to task to perform. Then the data is transmitted i.e. nothing but Moving left negative y-axis some milli volts voltage will be generated and it is compared Moving right Perpendicular to y-axis according to the program. Then the data will be transmitted to Moving Perpendicular to the receiver section and displayed in the hyper terminal for forward negative z-axis monitoring purpose with the help of zigbee modulus. Moving Moving forward:If the baby is moving forward the position of Perpendicular to z-axis backward the the plane of the mems sensor will be perpendicular to Will move negative z-axis respectively. The mems sensor will sense all the Climbing up Perpendicular to x-axis upwards positions accurately. Here in this activity the ultra-sonic sensor Climbing Will move has no task to perform. Then the data is transmitted i.e. nothing Perpendicular to x-axis down downwards but some milli volts voltage will be generated and it is compared according to the program. Then the data will be transmitted to the receiver section and displayed in the hyper The above sensors data will stored at regular intervals of time terminal for monitoring purpose with the help of zigbee so as to analyze the baby behavior. So as we can take modulus. preventive measures an avoiding those things which greatly Moving backward:If the baby is moving backward the position effecting the baby. Here temperature sensor plays an important of the the plane of the mems sensor will be perpendicular to z- role in sensing the temperature of the baby, the abnormal axis respectively. The mems sensor will sense all the positions condition of the temperature is programmed between 28°F to accurately. Here in this activity the ultra-sonic sensor has no 100 °F respectively. So these are the abnormal which are taken task to perform. Then the data is transmitted i.e. nothing but into consideration. Here a fire sensor is provided so as rescue some milli volts voltage will be generated and it is compared the baby if he is involved in any fire accident or if he is near to according to the program. Then the data will be transmitted to fire. Each and every device which is harmful to the baby in the the receiver section and displayed in the hyper terminal for house is given with an rfid card so as to prevent the baby from monitoring purpose with the help of zigbee modulus. injuries. If in any situation the baby is near to these devices the Climbing up:If the baby is climbing up the position of the the alarm will be blown and baby will rescued from all the plane of the mems sensor will be perpendicular to x-axis dangerous articles in the home. For smaller devices a RFID pil respectively. The mems sensor will sense all the positions can be provided as to facilitate comfortable working. So from accurately. Here in this activity the ultra-sonic sensor has task the proposal we are able to provide a secure home for the to perform. The ultra-sonic sensor will move up for a fixed babies below 2 years of age. This system is more accurate distance indicating that the baby is climbing up.Then the data is compared with all the present existing systems in the market. transmitted i.e. nothing but some milli volts voltage will be This proposal has a greatest advent in present day world which generated and it is compared according to the program. Then makes a great sense. the data will be transmitted to the receiver section and III. RESULTS displayed in the hyper terminal for monitoring purpose with the A. Experimental Setup help of zigbee modulus.Then the data is transmitted i.e. nothing We have volunteered two families to conduct the experiment in but some milli volts voltage will be generated and it is 40 m²bed room and 20 m² gardens. In a controlled environment ISBN NO : 378 - 26 according - 13840 - 9 to the program. Then INTERNATIONAL & TECHNOLOGY FOR SKILL DEVELOPMENT compared the data willASSOCIATION be 47 setting,OF allENGINEERING data were collected from ten babies whoare 16, 17,
Proceedings of International Conference on Developments in Engineering Research
20, 25, 27 months-old baby boys and 21, 23, 24, 26,and 29 months-old baby girls. A supporter who observed the experiments annotates raw data by clicking buttons or typing name of the activities through the monitoring application. All experiments were performed in a real home environment consisting of one wearable sensor device for the child and the monitoring application operated on a laptop computer. Baby boys wiggled more than baby girls. They squirmed more and get restless on the floor, and crawl over longer distances. While the average boy did not move around much more than the typical girl, the most active children were almost always boys, and the least active children were girls. The following are the results of different activities of a child in different conditions. i) This is the result when the baby is moving forward.
Fig.4. baby in forward motion
ii) This is the result when the baby is moving backward.
Fig.5. baby in backward motion
iii) This is the result when the baby is climbing down.
www.iaetsd.in
Fig.6. baby is climbing up.
v) This is the result when the baby is moving right.
Fig.7. baby is moving right
vi) This is the result when the baby is moving left.
Fig.8. baby is moving left
vii) This is the result when the baby is baby is danger, sms will be sent to the parents.
Fig.5. baby isclimbing down Fig.9. sms alert to the parents
iv) This is the result when the baby is climbing up. ISBN NO : 378 - 26 - 13840 - 9
vii) This is the result at the receiver section
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 48
Proceedings of International Conference on Developments in Engineering Research
Fig.9. Result at the receiver section
IV. CONCLUSION This paper has presented the activity recognition method for Children using only a triaxial accelerometer and anultra sonic sensor. Results showed that using anultra-sonic sensor could reduce the incidence of false alarms. The early warning system will give the parents enough time to save their babies, and, thus, minimize any instances of falling accidents or sudden infant death syndrome. ACKNOWLEDGMENT The authors would like to thank all the volunteers MATHAJI REHABITATION CHILD CENTRE who have supported greatly in our experiments. They would also like to thankK. Dhanunjayafor previewing our manuscript.
www.iaetsd.in
[4] S. Boughorbel, J. Breebaart, F. Bruekers, I. Flinsenberg, and W.ten Kate. (2010). Child-activity recognition from multisensor data.in Proc. 7th Int. Conf. Methods Tech. Behav. Res., ser. MB ’10.New York, NY, USA: ACM, [Online].pp.38:1– 38:3, Available:http://doi.acm.org/10.1145/1931344.1931382 [5] A. Fleury, M. Vacher, and N. Noury, “Svm-based multimodal classification of activities of daily living in health smart homes: Sensors, algorithms,and first experimental results,” IEEE Trans. Inf. Technol. Biomed., vol. 14,no. 2, pp. 274–283, Mar. 2010. [6] C. Zhu andW. Sheng. (2009).Multi-sensor fusion for human daily activityrecognition in robot-assisted living. in Proc. 4th ACM/IEEE Int. Conf.Human Robot Int., ser. HRI ’09. New York, NY, USA: ACM, [Online].pp. 303–304, Available: http://doi.acm.org/10.1145/1514095.1514187 [7] L. Atallah, B. Lo, R. King, and G.-Z. Yang, “Sensor positioning for activity recognition using wearable accelerometers,” IEEE Trans. Biomed.Circuits Syst., vol. 5, no. 4, pp. 320–329, Aug. 2011. [8] L. Bao and S. S. Intille. (2004). Activity recognition from user-annotatedacceleration data. Pervas. Comput., [Online]. pp. 1–17, Available: http://www.springerlink.com/content/9aqflyk4f47khyjd [9] R. Luo and M. Kay, “Multisensor integration and fusion in intelligentsystems,” IEEE Trans. Syst., Man, Cybern., vol. 19, no. 5, pp. 901–931,Sep./Oct. 1989. [10] R. R. Brooks and S. S. Iyengar, Multi-sensor Fusion: Fundamentals andApplications with Software. Englewood Cliffs, NJ: Prentice-Hall, 1998. [11] R. Luo, C.-C. Yih, and K. L. Su, “Multisensor fusion and integration:Approaches, applications, and future research directions,” IEEE SensorsJ., vol. 2, no. 2, pp. 107–119, Apr. 2002. [12] N. Ravi, N. Dandekar, P. Mysore, and M. L. Littman. (2005). Activityrecognition from accelerometer data. in Proc. 17th Conf. Innovated. Appl.Artif. Intel., AAAI Press, [Online]. vol. 3, pp. 1541–1546, Available:http://dl.acm.org/citation.cfm?id=1620092.1620107 [13] B. Baas and S. U. D. of Electrical Engineering. (1999). An Approachto Low-Power, High-Performance, Fast Fourier Transform ProcessorDesign. Stanford, CA: Stanford University. [Online]. Available:http://books.google.com/books?id=61KFHwAACAA J [14] R. Kohavi. (1995). A study of cross-validation and bootstrap foraccuracy estimation and model selection. in Proc. 14th Int. JointConf. Artif. Intel. Francisco, CA, USA: Morgan Kaufmann PublishersInc., [Online]. vol. 2, pp. 1137–1143, Available: http://dl.acm.org/citation.cfm?id=1643031.1643047 [15] J. R. Quinlan, C4.5: Programs for Machine Learning (Morgan Kaufmann Series in Machine Learning), 1st ed. San Mateo, CA: Morgan Kaufmann,Oct. 1992. [16] J. C. Platt. (1999). Fast training of support vector machines usingsequential minimal optimization, Advances in kernel methods.,B. Sch¨olkopf, C. J. C. Burges, and A. J. Smola, Eds., Cambridge, MA: MIT Press. [Online]. pp. 185–208, Available: http://dl.acm.org/citation.cfm?id=299094.299105 [17] S. S. Keerthi, S. K. Shevade, C. Bhattacharyya, and K. R. K. Murthy.(2001, Mar.). Improvements to platt’s SMO algorithm for SVM classifierdesign. Neur. Comput., [Online]. 13(3), pp. 637–649, Available:http://dx.doi.org/10.1162/089976601300014493 .
REFERENCES [1] A. M. Khan, Y.-K. Lee, S. Y. Lee, and T.-S. Kim. (2010, Sep.). A triaxial Accelerometer-based physical-activity recognition via augmentedsignal features and a hierarchical recognizer. Trans. Info. Tech. Biomed., [Online]. 14(5), pp. 1166–1172, Available: http://dx.doi.org/10.1109/ TITB.2010.2051955 426 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 17, NO. 2, MARCH 2013 [2] N. Li, Y. Hou, and Z. Huang. (2011). A real-time algorithm based on triaxial accelerometer for the detection of human activity state in Proc. 6th Int. Conf. Body Area Netw., ser. BodyNets ’11. ICST, Brussels, Belgium, Belgium: ICST (Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering), [Online]. pp. 103–106, Available: http://dl.acm.org/citation.cfm?id=2318776.2318801 [3] A. G. Bonomi. (2011). Physical activity recognition using a wearableaccelerometer. in Proc. Sens. Emot., ser. Philips Research Book Series, J. Westerink, M. Krans, and M. Ouwerkerk, Eds., Springer Netherlands,[Online]. vol. 12, pp. . 41–51. Available: http://dx.doi.org/10.1007/978-90-481-3258ISBN NO : 378 - 26 - 13840 - 9 INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 4_3 49
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
An Efficient Way of Classifying and Clustering Documents Based on SMTP U.Umamaheswari [1],
Mr.G.Shivaji Rao M.E.[2],
M.E scholar, [1] Department of Computer Science and Engineering, Sree Sowdambika College of Engineering, Aruppukottai-626 134, Tamil Nadu, India. 13umamaheswari@gmail.com
Assistant Professor, Department of Computer Science and Engineering, Sree Sowdambika College of Engineering, Aruppukottai-626 134, Tamil Nadu, India.
[2]
shivajirao88@gmail.com
Abstract In text processing, the similarity measurement is the important process. It measures the similarities between the two documents. In this project we proposed the new similarity measurement. The computation of similarity measurement is based on the feature of two documents. Our proposed system contains three case to compute the similarity. The three cases are, both two documents contains features, only one document contains feature, there is no feature into the two documents. In first case, the similarity is increased when the differences of feature value is decreased between the two documents. Then the given differences are scaled. In second case, fixed value is given to the similarity. In third cases there is no contribution to the similarity. Finally our proposed measure method achieves the better performance compared than other measurement methods. Index Termsâ&#x20AC;&#x201D;Document classification, document clustering, entropy, accuracy, classifiers, clustering algorithms
I. INTRODUCTION Text processing plays an important role in information retrieval, data mining, and web search. A document is any content drawn up or received by the Foundation concerning a matter relating to the policies, activities and decisions falling within its competence and in the framework of its official tasks, in whatever medium (written on paper or stored in electronic form , including e-mail, or as a sound, visual or audio-visual recording). The term classification means the allocation of an appropriate level of security (as confidential or restricted) to a document the unauthorised disclosure of which might prejudice the interests of the Foundation, the EU or third parties. Documents are confidential when their unauthorised disclosure could harm the essential interests of an individual, the Foundation or the EU. Documents are restricted when their unauthorised disclosure could be Disadvantageous to the Foundation, the EU or a third party. Documents are restricted when their unauthorised disclosure could be disadvantageous to the Foundation, the EU or a third party. The term originator means the duly authorised author of a classified document. The term downgrading means a reduction in the level of classification. The term declassification means the removal of any classification.
ISBN NO : 378 - 26 - 13840 - 9
RULES FOR CLASSIFICATION Foundation documents that are not public shall be classified in one of the following categories: confidential or restricted. Criteria and guidance for classification are set out in Annex 1 to this decision. The classification of a document shall be decided by the originator based on these rules. All classified documents shall be recorded in a register of classified documents. Applications for access to classified documents shall be examined by the Director. If a classified document is to be made available in response to a request from a member of the public, it shall be first declassified by a decision of the Director. Documents shall be classified only when necessary. The classification shall be clearly indicated and shall be maintained only as long as the document requires protection. The classification of a document shall be determined by the level of sensitivity of its contents. Classification of documents shall be periodically reviewed. By request of the Document Management Officer (DMO), the originator of a document shall indicate if that document or information may be downgraded and declassified. Where a document or information is declassified, details shall be recorded in the register and the document shall be archived appropriately. Where classification is retained, details of the review shall be entered
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 50
Proceedings of International Conference on Developments in Engineering Research
in the register. All classified documents shall be retained in a manner that ensures they are not disclosed to unauthorised individuals. Security and access levels to classified documents shall be recorded in the register by the DMO. Either the originator or the DMO may retain classified documents. Where the originator retains documents, they shall be physically safeguarded as specified by the DMO. Individual pages, paragraphs, sections, annexes, appendices, attachments and enclosures of a given document may require different classifications and shall be classified accordingly. The classification of the document as a whole shall be that of its most highly classified part. The originator shall indicate clearly at which level a document should be classified when detached from its enclosures. The classification shall appear at the top and bottom centre of each page, and each page shall be numbered. Each classified document shall bear a reference number and a date. All annexes and enclosures shall be listed on the first page of a document classified as confidential. The classification shall be shown on restricted documents by mechanical or electronic means. Classification shall be shown on confidential documents by mechanical means or by hand or by printing on pre-stamped, registered paper. DOCUMENT CLASSIFICATION Document classification or document categorization is a problem in library science, information science and computer science. The task is to assign a document to one or more classes or categories. This may be done "manually" (or "intellectually") or algorithmically. The intellectual classification of documents has mostly been the province of library science, while the algorithmic classification of documents is used mainly in information science and computer science. The problems are overlapping, however, and there is therefore also interdisciplinary research on document classification. The documents to be classified may be texts, images, music, etc. Each kind of document possesses its special classification problems. When not otherwise specified, text classification is implied. Documents may be classified according to their subjects or according to other attributes (such as document type, author, printing year etc.). In the rest of this article only subject classification is considered. There are two main philosophies of subject classification
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
of documents: The content based approach and the request based approach. II. RELATED WORK The documents are represented as vectors, each elements of vectors indicates the corresponding feature values of the documents. In existing system the feature values are termed as frequency, relative term frequency and tf-idf (term frequency and inverse document frequency). The document size is large and most of the vector value is zero in this system. In existing system non symmetric measure is used to measure the similarities. Canberra distance metric is one of the existing methods which are used to measure the similarity. This method is applicable when the vector elements are always non-zero. In existing system cosine similarity measure is used to measure the similarities between the two documents. It takes cosine of angle between the two vectors. The phase based similarity measure is used in one of the existing system. So, we are our some problem in following as the document is large in existing system so it takes more memory to store the documents. It contains zero vector values so it makes severe challenges to measure the similarity. It is not efficient and does not provide accurate results. The performance is low. It takes more time to produce the result. In our proposed system new measure is used to measure the similarities between the two documents. This new measure is symmetric measure. The similarities between the two documents are measured with respect to the features. Three feature cases are used to measure the similarities. In first case the two documents contains features value, in second case, only one document contains the feature value and in third case there is no feature value in both documents. This measure is applied in many applications such as single label and multi label classification, clustering and so on. In our proposed system we k-NN based single label classification (SL-KNN) and k-NN base multi label classification (ML-KNN) for classification purpose. The documents are formed as cluster in our concept for this purpose we used Hierarchical Agglomerative Clustering (HAC). It is the one type of k means clustering algorithm. We three type of data sets such as WebKB, Reuters 8 and RCV1. These data sets are stored in the form of XML. So, In our project improve following as It achieves better
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 51
Proceedings of International Conference on Developments in Engineering Research
performance compared than other measure, It provides efficient results, It achieves accuracy in similarity measurement, It take less time for similarity measure. III.PROPOSE SIMILARITY MEASURE In our proposed system new measure is used to measure the similarities between the two documents. This new measure is symmetric measure. The similarities between the two documents are measured with respect to the features. Three feature cases are used to measure the similarities. In first case the two documents contains features value, in second case, only one document contains the feature value and in third case there is no feature value in both documents. This measure is applied in many applications such as single label and multi label classification, clustering and so on. In our proposed system we k-NN based single label classification (SL-KNN) and k-NN base multi label classification (ML-KNN) for classification purpose. The documents are formed as cluster in our concept for this purpose we used Hierarchical Agglomerative Clustering (HAC). It is the one type of k means clustering algorithm. We three type of data sets such as WebKB, Reuters 8 and RCV1. These data sets are stored in the form of XML.
Fig: System architecture The following properties, among other ones, are preferable for a similarity measure between two documents: 1. The presence or absence of a feature is more essential than the difference between the two values associated with a present feature. Let as Consider two features wi and wj and two documents d1 and d2. wi = d2(some relationship) wi ≠ d1(no relationship) In this case, d1 and d2 are dissimilar in terms of wi. If wj appears in both d1 and d2. Then wj has some relationship with d1 and d2 simultaneously. In this case, d1 and d2 are similar to some degree in terms of wj.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
For the above two cases, it is reasonable to say that wi carries more weight than wj in determining the similarity degree between d1 and d2. 2. The similarity degree should increase when the difference between two nonzero values of a specific feature decreases. 3. The similarity degree should decrease when the number of presence-absence features increases. 4. Two documents are least similar to each other if none of the features have nonzero values in both documents. Let d1 = < d11, d12, . . . , d1m > and d2 = < d21, d22, . . . , d2m >. If d1id2i = 0, d1i + d2i > 0 for 1 ≤ i ≤ m, then d1 and d2 are least similar to each other. As mentioned earlier, d1 and d2 are dissimilar in terms of a presence-absence feature. 5. The similarity measure should be symmetric. That is, the similarity degree between d1 and d2 should be the same as that between d2 and d1. 6. The value distribution of a feature is considered, i.e., the standard deviation of the feature is taken into account, for its contribution to the similarity between two documents. A feature with a larger spread offers more contribution to the similarity between d1 and d2. IV. METHODOLOGY 1. DATA SET SELECTION Here we used three types of data sets. They are WebKB, Reuters 8 and RCV1. These data sets are stored in the form of HTML and text or SGM. Each data set is divided into two types. They are training and testing data set. WebKB contains web pages as the document which is collected by the world wide knowledge base. It does not contain predefined training set and testing set. So we randomly divide these data sets as training and testing set. Reuter’s data set contains predesigned training set and testing sets. If the resultant data set contains 71% of training set and 29% of testing set then it is called as Reuters 8. RCV1 also contains predesigned training and testing data sets. 2. CLUSTERING After the data set collection then we perform the cluster formation. Cluster contains the more than one data set. For cluster formation
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 52
Proceedings of International Conference on Developments in Engineering Research
we used k-means algorithm. In our proposed system, we used HAC (hierarchical agglomerative clustering). It is the one type k means clustering algorithm and it used bottom up approach for clustering. The HAC is mostly used to measure the performance. 3. PERFORM CLASSIFICATION After form the cluster then we performs the classifications. We used SL-kNN and MLkNN method is used to perform the classifications. SL-KNN is used for only one category and ML-KNN is used for more than two categories. These are used to measure the similarity between the documents. It gives best result for similarity. Classification is used to retrieve process. It increases the fast of retrieve the data. It easily identifies the similarity between the documents. It gives related result efficiently. 4. FIND SIMILARITIES After select the classification methods then we finds the similarities between the data sets. The similarity measurement is based on the features of documents. It finds the word weight and distance to measure the similarities. Similarity measure has three properties they are, the presence and absence feature is more important than the difference of two values that are presented in the feature, the similarity degree is increased when the difference between the non zero values of feature is decreased, the similarity degree is decreased when the presence and absence of feature is increased. 5. PERFORMANCE COMPARISON After perform all the process then we compared the performance of process. In our proposed system, the performance is high compared than all existing system. V.CONCLUSION The main concept of our proposed system is to find the similarity between the documents. For similarity measure we used KNN based single label classification and KNN based multi label classification. Here we collect three data sets such as WEBKB, reuters-8 and RCV1. Classification is used for retrieve process and it increases the fast of retrieve the data. It easily identifies the similarity between the documents. It gives related result efficiently. Here we used HAC for clustering and it is the one type of k means clustering algorithm. It also measures the performance. In our proposed system, the data sets are stored in the form the XML. Here we
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
used tf-idf (term frequency and inverse document frequency) to retrieve the document terms and It measures the similarity by calculating the weight. Here we used optimization and it identifies the no of cluster that we formed in the document. It increases the accuracy of similarity measurement.
VI.REFERENCES 1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Yung-Shen Lin, Jung-Yi Jiang, and Shie-Jue Lee, Member, IEEE,” A Similarity Measure for Text Classification and Clustering”, in IEEE Transactions On Knowledge And Data Engineering, Vol. 26, No. 7, July 2014 1575 D. Cai, X. He, and J. Han, “Document clustering using locality preserving indexing,” IEEE Trans. Knowl. Data Eng., vol. 17, no. 12, pp. 1624–1637, Dec. 2005. H. Chim and X. Deng, “Efficient phrase-based document similarity for clustering,” IEEE Trans. Knowl. Data Eng., vol. 20, no. 9, pp. 1217–1229, Sept. 2008. S. Clinchant and E. Gaussier, “Information-based models for ad hoc IR,” in Proc. 33rd SIGIR, Geneva, Switzerland, 2010, pp. 234–241. K. M. Hammouda and M. S. Kamel, “Hierarchically distributed peer-to-peer document clustering and cluster summarization,” IEEE Trans. Knowl. Data Eng., vol. 21, no. 5, pp. 681–698, May 2009. Y. Zhao and G. Karypis, “Comparison of agglomerative and partitional document clustering algorithms,” in Proc. Workshop Clustering High Dimensional Data Its Appl. 2nd SIAM ICDM, 2002, pp. 83–93. T. Zhang, Y. Y. Tang, B. Fang, and Y. Xiang, “Document clustering in correlation similarity measure space,” IEEE Trans. Knowl. Data Eng., vol. 24, no. 6, pp. 1002–1013, Jun. 2012. M. L. Zhang and Z. H. Zhou, “ML-kNN: A lazy learning approach to multi-label learning,” Pattern Recognit., vol. 40, no. 7, pp. 2038–2048, 2007. Strehl and J. Ghosh, “Value-based customer grouping from large retail data-sets,” in Proc. SPIE, vol. 4057. Orlando, FL, USA, Apr. 2000, pp. 33–42. F. Sebastiani, “Machine learning in automated text categorization,” ACM CSUR, vol. 34, no. 1, pp. 1–47, 2002. T. W. Schoenharl and G. Madey, “Evaluation of measurement techniques for the validation of agentbased simulations against streaming data,” in Proc. ICCS, Kraków, Poland, 2008. D. D. Lewis, Y. Yang, T. Rose, and F. Li, “RCV1: A new benchmark collection for text categorization research,” J. Mach. Learn. Res., vol. 5, pp. 361–397, Apr. 2004. V. Lertnattee and T. Theeramunkong, “Multidimensional text classification for drug information,” IEEE Trans. Inform. Technol. Biomed., vol. 8, no. 3 pp. 306–312, Sept. 2004. S.-J. Lee and C.-S. Ouyang, “A neuro-fuzzy system modeling with self-constructing rule generation and hybrid SVD-based learning,” IEEE Trans. Fuzzy Syst., vol. 11, no. 3, pp. 341–353, Jun. 2003.
15. G. Amati and C. J. V. Rijsbergen, “Probabilistic models of information retrieval based on measuring the divergence from randomness,” ACM Trans. Inform. Syst., vol. 20, no. 4, pp. 357– 389, 2002.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 53
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
An Efficient Way of Detecting a Numbers in Car License Plate Using Genetic Algorithms C.Subha
S.Sudha M.E.,
M.E Computer Science Sree Sowdambika College Of Engineering Aruppukottai Tamilnadu St, India mirthula31@gmail.com
Assistant Professor Department of Computer Science and Engineering Sree sowdambika College of Engineering Aruppukottai, Tamilnadu St, India sudhajiin@yahoo.co.in
Abstract - To detect the numbers and characters inside the license plate using image processing and genetic algorithm (GA). For this Number plate detection many algorithms are used. But in my project mainly focusing on the genetic algorithm for provide perfect accuracy compare to any other systems. This paper describes a detection method in which the vehicle plate image is captured by the cameras and the image is processed to get the plate’s numbers and characters. The system is implemented using MATLAB and various images are processed with to verify the distinction of the proposed system. Index Terms - Genetic algorithm (GA), Image processing, License plate (LP), Number plate localization, Perfect accuracy.
I.INTRODUCTION Nowadays number of automobiles grows quickly, the traffic problems arise as well, for example car robbery, over speeding and moving on the red light. To avoid these problems an efficient real time working vehicle identification system is needed. Most usually suitable technique is license plate (LP) detection based on image processing by capturing license plates using cameras. All the implemented techniques can be classified according to the selected features. Color information based systems have been built to detect specific plates having fixed colors. Shape- based techniques were developed to detect the plate based on its rectangular shape. Edge-based techniques were also implemented to detect the plate based on the high density of vertical edges inside it. GAs has been used infrequently because of their large computational needs. Variety of research has been tried at different levels under some constraints to minimize the search space of genetic algorithms (GAs). Researchers in based their GA on pixel color features to segment the image depending on stable colors
ISBN NO : 378 - 26 - 13840 - 9
followed by shape dependent policy to identify the plate’s area. In, GA was used to search for the best fixed rectangular area having the same texture features. GA was used in to identify the LP symbols not to detect the LP. Detecting license character and at the same time differentiating it from similar patterns based on the geometrical relationship between the symbols constituting the license numbers are selected approach in this research. Consequently, a new approach genetic algorithm is initiate in this paper that detects LP symbols without using any information linked with the plate’s external shape or interior colors to allow for the detection of the license numbers in case of shape or color distortion either physically or due to capturing conditions. Further processes are explained in the next sections. II.PROPOSED TECHNIQUE The proposed system is comprised of two phases: image processing phase and GA phase. Each phase is composed of many steps. The Fig. 1 depicts the various image processing steps that finally produce image objects to the GA portion. GA selects the best LP symbol locations depending on the input geometric relationship matrix (GRM). III.IMAGE PROCESSING PHASE In this phase, an input color image is used to a sequence of processes to extract the relevant 2-D objects that may represent the symbols. It has different stages, as depicted in Fig. 1. A. Color image to Grayscale conversion he input image is used as a color image to bring other information relevant to the concerned vehicle. Color (RGB) to grayscale (gs) conversion is
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 54
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig. 3. (a) Converted binary image for image in Fig. 2, using Otsuâ&#x20AC;&#x2122;s method. (b) Car image with variable illumination. (c) Output when using Otsuâ&#x20AC;&#x2122;s method for image in (b). (d) Output when applying local adaptive threshold method for same image in (b). Fig. 1. Overall system Flowchart for localization of LP symbols.
performed using the standard NTSC method by removing the hue and saturation information while holding the luminance as follows: gs=0.299*R+0.587*G+0.114*B
(1)
In LP detection, closing and opening operations are applied to fill noisy holes and remove objects. Dilation: This is the b asic operators in the part of morphology. It is usually applied to binary image, but there are versions run on grayscale image. the basic effect of the operator on a binary image is to progressively extend the boundaries of regions of foreground pixels (ie, white pixels). Applications of dilation for bridging gaps in an image. It can remove unwanted information. Opening of an image is erosion followed by a dilation using the same structuring element. Shown in fig 4. Erosion: This is also very important operator for the morphological operation. The basic effect of the operator on a binary image is to erode away the boundaries of regions of foreground pixels (ie, white pixels). Shown in fig 5.
Fig. 2. Converted grayscale image.
B. Grayscale to Binary Using Dynamic Adaptive Threshold Converting the input image into a binary image is one of the most important stages in localizing LPs to overcome the illumination problems. In my system, a local adaptive threshold technique has been implemented to determine the threshold at each pixel depending on the average gray level. This process as shown in Fig.3 C. Morphological Operations Morphological operations, like dilation and erosion, are important processes needed for pattern recognition systems to eliminate noisy object. Fig 4. Effect of dilation using 3x3 square structuring element
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 55
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
A. Chromosome Encoding In chromosome encoding an integer encoding scheme is selected and each gene assigned to an integer. Seven genes are forming a chromosome as shown in fig. 7. An output is extracted as a M objects.
Fig 5. Effect of erosion using a 3×3 square structuring element Strip away a layer of pixels from an object, shrinking it in the process.
Fig 7. Chromosome of seven genes for representation of Saudi license plate.
D. Connected Component Analysis
B. Fitness Function
CCA is one of the technique in image processing that scans an image and groups pixels in components depends on pixel connectivity. The result of this stage is an array of N objects.
Simple function of the fitness measure is used by some genetic algorithms to select individuals. In this proposed system fitness is used as the inverse of the estimated objective distance between the prototype chromosome and the current chromosome.
E. Size Filtering
C. Selection Method
The output of the CCA stage are filtered on the basis of their widths Wobj and heights Hobj lie between their respective thresholds as follows: Wmin ≤ Wobj ≤ Wmax
and
Hmin ≤ Hobj ≤ Hmax
(2)
Hmin and Wmin are the value below which a symbol cannot be recognized (for example 8 pixels) and Wmax can be set to the image width divided by the number of symbols. Hmax is estimated as Wmax divided by the aspect ratio of the used font. The result of this stage is an array of M objects. The output of this stage is given in Fig. 6.
In this selection method, the stochastic universal sampling (SUS) method each individual is formed to a continuous segment of a line. Depending on the percentage of individuals to be selected by a number of pointers over the line. D. Mutation Operators This mutation method is used to remove unfit members in genetic iterations. It can eliminate some features of genetic material. To maintain the mating pool variety by Gas ensures that the new parts of the search space. They are two kinds of mutation operators. 1) Substitution Operator 2) Swap Operator E. Crossover Operator In genetic algorithm crossover operator is used to produce new chromosome (offspring) by groups two chromosomes (parents). This new chromosome is better than the both parents if it takes the best characteristics from each of the parents. In my project, the two parents chromosomes are combined into the array Carray as shown in fig.8. In my project USPS crossover operator is used.
Fig. 6. M objects (64) output after size filtering of N objects in Fig. 5(2).
IV.GENETIC ALGORITHM In this phase M objects are given to the input. This phase is used to resolve the 2D compound object detection problem. It contains many steps.
ISBN NO : 378 - 26 - 13840 - 9
F. Replacement Strategy A lot of alternate strategies are used to replacing only a portion of the population between generations. The most frequent strategy is to probabilistically replace the unfit individuals in the earlier generation. In elitist strategy the
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 56
Proceedings of International Conference on Developments in Engineering Research
greatest fit individuals of the previous generation are appended to the recent population. In my proposed system, the best 10% of the parents are selected and appended to the offspring(90%) to produce the new generation (100%).
www.iaetsd.in
Moreover, an enhancement in the performance of the developed GA was achieved by applying the new USPS crossover operators, which greatly improved the convergence rate of the whole system. REFERENCES
Fig. 8 . Proposed crossover operator steps.
V.CONCLUSION In this paper describes the localization of license plate in a efficient manner. For this purpose i used genetic algorithm (GA). The license plate contain many unwanted details. These are first remove by the image processing phase and then localized by the genetic algorithm phase. The results were encouraging and a new approach for solving the LP detection problem relying only on the geometrical layout of the LP symbols. Also, a flexible system was introduced that can be simply adapted for any LP layout by constructing its GRM matrix. The proposed system possessed high immunity to changes in illumination either temporarily or spatially. A high percentage success rate was achieved with the aid of the adaptability aspect of the GAs. A very important attainment is overcoming most of the problems arising in techniques based on CCAT by allowing the GA.
ISBN NO : 378 - 26 - 13840 - 9
[1] A. Ahmadyfard and V. Abolghasemi, “Detecting license plate using texture and color information,” in Proc. Int. Symp. Telecommun., 2008, pp. 804–808. [2] G. Li, R. Yuan, Z. Yang, and X. Huang, “A yellow license plate location method based on RGB model of color image and texture of plate,” in Proc. 2nd Workshop Digit. Media Its Appl. Museum Heritages, 2007, pp. 42–46. [3] X. Shi, W. Zhao, Y. Shen, and O. Gervasi, “Automatic license plate recognition system based on color image processing,” in Lecture Notes on Computer Science, Berlin, Germany: SpringerVerlag, 2005, vol. 3483, pp. 1159–1168. [4] M. Deriche, “GCC license plates detection and recognition using mor- phological filtering and neural networks,” Int J. Comp. Sci. Info Security, vol. 8, no. 8, pp. 263–269, Dec. 2010. [5] O. Villegas, D. Balderrama, H. Dom´inguez, and V. Sa´nchez, “License plate recognition using a novel fuzzy multilayer neural network,” Int. J. Comput., vol. 3, no. 1, pp. 31–40, 2009. [6] S. H. M. Kasaei, S. M. M. Kasaei, and S. A. Monadjemi, “A novel morphological method for detection and recognition of vehicle license plate,” Amer. J. Appl. Sci., vol. 6, no. 12, pp. 2066–2070, 2009. [7] A. Theja, S. Jain, A. Aggarwal, and V. Kandanvli, “License plate extraction using adaptive threshold and line grouping,” in Proc. ICSPS, Jul. 2010, vol. 1, pp. 211–214. [8] P. Tarabek, “Fast license plate detection based on edge density and integral edge image,” in Proc. Int. Conf. Appl. Mach. Intell. Inform.,2012, pp. 37–40. [9] V. Abolghasemi and A. Ahmadyfard, “A fast algorithm for license plate detection,” in Proc. Int. Conf. Visual Inform. Syst., 2007, vol. 4781, pp. 468–477. [10] S. Roomi, M. Anitha, and R. Bhargavi, “Accurate license plate local- ization,” in Proc. Int. Conf. Comput. Commun. Electr. Technol., 2011, pp. 92–99. [11] S. K. Kim, D. W. Kim, and H. J. Kim, “A recognition of vehicle license plate using a genetic algorithm based segmentation,” in Proc. Int. Conf. Image Process., 1996, vol. 1, pp. 661–664. [12] J. Xiong, S. Du, D. Gao, and Q. Shen, “Locating car license plate under various illumination conditions using genetic algorithm,” in Proc. ICSP,2004, vol. 3, pp. 2502–2505. [13] Z. Ji-yin, Z. Rui-rui, L. Min, and L. Yinin, “License plate recognition based on genetic algorithm,” in Proc. Int. Conf. Comput. Sci. Software Eng., Dec. 2008, vol. 1, pp. 965–968. [14] V. P. de Arau´ jo, R. D. Maia, M. F. S. V. D’Angelo, and G. N.R. D’Angelo, “Automatic plate detection using genetic algorithm,” in Proc. 6th WSEAS Int. Conf. Signal Speech Image Process., Sep. 2006, pp. 43–48.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 57
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
RF Controlled Sailing Robot for Oceanic Missions R. Srikanth Kumar1, K. Dhanunjaya2 1 2
PG Student, Department of ECE, ASCET, Gudur, Andhra Pradesh, India
Head of the Department, Department of ECE, ASCET, Gudur, Andhra Pradesh, India
Abstract: Ocean exploration and navigational research by supporting expeditions with computer vision techniques have shown potential for sailing robots development in order to make measurements at the surface. Sailing robots explores the science and technologies for the identification of underwater features. Key applications of sailing robot are measuring depth and sensing metals under the water. An idea presented is the robot that can sail on water, controlled by laptop keypad. Other modules are gps and camera.
necessary. The data received from sensors is also transmitted to laptop through these modules.
Keywords: sailing robot, laptop, gps, camera I. INTRODUCTION For oceanic missions such as finding metals, spy applications, measuring depth, etc., one has to travel on the boat. But it is risky, because climate can change suddenly on the ocean. For example cyclone may occur suddenly. Also while spying, the opposite one may fire and can result in death of driver controlling the boat. To overcome this, the boat has to be designed in such a way that it can be operated without a driver. So it has to use any means of wireless communications. In this paper RF pro wireless communication is presented. II. RF WIRELESS COMMUNICATION The commands for respective movements of sailing robot are send from laptop keypad through RF PRO wireless communication. The module used for this purpose is RFSv4.3. It provides easy and flexible wireless data transmission between devices. RFSv4.3 uses 2.4 GHz carrier frequency. On pressing keys on laptop keypad respective action takes place. For example on pressing 1 the sailing robot moves forward. For this project two RFSv4.3 modules are
ISBN NO : 378 - 26 - 13840 - 9
III. SENSORS A. Depth sensor:
One of the applications of sailing robot is to measure depth under water. For this an ultrasonic sensor is used. The module used is Ultrasonic ranging module HC-SR04. The basic principle depends on echo reception. A short 10s pulse is supplied to the trigger input. The module will send out an 8 cycle burst of ultrasound at 40 KHz. After hitting ground the pulse signal is reflected back. Now the range can be calculated through the time interval between sending trigger signal and receiving echo signal. The formula is, Distance = (time taken Ă&#x2014; velocity of sound)/2
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 58
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
B. Obstacle Sensor: Although the sailing robot is monitored through camera, there may be chances of collision with obstacles because of controller inactive. So to avoid this an ultrasonic sensor is used to detect the obstacles. When an obstacle is detected the sailboat automatically stops. C. Metal sensor:
VII. REFERENCES 1. www.google.com To detect metals under the water, a metal sensor is used. The metal sensor used is ID18 â&#x20AC;&#x201C; 3008NA. It is water proof. This module is useful in finding missing ships. IV. GPS
2. www.keil.com 3. www.nxp.com 4. www.robosoftsystems.co.in 5. www.royaltek.com 6. www.ideal.net.in
As the climate on ocean is unpredictable, there is a chance of missing of sailboat. This may occur because of cyclone or some other reasons. So a GPS module is used to detect the sailboat when missed. The GPS module used is Royaltek REB-1315S4. GPS module can also be used for navigation purpose. V. CAMERA This module sends continuous video to the laptop for continuous monitoring. The camera used in this project can send video from 50 to 100 meters distance if it in line of sight with receiver. This can also be used for spy applications. VI. X-CTU TERMINAL All the values received from sensors can be displayed on laptop using X-CTU terminal.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 59
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Solar Power Satellites
K. Deepak
K. Poojitha
IV B.Tech EIE Sree Vidyanikethan Engineering College
Abstract
IV B.Tech EIE Sree Vidyanikethan Engineering College power station in space to transmit electricity to Earth by way of radio waves-the Solar
Solar Power Satellites (SPS) converts solar energy in to micro waves and sends that microwaves in to a beam to a receiving antenna on the Earth for conversion to ordinary Electricity. SPS is a clean, large-scale, stable electric power source. For SPS Wireless power transmission is essential. WPT contains microwave beam, which can be directed to any desired location on Earth surface. This beam collects Solar Energy and converts it into Electrical Energy. This concept is more advantageous than conventional methods. The SPS will be a central attraction of space and energy technology in coming decades. It is not a pollutant but more aptly, a man made extension of the naturally generated electromagnetic spectrum that provides heat and light for our sustenance. Keywords: Solar Power Satellites; Microwaves;
Power Satellites. Solar Power Satellites (SPS) converts solar energy in to micro waves and sends that microwaves in to a beam to a receiving antenna on the Earth for conversion to Ordinary Electricity. SPS is a clean, large-scale, stable electric power source. Solar Power Satellites is known by a variety of other names such as Satellite Power System, Space Power Station, Space Power System, Solar Power Station, Space Solar Power Station etc. One of the key Technologies needed to enable the future feasibility of SPS is that of Microwave Wireless Power Transmission. WPT is based on the energy transfer capacity of microwave beam I e, energy can be transmitted by a well-focused microwave beam. Advances in Phased array antennas and rectennas have provided the building blocks for a realizable WPT system
1. Introduction The new millennium has introduced increased pressure for finding new renewable energy sources. The exponential increase in population has led to the global crisis such as global warming, environmental pollution and change and rapid decrease of fossil reservoirs. Also the demand of electric power increases at a much higher pace than other energy demands as the world is industrialized and computerized. Under these circumstances, research has been carried out to look in to the possibility of building a
ISBN NO : 378 - 26 - 13840 - 9
1.1 Why SPS Increasing global energy demand is likely to continue for many decades. Renewable energy is a compelling approach â&#x20AC;&#x201C; both philosophically and in engineering terms. However, many renewable energy sources are limited in their ability to affordably provide the base load power required for global industrial development and prosperity, because of inherent land and water requirements. The burning of fossil fuels resulted in an abrupt decrease in their availability. It also led to the green-house effect and many other environmental
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 60
Proceedings of International Conference on Developments in Engineering Research
problems. Nuclear power seems to be an answer for global warming, but concerns about terrorist attacks on Earth bound nuclear power plants have intensified environmentalist opposition to nuclear power. Earth based solar panels receives only a part of the solar energy. So it is desirable to place the solar panel in the space itself, where, the solar energy is collected and converted in to electricity which is then converted to a highly directed microwave beam for transmission. This microwave beam, which can be directed to any desired location on Earth surface, can be collected and then converted back to electricity. This concept is more advantageous than conventional methods. Also the microwave energy, chosen for transmission, can pass unimpeded through clouds and precipitations. 1.2 SPS –The Background The concept of a large SPS that would be placed in geostationary orbit was invented by Peter Glaser in 1968.The SPS concept was examined extensively during the late 1970s by the U.S Department of Energy (DOE) and the National Aeronautics and Space Administration (NASA). The DOE-NASA put forward the SPS Reference System Concept in 1979. The central feature of this concept was the creation of a large scale power infrastructure in space, consisting of about 60 SPS, delivering a total of about 300GW.But, as a result of the huge price tag, lack of evolutionary concept and the subsiding energy crisis in 1980-1981, all U.S SPS efforts were terminated with a view toreasses the concept after about ten years. During this time international interest in SPS emerged which led to WPT experiments in Japan.
www.iaetsd.in
difference between existing satellites and SPS is that an SPS would generate more power-much more power than it requires for its own operation. The solar energy collected by an SPS would be converted into electricity, then into microwaves. The microwaves would be beamed to the Earth’s surface, where they would be received and converted back into electricity by a large array of devices known as rectifying antenna or rectennas. Each SPS would have been massive; measuring 10.5 km long and 5.3km wide or with an average area of 56 sq. km. The surface of each satellite would have been covered with 400 million solar cells. The transmitting antenna on the satellite would have been about 1 km in diameter and the receiving antenna on the Earth’s surface would have been about 10 km in diameter. In order to obtain a sufficiently concentrated beam; a great deal of power must be collected and fed into a large transmitter array. The power would be beamed to the Earth in the form of microwave at a frequency of 2.45 GHz. Microwaves have other features such as larger band width, smaller antenna size, sharp radiated beams and they propagate along straight lines. Microwave frequency in the range of 2-3 GHz are considered optimal for the transmission of power from SPS to the ground rectennas site.The amount of power available to the consumers from one SPS is 5GW. The peak intensity of microwave beam would be 23 m W/cm². SPS has all the advantage of ground solar, it generates power during cloudy weather and at night. In other words SPS receiver operates just like a solar array. It receives power from space and converts it into electricity. This reduces the size and complexity of satellite.
1.3 SPS-A General Idea Solar Power Satellites would be located in the geo-synchronous orbit. The
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 61
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
1. The conversion of direct power from the photovoltaic cells, to microwave power on the satellites on geosynchronous orbit above the Earth. 2. The formation and control of microwave beam aimed precisely at fixed locations on the Earth’s surface. 3. The collection of the microwave energy and its conversion into electrical energy at the earth’s surface. Fig1: Basic Design Of SPS
2. Wireless Power Transmission Transmission or distribution of 50 or 60 Hz electrical energy from the generation point to the consumer end without any physical wire has yet to mature as a familiar and viable technology. The 50 Hz ac power tapped from the grid lines is stepped down to a suitable voltage level for rectification into dc. This is supplied to an oscillator fed magnetron. The microwave power output of the magnetron is channeled into an array of parabolic reflector antennas for transmission to the receiving end antennas. To compensate for the large loss in free space propagation and boost at the receiving end the signal strength as well as the conversion Efficiency, the antennas are connected in arrays. A simple radio control feedback system operating in FM band provides an appropriate control signal to the magnetron for adjusting its output level with fluctuation in the consumers demand at the receiving side. The overall efficiency of the WPT system can be improved by-Increasing directivity of the antenna array. Using dc to ac inverters with higher conversion efficiency .Using schottky-diode with higher ratings. 2.1 Microwave Power Transmission In SPS The microwave transmission system have had three aspects:
ISBN NO : 378 - 26 - 13840 - 9
The key microwave components in a WPT system are the transmitter, beam control and the receiving antenna called rectennas. At the transmitting antenna, microwave power tubes such as magnetrons and klystrons are used as RF power sources. Rectenna is a component unique to WPT systems. The following section describes each of these components in detail. 2.2 Transmitter The key requirement of a transmitter is its ability to convert dc power to RF power efficiently and radiate the power to a controlled manner with low loss. The transmitter’s efficiency drives the end-to-end efficiency as well as thermal management system. The main components of a transmitter include dc-to-RF converter and transmitting antenna. Power distribution at the transmitting antenna = (1-r²), where r is the radius of antenna [7].There are mainly three dc-to-RF power converters: magnetrons, klystrons and solid state amplifiers.
Fig. 2: Klystron amplifier
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 62
Proceedings of International Conference on Developments in Engineering Research
Klystron Fig. 2 shows the klystron amplifier [15]. Here a high velocity electron beam is formed, focused and send down a glass tube to a collector electrode which is at high positive potential with respect to the cathode. As the electron beam having constant velocity approaches gap A, they are velocity modulated by the RF voltage existing across this gap. Thus as the beam progress further down the drift tube, bunching of electrons takes place. This variation in current enables the klystron to have significant gain. Thus the catcher cavity is excited into oscillations at its resonant frequency and a large output is obtained. The tube body and solenoid operate at 300°C and the collector operates at 500°C. The overall efficiency is 83%. The microwave power density at the transmitting array will be 1 kW/m² for a typical 1 GW SPS with a transmitting antenna aperture of 1 km diameter. If we use 2.45 GHz for MPT, the number of antenna elements per square meter is on the order of 100. 2.3 Rectenna Brown was the pioneer in developing the first 2.45GHz rectenna. Rectenna is the microwave to dc converting device and is mainly composed of a receiving antenna and a rectifying circuit. Fig .4 shows the schematic of rectenna circuit. It consists of a receiving antenna, an input low pass filter, a rectifying circuit and an output smoothing filter. The input filter is needed to suppress radiation of high harmonics that are generated by the non-linear characteristics of rectifying circuit. Because it is a highly nonlinear circuit, harmonic power levels must be suppressed.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
Fig. 3: Schematic of rectenna circuit.
For rectifying Schottky barrier diodes utilizing silicon and gallium arsenide are employed. Diode selection is dependent on the input power levels. The breakdown voltage limits the power handling capacity and is directly related to series resistance and junction capacitance through the intrinsic properties of diode junction and material. For efficient rectification the diode cut off frequency should be approximately ten times the operating frequency. Diode cut off frequency is given by ƒ=1/ [2_RsCj], where ƒ is the cut off frequency, Rs is the diode series resistance, Cj is the zerobias junction capacitance. 2.5 Recently Developed MPT Systems The Kyoto University developed a system called Space Power Radio Transmission System (SPORTS).The SPORTS is composed of solar panels, a microwave transmitter subsystem, a near field scanner, a microwave receiver. The solar panels provide 8.4 kW dc power to the microwave transmitter subsystem composed of an active phased array. It is developed to simulate the whole power conversion process for the SPS, including solar cells, transmitting antennas and rectenna system. Another MPT system recently developed by a team of Kyoto University, NASDA and industrial companies of Japan, is an integrated unit called the Solar Power Radio Integrated Transmitter (SPRITZ), developed in 2000. This unit is composed of a solar cell panel, microwave generators, transmitting
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 63
Proceedings of International Conference on Developments in Engineering Research
array antennas and a receiving array in one package. 3. Construction of SPS from Non Terrestrial Materials: Feasibility and Economics SPS, as mentioned before is massive and because of their size they should have been constructed in space. The aluminum and silicon can be refined to produce solar arrays. Among them are the shallow gravity wells of the Moon and asteroids; the presence of an abundance of glass, metals and oxygen in the Apollo lunar samples; the low cost transport of those materials to a higher earth orbit by means of a solar powered electric motor; the availability of continuous solar energy for transport, processing and living. One major new development for transportation is required: the mass driver. The mass driver is a long and narrow machine which converts electrical energy into kinetic energy by accelerating 0.001 to 10 kg slugs to higher velocities. The mass driver conversion efficiency from electrical to kinetic energy is close to 100 percent. 3.1 Microwaves-Environmental Issues The price of implementing a SPS includes the acceptance of microwave beams as the link of that energy between space and earth. Because of their large size, SPS would appear as a very bright star in the relatively dark night sky. SPS possess many environmental questions such as microwave exposure, optical pollution that could hinder astronomers, the health and safety of space workers in a heavy-radiation (ionizing) environment, the potential disturbance of the ionosphere etc. The atmospheric studies indicate that these problems are not significant, at least for the chosen microwave frequency. On the earth, each rectenna for a fullpower SPS would be about 10 km in diameter. This significant area possesses
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
classical environmental issues. These could be overcome by sitting rectenna in environmentally insensitive locations, such as in the desert, over water etc. However, the issues related to microwaves continue to be the most pressing environmental issues. On comparing with the use of radar, microwave ovens, police radars, cellular phones and wireless base stations, laser pointers etc. public exposures from SPS would be similar or even less. Based on well-developed antenna theory, the environmental levels of microwave power beam drop down to 0.1ìW/cm². Serious discussions and education are required before most of mankind accepts this technology with global dimensions. Microwaves, however is not a ‘pollutant’ but, more aptly, a man made extension of the naturally generated electromagnetic spectrum that provides heat and light for our substance. 4. Advantages and Disadvantages The idea collecting solar energy in space and returning it to earth using microwave beam has many attractions. The full solar irradiation would be available at all times expect when the sun is eclipsed by the earth. Thus about five times energy could be collected, compared with the best terrestrial sites. The power could be directed to any point on the earth’s surface. The zero gravity and high vacuum condition in space would allow much lighter, low maintenance structures and collectors. The power density would be uninterrupted by darkness, clouds, or precipitation, which are the problems encountered with earth based solar arrays. The realization of the SPS concept holds great promises for solving energy crisis. The concept of generating electricity from solar energy in the space itself has its inherent disadvantages also.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 64
Proceedings of International Conference on Developments in Engineering Research
Some of the major disadvantages are: The main drawback of solar energy transfer from orbit is the storage of electricity during off peak demand hours. The frequency of beamed radiation is planned to be at 2.45 GHz and this frequency is used by communication satellites also. The entire structure is massive. High cost and require much time for construction. Radiation hazards associated with the system. Risks involved with malfunction. High power microwave source and high gain antenna can be used to deliver an intense burst of energy to a target and thus used as a weapon. 5. Conclusion and Future Scope The SPS will be a central attraction of space and energy technology in coming decades. However, large scale retro directive power transmission has not yet been proven and needs further development. Another important area of technological development will be the reduction of the size and weight of individual elements in the space section of SPS. Large-scale transportation and robotics for the construction of large-scale structures in space include the other major fields of technologies requiring further developments. The electromagnetic energy is a tool to improve the quality of life for mankind. It is not a pollutant but more aptly, a man made extension of the naturally generated electromagnetic spectrum that provides heat and light for our sustenance. From this view point, the SPS is merely a down frequency converter from the visible spectrum to microwaves.
www.iaetsd.in
Microwave wireless power transmission technology”, IEEE microwave magazine, pp.46-57, Dec 2002. [3] J.C. Mankins,”A fresh look at space solar power: new architectures, concepts and technologies” in 38th Astronautical Federation. [4] Seth Potter, “Solar power satellites: an idea whose time has come [online] Available on www.freemars.org/history/sps.html, last updated on Dec.1998 [5] Consumer Energy Information: EREC Reference Briefs [online] Available on www.eere.gov/consumerinfo/rebriefs/123.ht ml,last updated on Apr.03. [6] Mc GrawHill Encyclopedia of Science and Technology, vol.16, pp.41.
References[1] Hiroshi Matsumoto, “Research on solar power satellites and microwave power transmission in Japan”, IEEE microwave magazine, pp.36-45, Dec 2002. [2] James O. Mcspadden & John C. Mankins,”Space solar power programs and
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 65
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Autonomous pick and place rover for long distance surveillance using ultrasonic sensors V.Sampath kumar1, Amarendra Jadda2 2
1 PG Student, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India. Assoc. Professor, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India
1 verappa1989@gmail.com amarendra.jadda@gmail.com
2
Abstract: The technology advancements takes place in the field of robotics. This paper presents the autonomous rover where it can be used for long distance surveillance. It is based on the ultrasonic sensors,Temperature sensors and humidity sensors meant for to detect the obstacles as well as to estimate the parameters such as temperature, humidity respectively. The wireless communication protocols used in between PC and Robo are zigbee and RF protocols.This robo is capable of pick and place of objects using the special type of grippers. In this paper, the usage of LPC2148 is another added advantage which makes the robotic system highly Reliable. Index Index Terms---- Autonomous Robot, Sensors, RF protocol, Zigbee protocol, DC motor, Camera, Pick and place robot
A.Pick and Place module: The name itself indicates that the objects are to be picked from one place and placed on another place. The prime objective is to pick the objects based on the environment which can handles any shape of the object. The picking arm can rotate upto 360 degrees by making the base of the robot fixed. B.Sensor module: I.
I.Introduction: Robots play a vital role because they will work under any conditions without the tiredness. They are used in Industries, IC manufacturing, Pharmaceutical etc.,. for different purposes. They may be static or dynamic robots operates by the commands given manually or simply by automated instructions. This robos can perform multiple tasks by attaching devices like sensors, cameras inorder to calculate the various parameters based on the application required. These robos can be controlled by either the microcontrollers or processors which are dependent on the protocols used such as zigbee and RF protocals. The dynamic robos are taking directions under the guidelines of the operator. II.
modules are used which are communicating with the PC based on the RF, Zigbee protocols respectively. LPC 2148 is used for controlling of the robo by giving the instructions. The various types of sensors are used such as temperature sensor, humidity sensor and ultrasonic sensor.
Temperature Sensor:
In General temperature sensors are used to sense the temperature and thereby transducers are converting these sensed temperature values to the electrical values. Here the temperature sensor used is LM35 which is analog in nature and its functional range lies in between 0 degree Celsius to 100 degree Celsius. It has a sensitivity of 10mV per degree Celsius.
Related Work:
This paper focuses on the dynamic approach of calculating various parameters based on the sensors used along with the pick and place mechanism. In this approach the camera and sensor
ISBN NO : 378 - 26 - 13840 - 9
Fig 1. Pin diagram of LM35
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 66
Proceedings of International Conference on Developments in Engineering Research
II.
www.iaetsd.in
Humidity Sensor:
Ă&#x152;t is used to measure the moisture content in the environment, designed to give the enhanced parameters of sensitivity and response time. Humidity sensor is highly stable and reliable. These sensors are constructed by a planar capacitor and another polymer layer to protect against the environmental effects that may be hazards. C.
Ultrasonic Sensor:
Fig 3. Zigbee Module
These sensors are used to find the obstacles if any that found in front of robot. Ultrasonic sensors are based on the principal of the Radar or Sonarwhere it gets the echo signal. It will act as a transceiver in which it generates the high frequency signal and analyse the reflected back echo signal. It measures the distances up to 2.5 meters at an accuracy of 1 centi- meter.
It provides low latency and duty cycle that supports for multiple network topologies such as static, mesh and star topologies. It will support upto 65,000 nodes on a network. It avoids the collision of packets and provides the acknowledgment status. IV.
Hardware Module:
It consists of the components such as DC motor, LPC 2148, and an LC293D. DC motor is the one which converts electrical energy into mechanical energy used for various industrial applications. The driver circuit is used for driving these motors where in which here LC293D is used as a driver circuit. The brief content of the these modules as follows: A.
DC motors:
The different type of protocols are used here in order to communicate between the PC and the Robot. The protocols that are used are Zigbee protocol and RF protocol respectively.
There are many types of DC motors are available in which most of the cases two types of motors most commonly used such as brushed and brushless types. Here in this paper mainly focusing on the brushed DC motor which will operate at 12V DC and 0.6A. The speed of the DC motor is controlled by either changing the voltage applied to armature or by varying the field current. Most of the DC motors are driven by the Driver ICs at present.
A.
B.
Fig2. Block Diagram of Ultrasonic Sensor
III.
Communication Protocols:
Zigbee protocol:
Itâ&#x20AC;&#x2122;s a wireless protocol based on the IEEE 802.15 standard that suits for low power devices communication. Generally it is intended for short distance communication but we may extend to long distance using intermediate devices.
ISBN NO : 378 - 26 - 13840 - 9
LC293D:
This IC acts as driver circuit in which it is available in the 16-bit DIP. It will control the two small motors simultaneously in either forward or reverse direction, in which 4 pins is sufficient in order to perform this tasks.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 67
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig 5. IC L293D
Fig4. DC Motor
There exists certain features and drawbacks of this IC that is used such as output current capability is limited to 600mA per channel in the sense this IC cannot drive the bigger motors. Most probably this would be fit for the small motors which suits for the Robot arm movements. If the output current exceed the 1.2 A repeatedly then the IC will be destroyed.
C.
Camera Module:
In this paper camera used is wireless camera used for both the transmitter and receiver section. Its range lies in between 100 and 150 metres based on the principal of the RF protocol. The camera used here operates at the range of 1.2 GHZ. The transmitter and receiver section uses the power supply voltage of 9v battery.
The supply voltage will maintain up to 36V maximum. If the output current exceed beyond the 1.2A then the IC gets destroyed so here in order to protect the IC from over temperature an internal sensor will be used for sensing the temperature. If the temperature will exceeds beyond the set point then automatically this sensor will send an alert and stops the motors. The additional feature of this IC is using clamp diode which protects the IC from the voltage spikes. In this IC if the voltage exceeds 1.5v then only the pin will gets enabled that makes suitable for high frequency applications like switching applications. Other than driving the motors it also be used to drive the solenoids, stepper motors etc., This is mainly based on the concept of H-bridge in which the voltage will be flow in either of the directions. By changing the voltage it will make the Robot to rotate in clockwise or anticlockwise direction. In this LC293D two H-bridge circuits are present inside the IC which will be able to rotate two dc motors independently. Here two pins are used namely pin 1 and pin9 respectively. For driving the motor with left H-bridge we enable the pin 1 whereas driving right H-bridge we make the pin 9 to be enable. If both the pins are disable then the motor used will be suspended in which it acts like switch.
ISBN NO : 378 - 26 - 13840 - 9
Fig 6. Wireless Camera D.
LPC 2148 Processor:
ARM is a family of instruction set architectures based on RISC architecture developed by ARM Holdings. This ARM holdings had reported that shipments was around 6100 million ARM-based processors to manufacturers of chips based on ARM architectures where the maximum percentile usage of smart phones and the remaining of 35 per cent of digital TVs and set-top boxes. As of 2014, ARM is the usage of 32-bit instructions is the maximum for major number of applications.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 68
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig 9. Side View of Robot
The Robot thus shows in various directions of different views where we used the sensors and camera based modules for surveillance. Fig 7. H-Bridge
V.
Experimental Results:
The robot used here is used to pick the objects and is rotated in 360 degrees rotation. The In this project is to implement a 32-bit low-power ARM processor into the development of a Mobile Robot using ARM Processor for Line Following Application mobile robot equipped with IR sensor capabilities. The scope of the project is LPC2148 Arm Processor DC brushless motor 100 RPM Motor is used. At the end of this project, the ARM processor has successfully for motor control and respond to the digital input from the interface of IR sensor to the Microcontroller. Fig 10. Front View of Robot
snapshots of the robot used in various views such as front view, side view and top view respectively.
VI.
Conclusion:
The Robot in this paper used various sensors which senses the temperature and humidity respectively. The camera is used to estimate the position and orientation of robot. This will be used to work at hostile environments where not possible for the humans to do that. In this paper this Robot will handles over the distance of 100 metres maximum. We may expect that this coverage of surveillance can be increased to a long distances where a major number of applications comes without the human intervention.
Fig 8. Top View of Robot
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 69
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
References: [1] Jacquier E, Polson NG, Rossi PE (1993) Priors and models for multivariate stochastic volatility.Unpublished manuscript, Graduate School of Business,University of Chicago. [2] L. Righetti, A. Ijspeert, Design methodologies for central pattern generators: an application to crawling humanoids, in: G.S. Sukhatme, S. Schaal, W.Burgard, D. Fox (Eds.), Proceedings of Robotics: Science and Systems,MIT Press, 2006. [3] J. Pisokas, U. Nehmzow, Experiments in subsymbolic action planning with mobile robots, in: Adaptive Agents and MultiAgent Systems II, in: Lecture Notes in Artificial Intelligence, Springer, 2005, pp. 80–87. [4] Franklin Hanshar and Beatrice Ombuki-Berman. Dynamic vehicle routing using genetic algorithms.Applied Intelligence, 27(1):52–91, 2007. [5] Thomas L. Dean and Michael P.Wellman. Planning and control. Morgan Kaufmann Publishers Inc., 1991. [6] G. Antonelli and S. Chiaverini, “Linear estimation of the physical odo- metric parameters for differential-drive mobile robots,” Auton. Robots, vol.23, no. 1, pp. 59–68, Jul. 2007. [7] A. Martinelli, N. Tomatis, and R. Siegwart, “Simultaneous localization and odometry self calibration for mobile robot,” Auton. Robots, vol. 22, no. 1, pp. 75–85, Jan. 2007. Amarendra Jadda is working as Assiciate Professor in ECE Dept, ASCET, GUDUR. He has been guiding UG & PG students since three years in this institution. He pursued his M.Tech From JNTUH, Kukatpally Hyderabad. He presented papers in four international journals & four international conferences. His intresting fields are Communications &Signal Processing, Embedded Systems.
Sampath kumar veerappa did his B.Tech in Electronics and communication engineering from Prathyusha Engineering College, Chennai. He is currently pursuing his M.Tech degree in Embedded systems from audisankara college of engineering, Gudur.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 70
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Intelligent Bus Alert System For Blind Passengers N.Surendra Yadav1, Tulasi Sanath Kumar2 1
2
PG Student, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India. Asst. Professor, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India 1 2
cupid.suri@gmail.com tulasisanath@gmail.com
Abstract:Talking signs, guide cane, echolocations are all useful innavigating the visually challenged people to reach theirdestination, but the main objective is not reached that it fails tojoin them with traffic. In this project we propose a bus systemusing wireless sensor networks (WSNs).The blind people in thebus station is provided with a Zig-Bee unit which is recognized bythe Zig-Bee in the bus and the indication is made in the bus thatthe blind people is present in the station. So the bus stops at theparticular station. The desired bus that the blind want to take isnotified to him with the help of speech recognition systemHM2007. The blind gives the input about the place he has toreach using microphones and the voice recognition systemrecognizes it .The input is then analyzed by the microcontrollerwhich generates the bus numbers corresponding to the locationprovided by the blind. These bus numbers are converted intoaudio output using the voice synthesizer APR 9600. TheZig-Beetransceiver in the bus sends the bus number to the transceiverwith the blind and the bus number is announced to the blindthrough the headphones. The blind takes the right bus parked infront of him and when the destination is reached it is announcedby means of the GPS-634R which is connected with the controllerand voice synthesizer which produces the audio output. Thisproject is also aimed at helping the elder people for independentnavigation.
of navigation for the blind are very complex andtroublesome especially when they walked down in street andalso navigate to distant places by public transport system. Fora visually impaired person, doing things such as reading trafficsignals and street signs can be extremely challenging, if not itis impossible to do.
I. INTRODUCTION
In order to overcome these challenges, a visually impairedperson might use walking cane, guide dog, and sighted guide. These alternatives also called as assistive devices can behelpful to the blind but not so effective. The sighted guide canbe immensely effective, as well provide social comfort, but itrestricts the independence of the blind individual. Guide dogsand walking canes allow for a more independent means oftravelling, but they are limited in unfamiliar environments.RFID is feasible and cost effective but it is more suitable forindoor communication only. Also it provides only one waycommunication and a very short range of identification. Asystem with an augmented walking cane, a pair of augmentedglasses and identifiable items tagged with semacode/datamatrix tags is used for outdoor navigation of blind people. If aman has to take the bus, he walks along the pavement and hiswalking cane recognizes a tag. But the image quality of theweb camera is fairly poor. Tag recognition in darkness or inbad lightning conditions might be a problem. Another issue isthat camera needs a visual, so if a tag is hidden behind aperson or another object, then the camera cannot detect it.Tags on all environments will properly contaminate theenvironment and meet resistance from many citizens.
Out the 6.7 billion people that populate the world,161 million are visually impaired. Each visually impairedindividual faces different challenges based on their specificlevel of vision. With the rise of various support-basedorganizations, more visually impaired people have been giventhe opportunity to education and many other means. But stillthe issues
To overcome the drawbacks of currently availableassistive devices, we propose a Wireless sensor network systemwith Zig-Bee for blind identification by the bus and embeddedsystem for providing the bus number and [mally GPS fordestination indication. Wireless sensor network (WSNs)consists of sensors that continuously
Index Terms:Wireless sensor networks, Speech Recognition System,Voice Synthesizer, GPS, Zigbee.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 71
Proceedings of International Conference on Developments in Engineering Research
monitors theenvironmental conditions and send their data to the mainnetwork [51. Zig-Bee is an embedded device for use in a WSNwhich is tiny in size. These nodes have processing andcomputational capability and generally consist of an RFtransceiver, memory, on board sensors/actuators and a powersource [71. Zig-Bee have CC2420 which is a true single-chip 2.4GHz IEEE 802.15.4 compliant RF (Radio Frequency)transceiver designed for low-power and low-voltage wirelessapplications so we can send or receive useful informationthrough using this chip. The number of the bus parked in frontof the blind is send to the Zig-Bee in the blind system. Anotherfunction of Zig-Bee is identification of blind in the bus station.If both the numbers match the buzzer in the bus unit alarmsand indicates the driver that there is blind in the bus station. The software part is Embedded C with MPLAB IDE(Operating System) for programming the controller. MPLABIDE runs as a 32-bit application on MS Windows, is easy touse and includes a host of free software components for fastapplication development and super-charged debugging.MPLAB IDE also serves as a single, unified graphical userinterface for additional Microchip and third party software andhardware development tools. II. DESIGN PRINCIPLES OF ELECTRONIC TRAVELLING AIDS The most important travelling aid for a visually impairedperson is still the white cane. It is after all an excellentexample of a good travelling aid: multifunctional, cheap andreliable. It also tells others that the person is visuallyimpaired. Another irreplaceable travelling aid is a guide dog.Among other things the dog is also a friend and a companion. In studies about visually impaired person navigation it hasbeen noted that even a small amount of extra informationabout the environment makes a remarkable increase inperformance. Also it seems that a good travelling aidshould produce only small amounts of meaningfulinformation and the ETA should not block hearing or othersenses so that the visually impaired can still use theirtraditional methods to acquire information about theenvironment. If the user needs to concentrate heavily on usingthe ETA, he or she has no capacity left for normalenvironment perception. Therefore, instead of trying to develop ETAs to replaceprimary travelling aids, one should develop complementarysystems.Navigation systems have usually worked well in smallscaleimplementations,
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
but a large-scale implementation maybe extremely expensive (especially with beacon basednavigation systems). The amount of visually impaired personsof the population is small (~ 1,6 %) and therefore largeinvestments to special infrastructure are not sensible.As an example, there have often been suggestions aboutequipping buses with radio transmitters to help the visuallyimpaired to know when the bus is coming. The visuallyimpaired would in turn carry a radio receiver.In Prague there is a pilot system in operation. However, forexample in Finland, where we have about 80 000 visuallyimpaired and 10 000 buses, a similar system would cost atleast 10 M€ just for the bus transmitters. Other methods need to be found to ensure that the visuallyimpaired persons have equal possibilities to access sameservices than all the other citizens. 2.1. Visually Impaired Persons and Public Transport Generic travelling difficulties for the visually impairedpersons are localization and environment perception, selecting and maintaining the correct heading, detecting obstaclesabove waist and detecting unexpected roadworks. If we examine problems a visually impaired person meetswhen using public transport, we recognize the following list(the list depends slightly of the transportation): • planning a trip • finding a stop or station • finding an entrance to the station • navigating inside the station • finding the right platform and waiting place • knowing when the right vehicle arrives • finding a vehicle entrance • payment • finding a seat • receiving passenger information during the trip • depart on the right stop • finding the destination. Most of these tasks are trivial for the sighted persons, butvery difficult for the visually impaired. For example therehave been cases when a blind has spent several hours at a busstop, because he couldn't recognize the arrival of the rightvehicle. For true door-to-door navigation for a visually impairedthere are requirements for continuous positioning, continuous(Internet) access to real time
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 72
Proceedings of International Conference on Developments in Engineering Research
public transport information accurate map data roadworkinformation.
andavailability of together with
Electronic maps are designed for car navigation and notsuitable for pedestrian route planning purposes. Informationabout pavements and pedestrian crossings is collectedseparately and not included in typical map data. Door-to-doorguidance requires map data pointing entrances to houses andcontinuous guidance would require indoor maps and indoorpositioning. These are generally unavailable today. Nevertheless, in our studies we did not find many specificinformation needs for the visually impaired group alone. Theinformation needed and sought for is in most cases useful forall passengers or already accessible to some user groups. Onlythe means for a visually impaired to access the informationwould be different. Therefore it is very important to offeradditional interface, when new passenger informationservices are designed. III. NOPPA ARCHITECTURE AND PROTOTYPE Our approach is to improve public transport accessibilityby creating access to passenger information with a personalmobile device rather than building physical infrastructure.NOPPA architecture (fig.1) is based on public and/orcommercial information services and databases available via the Internet, a client-server approach with near continuousTCP/IP connection (GPRS for practical reasons) andprogrammable mobile devices with capabilities for speechuser interface (mainly speech synthesis) and satellitepositioning.
www.iaetsd.in
The terminal devices in NOPPA user tests have by far beenvarious PDAs with Microsoft's Pocket PC / Pocket PC PhoneEdition operating system and there is also an early version formobile phones with Symbian Series 60 operating system. Design goals for NOPPA were: • Easy and fast to use (preferably faster than any traditionalmethod) • Affordable to the user • Access to public transportation passengerinformation systems
and
• Applicable both indoors and outdoors • Integration of personalnavigation
products
and
services
for
• Modular, easy to update, easy to add functions • Speech user interface • Easy to customize for various user groups and purposes. The Information Server is an interpreter between the userand Internet information systems. It collects filters andintegrates information from different sources and delivers theresults to the user. The server handles speech recognition (e.g. from 13200street and destination names) and functions requiring eitherheavy calculations or large data transfer from the Internet.The data transfer between the server and the client is kept inminimum. The client terminal (fig.2) holds speechsynthesis, user interface, positioning and most of routeguidance. The user interface is menu-based and selections aredone with hardware buttons and speech input. As the mobile devices gain more memory and fasterprocessors some of the speech recognition work can be donein the user terminal which will further reduce the need fordata transfer between the client and the server. It will alsoenable menu selections with speech user interface when thereis no server connection. Nevertheless, the speech recognitionrequires a very large vocabulary (street names) which also hasto be updated from time to time, so it may be unpractical tocompletely do the processing in the terminal.
Fig.1. Architecture of the NOPPA system
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 73
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
NOPPA terminal software with speech synthesis needs tobe installed on the device, completely replacing theunderplaying operating system's user interface. If the operatingsystem supports a screen reader for example, more functions(such as phone calls, SMS and MMS) can be left to originalsoftware. The first prototype followingcharacteristics:
system
has
the
• Speech recognition and synthesis • 6 simultaneous users per single server computer (a 2 GHz PC) for speech processing time limits • Access to three route planners (commuter and intercitytraffic both bus and train, also a possibility to calculatecar navigation type of routes) • Guidance and route following during a trip • Personal in-vehicle stops announcements • Roadwork information (connection to a city's database) • Access to some bus, tram and train real time informationsystems (only early development) • Flight departure information at the largest airport in Finland, real time • Several news services, local weather • Watch • Memo • GSM phone implementation)
and
SMS
services
The system must be designed to handle connection failures so that they don't break guidance or prevent usingother functions (phone call, SMS, memo etc).Commercial sensitive GPS receivers are able to operateinside a bus and a tram, but still greatly benefit from antennaplacement near window. Also GPS receivers' slow time tofirst fix (TTFF, typically 30-60 seconds) can be a problemwhen turning a GPS first on or leaving a building after beinga long time inside with no update to receiver satellite data.
(basic
• Bluetooth and GPRS connectivity (also WLAN possible) • GPS and GSM positioning, optional pedometer andcompass unit • Indoor navigation features based on Bluetooth, WLANpositioning or compass/pedometer • Own recorded walking routes, basic GPS functions • Search of current address • POI (Point of Interest) and AOI (Area of Inters)databases
ISBN NO : 378 - 26 - 13840 - 9
Fig.2. NOPPA Pocket PC terminal and a Bluetooth GPS receiver The prototype is now at user evaluation phase. Usability,reliability and recovery after an error are known to beimportant issues. For example a continuous GPRS serverconnection is not possible when moving in a train, elevator orbasement.
The speech output in guidance and in describing a routemust be carefully planned to avoid misunderstandings and tohelp create a mental image of the route. The program shouldnot try to give more accurate guidance than it safely can. Forexample when standing near a bus stop, if the program wouldadvice that "the bus stop is 10 meters forwards", the usermight very well end up standing on a driveway. Combinedinaccuracy of GPS positioning and map data is very oftenover 10 meters and the program should not really try to guidethat short a distance (at least not require the user to move),even though there would seem to be a clear differencebetween GPS and target coordinates. The difficulty is to tell the user without misunderstandings,that the calculated target is maybe 20 meters forwards, but theuser has to find the exact
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 74
Proceedings of International Conference on Developments in Engineering Research
location himself and it may not beeven safe to move the full 20 meters. Often there is someinformation even in the short distances, so the user mightwant to hear the target distance and direction after all, insteadof just hearing "the target is near".In practice, one must take into account that map data canhave outdated information or inaccuracies, positioning can beunavailable or inaccurate, or wireless data transmission is notalways available. Therefore a lot of responsibility is left forthe user and guidance is complementary. IV. RESULTS AND DISCUSSION When the person reaches the bus station, he canfind the buses that pass through a particular location with thehelp of voice recognition system and voice synthesizer. When the bus approaches the bus station, there is anindication in the bus by the beep sound of a buzzer that thereis a blind person available in the bus station.This is achieved with the help of Zig-Bee unit both in the busunit and blind unit. Finally when the bus reaches the stationthe bus number is announced to the blind through headphones. There are currently available systems for the outdoornavigation but they will not assist in travelling to unfamiliarareas. Some systems use PDA which is not so economic andcannot be afforded by all. In most of the systems RFID tagsare used which are required in 1000s of numbers for trackingof route. Also it provides only one way communication. Thesystem we use is a mobile unit, weightless and economicallyfeasible.
www.iaetsd.in
VI. REFERENCES [1] G.Lavanya ME, Preethy. W, Shameem.A, Sushmitha.R, â&#x20AC;&#x153;Passenger BUS Alert System for EasyNavigation of Blindâ&#x20AC;?, 2013 International Conference on Circuits, Power and Computing Technologies. [2] Baudoin,G., Sayah,J, Venard, O. and EI Hassan, B.(2005), 'Simulation using OMNeT++ of the RAMPE systemanInteractive Additive Machine helping blinds in PublicTransports', EUROCON, Belgrade,pp.1-5. [3] Bolivar Torres, Qing Pang, (2010), 'Integration of anRFID reader to a Wireless sensor network and its use to identify an individual carrying RFID tag', InternationalJournal of ad hoc. sensor& ubiquitous computing ,voU,no.4,pp.1-15. [4] Brendan D Perry, Sean Morris and Stephanie Carcieri,(2009), 'RFID Technology to Aid in Navigation andOrganization for the Blind and Partially Sighted', pp. 1-52. [5] HerveGuyennet, KamalBeydoun and Violeta Felea,(2011), 'Wireless sensor network system helping navigation ofthe visually impaired', IEEE international conference onInformation and Communication Technologies: from Theoryto Applications, version 1, pp. 1-5. [6] Hyn Kwan Lee, Ki Hwan Eom , Min Chul Kim and TrungPham Quoc, (2010), 'Wireless Sensor Network Apply for theBlind U-bus System', International Journal of u- and eservice,Science and Technology,VoI.3,No.3,pp.l3-24.
V. CONCLUSION Primarily, the blind person in the bus station is identifiedwith RF communication. The blind informs the location heneeds through the microphone which is given to the voicerecognition system which produces the output of bus numbersin the voice synthesizer unit which is heard in headset. Thenthis location is transmitted to the transceiver in the bus. If thenames in the transceiver in the bus matches with that of thename send by the blind, then there is an alarm in the bus unitalerting the presence of blind and a voice to the user's headsetthat the particular bus has arrived. With the help of GPStracker connected with audio output the destination chosen bythe blind is intimated when the bus reaches the correctlocation. PDA's can be used for GPS tracking but it is not costeffective.
ISBN NO : 378 - 26 - 13840 - 9
[7] Jack Loomis,M. and Roberta Klatzky,L. (2008),'Navigation System for Blind', Massachusetts Institute ofTechnology, Vol 7, No.2, pp.193-203. [8] Jain.P.C ,Vijaygopalan.K.P. (2010), 'RFID and WirelessSensor Networks', Proceedings of ASCNT, CDAC, Noida,India, pp. 1 - 11. [9] Loc Ho ,Melody Moh, Teng-Sheng Moh and ZacharyWalker (2007) , 'A Prototype on RFID and Sensor Networksfor Elder Health Care', Taylor & Francis Group, LLC,pp.314-317. [10] Oyarzun, C.A and Sanchez, J.H. (2008), 'Mobile audioassistance in bus transportation for the blind', Department of Computer Science, University of Chile, pp.279-286.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 75
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
[11] Ravi Mishra and SornnathKoley, (2012),'Voice OperatedOutdoor Navigation System For Visually Impaired Persons',International Journal of Engineering and Technology,Vol 3,Issue 2,pp.l53157. [12] RiazAhamed, S.S (2009), 'The Role of ZigBeeTechnology In Future Data Communication System', Journalof Theoretical and Applied Information Technology,India, pp.129-134.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 76
Proceedings of International Conference on Developments in Engineering Research
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in 7
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 77
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Traffic Sign Recognition for Advanced Driver Assistance System Using PCA Kande Prasyam1, S. Himabindu2 1
2
PG Student, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India. Asst. Professor, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India 1
499prasyam@gmail.com 2 bindu437@gmail.com
Abstract: Traffic sign Recognition plays a vital role for the drivers in order to avoid the hurdles like speed breakers, narrow bridge or even accident zone etc. This paper presents the effective recognition of traffic signs using Principal component analysis. This could be done by placing a camera which captures the road sign images and it will be displayed as a video file in the GUI. This video file is converted into frames called array indexing. Here this technique uses different methods of image processing such as image segmentation, sign recognition and sign classification. The Eigen values of these images calculated and given to LPC 2148 processor where it will be interfaced with the audio amplifier and shows the sign direction in LCD. Index Terms---- Road sign, Principal Component analysis, Graphical user interface, Eigen values, Eigen vectors, LPC 2148.
I.
Introduction
Image itself a matrix, it will be arranged in the forms of Rows and columns. For comparing of the similar images Independent Component analysis is sufficient but for comparing different images with different Eigen values principal component analysis came into picture. Here the Eigen values are calculated and are compared with the database values so that how close the value matches that would be treated as image for sign recognition. The primary objective of this paper is to extract the details of type of sign that exists in sign board and intimates to the driver through the voice alert. The other advantage is the voice alert is in the language of the driver.
II.
ISBN NO : 378 - 26 - 13840 - 9
RELATED WORKS
This paper focuses on the image processing modules and hardware modules where for interfacing between them RS232 cable is used. The software module consists of three modules such as image segmentation, sign recognition and sign classification and hardware components such as LPC 2148, audio amplifier and LCD Display. A. Image Processing module: This module by name itself indicates that processing of image using different techniques such as RGB Colour segmentation, Recognition of signs and classification of them. Image segmentation: Initially sign board images are captured using camera and can be segmented in order to determine the exact boundaries of that image used for effective analysis. This colour segmentation is used to be converted to 2D image and then calculate the Eigen values easily Sign Recognition: After segmentation of image the sign is recognised from the sign board used for advanced driver assistance. These recognized signs given as an input to the classification module. Sign Classification: The classification stage includes compares the signs that are recognised and with the database images.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 78
Proceedings of International Conference on Developments in Engineering Research
Block Diagram: Sign board images
PC
Audio amplifier
RS232
LPC 2148
LCD Display
www.iaetsd.in
Where det() indicates the determinant and this above equation is also known as characteristic equation of A. If A is nxn , then there are n solutions or n roots of the characteristic polynomial. This characteristic polynomial is of order n. Hence there are n Eigen values of A satisfying the equation. AXi=λXi
Speaker
(4)
Where i=1,2,3,….n
Indications
Fig 1. Block Diagram of image processing module The sign board images gives as an input to the personal computer where it processes the image and compares the image with the database images and finally selects the image that is closer to database images. The extracted image after comparing with the database images will send to the RS232 cable. This RS232 cable sends the index number to LPC 2148 that is generated from Personal computer using Keil software. ARM processor interfaces with the audio amplifier as well as displaying the type of sign in LCD.
If all the Eigen values are distinct, there are n associated linealy independent eigenvectors, whose directions are unique, which meant for an n dimensional Euclidean space. Eigen Sign Approach: The Eigen values of the input signs captured is compared with the database signs. If any matching occurs with the database image then accordingly based on the index number it will shows the turn right, turn left, turn curve etc., based on the assignment of the index value to the corresponding sign.
Eigen Values and Eigen Vectors: Eigen vectors are non-zero vectors of a linear operator and result in a scalar multiple of them when operated on by the operator. The scalar then called the Eigen value where In association with the Eigen vector. The property of matrix is in which when a matrix acts on it only the vector magnitude is changed but not the direction. Consider the Eigen vector of X where A is a vector function AX=λX
(1)
may be known or unknown that is captured by a camera and we get the weights associated with the Eigen signs, that linearly approximate the sign or can be used to reconstruct the sign. Now these weights are compared with the weights of the known sign images that are available in database
so that it can be recognized as a known sign.The Euclidean distance between the image projection and known projections is calculated; the
By using equation 1 we get equation 2 (A-λI)X=0
When the Sign image to be recognized
(2)
sign image is then classified as one of the signs with minimum Euclidean distance.
where I is the n x n Identity matrix. Mathematically calculations: The above mentioned is a homogenous system of equations and we know that a non trivial solution exists only when
det (A-λI) = 0
ISBN NO : 378 - 26 - 13840 - 9
(3)
Let a sign image I(x,y) be a two dimensional N by N array of (8-bit) intensity values. An image may be considered as a vector
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 79
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
of dimension N2, for a typical image it will be a
have m eigenvectors instead of N2. Premultipying
256 by 256 size and would be the vector of
equation 6 by A, we have
dimension 65,536 or equivalently we will say that
AA TAvi = μi Avi
65,536-dimensional space. An ensemble of images,
(7)
then, maps to a collection of points in this huge
The right hand side gives us the M Eigen signs of
space. Principal component analysis would find the
the order N2 by 1.All such vectors would make the
vectors that best account for the distribution of the
image space of dimensionality M.
sign images within this entire space. As the accurate reconstruction of the sign Let us consider a set of sign images be
is
not required,
we can now
reduce the
T1,T2,T3,….TM. This sign images data set has to be
dimensionality to M’ instead of M. This is done by
mean adjusted before calculating the covariance
selecting the M’ Eigen signs which have the largest
matrix or Eigen vectors. The average sign is
associated Eigen values. These Eigen signs now
M
calculated as Ψ = (1/M) Σ1 Ti, Each image in the
span a M’-dimensional subspace instead of N2. A
data set differs from the average sign by the vector
new image T is transformed into its Eigen sign
Ф = Ti – Ψ.This is actually mean adjusted data. The
components.
covariance matrix is wk = ukT (T - ψ) C = (1/M) Σ
M 1
Φ i Φ iT
(8)
(5) where k = 1,2,….M’.
= AAT
The Euclidean distance of the weight
where A = [ Φ 1, Φ2, …. ΦM].
vector of the new image from the sign class weight The covariance matrix considered here is a
vector can be calculated as follows,
N2 by N2 matrix and would generate N2 εk = || Ω – Ωk||
eigenvectors and eigenvalues. It is impractical to calculate with image sizes like 256 by 256, or even lower than that.
where
(9)
Ωk is a vector describing the kth sign
class.Euclidean distance. The sign is classified as
An effective solution is needed to
belonging to class k when the distance εk is below
calculate the Eigen vectors. Set of images that are
some threshold value θε. Otherwise the face is
considered is less than the no of pixels in an image
classified as unknown. Also it can be found
(i.e M < N2), then we can solve an M by M matrix
whether an image is a sign image or not by simply
instead of solving a N2 by N 2 matrix. Consider the
finding the squared distance between the mean
covariance matrix as ATA instead of AAT
adjusted input image and its projection onto the face space.
The eigenvector vi can calculated as follows, ε2 = || Ф - Фf || ATAvi = μivi
(6)
(10)
where Фf is the face space and Ф = Ti – Ψis the mean adjusted input.
where μi is the eigenvalue. Here the size of covariance matrix would be M by M.Thus we can
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 80
Proceedings of International Conference on Developments in Engineering Research
Using these we can say whether the image as
www.iaetsd.in
Block Diagram:
known sign image, unknown sign image and not a sign image.
Sign image
Normalized sign image Feature vector
Image acquisition
B. Interfacing Module: In generally most of the personal computers are provided with two serial ports and one parallel port. A parallel port sends and receives data bits very faster nut the required number of wires is more whereas for serial communication it will send one bit at a time through the single wire and hence slower but the required number of wires are less. RS232 is meant for serial communication transmission of data in order to connect the DTE (Data Terminal Equipment) and DCE (Data Communications Equipment). For example, connecting a computer terminal with the printers, modems, UPS and other peripheral devices. RS232C is the latest one where RS232 is Recommend Standard number and C is the latest revision of the standard. It specifies that 25-pin D connector and most of the PCs are equipped with the male type D-connectors consists of only 9 pins. III. Hardware Module: After the selection of image based on the index value obtained from RS232 cable is given as an input to the hardware such as LPC2148. This LPC2148 will be interfaced with both the audio amplifier and LCD display.
Preprocessing
Sign database
Feature Extractor
Training sets
Classifier
‘known’ or ‘unknown’ Fig 2. Block diagram of typical sign recognition system
B. Liquid Crystal Display: It is very thin and flat panel used for electronically displaying information such as text, images as well as moving pictures. It has enormous applications include monitors for personal computers, televisions, instrument panels, and other devices ranging from aircraft cockpit displays to every-day consumer devices such as gaming devices, clocks, watches, calculators, and telephones. Hardware Circuitry:
A. LPC 2148: ARM7 is most widely used in embedded system application such as ranging from mobile phones to automotive braking systems. The number of transistors used in ARM7 is fewer which reduce the costs and power consumption. The ARM7 is based on a 16bit/32 bit with real-time emulation and embedded trace support. This also provided with the 512 kilobytes of embedded high speed flash memory. This LPC2148 also consists of 128-bit wide memory interface and unique accelerator architecture which will enable 32-bit code execution at maximum clock rate.
ISBN NO : 378 - 26 - 13840 - 9
Fig 3.
Before displaying the type of sign
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 81
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig 5. Traffic sign recognition results
Fig 4. Voice alert in accordance with the type of sign displayed in LCD
C. Audio amplifier: It is an electronic device that increases the strength of audio signals that pass through it. Audio amplifies up to the level that is suitable for driving loudspeakers. The different amplifiers that exist are car audio amplifier, PC audio amplifier, TV audio amplifier etc., and can be chosen based on the application. Finally this audio amplifier output is given to the speakers. IV. Experimental Results: The input images are captured by camera where it will be processed and compared with the database images. If any matching occurs then the corresponding sign image will be displayed and simultaneously the type of sign is also displayed at the bottom of the GUI. This would be done for various sign images and voice alert also provided with respect to the sign image displayed. LCD screen is also interfaced in order to display the type of sign that is present.
In the above figure shown that the sign board is displayed on the GUI as well as it will show the type of sign in the box at the bottom. Here it will display the Right hand curve sign image and like manner we will process and display the image using PCA in an effective way. V. Conclusion: This paper provides an effective recognition of traffic signs using PCA algorithm and thereby providing the voice alert to the drivers as well as display. After processing of the input sign image in GUI, then the allotted index number in accordance with the sign image is given to RS232 cable. Based on the index number the corresponding sign image direction will be audible in the speakers and displayed on LCD module. In the future we may expect the same feature of traffic sign recognition without the interfacing of GUI module with the LPC2148 using RS232 cable in TMS320CXXX in which the performance also be increased. References: [1] Prof. V.P. Kshirsagar, M.R.Baviskar, M.E.Gaikwad, ” Face Recognition Using Eigen faces”. [2] “Facial Recognition using Eigenfaces by PCA” by Prof. Y. Vijaya Lata, Chandra Kiran Bharadwaj Tungathurthi, H. Ram Mohan Rao, Dr.A. Govardhan, Dr.L.P. Reddy, International Journal of Recent Trends in Engineering, Vol. 1, No. 1, May 2009. [3] “Face Recognition using Eigenface Approach” by Vinaya Hiremath, Ashwini Mayakar,. [4] “Face Recognition using Eigenfaces and Neural Networks”
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 82
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
by Mohd Rozailan Mamat, Mohamed Rizon, Muhammad Firdaus Hasim, American Journal of Applied Sciences 2 (6): 1872-1875, 2006 ISSN 1546-9239, 2006 Science Publications. [5] L.D. Lopez and O. Fuentes, "Color-based road sign detection and tracking", Proc. Image Analysis and Recognition(ICIAR), Montreal,.CA, Agust 2007. [6] ARM Data Manual sheets http://www.keil.com/dd/docs/datashts/philips/lpc2141_42_44_4 6_48.pdf [7] A. D. L. Escalera, J. M. A. Armingol, and M. Mata, "Traffic sign recognition and analysis for intelligent vehicles ", Image and Vision Computing, vol. 21, pp. 247â&#x20AC;&#x201C;258, 2003.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 83
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Ethernet Based Intelligent Security System Ch.Jagadeesh1, Tulasi Sanath Kumar2 1
2
PG Student, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India. Asst. Professor, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India 1
jagadeeshchalla1508@gmail.com 2 tulasisanath@gmail.com
Abstract:In this paper we design and implement an embeddedsurveillance system by use of ultrasonic signal coding ofultrasonic sensors with multiple pyroelectric infrared sensors(PIR) to detect an intruder in a home or a storehouse. The PIRsensors are placed on the ceiling, and the ultrasonic sensormodule consists of a transmitter and a receiver which are placedin a line direction; however, ultrasonic sensors with the samefrequency are subject to interference by crosstalk with each otherand have a high miss rate. To overcome these disadvantages ofthe ultrasonic sensor, our design reduces the miss rate from theenvironmental interference by using an ultrasonic coding signal.Both ultrasonic sensors and PIR sensors are managed by themajority voting mechanism (MVM). Index Terms:Embedded Surveillance System; Majority VotingMechanism; PIR Sensor; Ultrasonic Sensor I. INTRODUCTION Recently surveillance systems have become more importantfor everyoneâ&#x20AC;&#x2122;s security. The embedded surveillance system,frequently used in a home, an office or a factory [2-4], uses asensor triggered to turn on a camera [5-6]. Some designs usedifferent types of sensors to achieve reliability by means of thedifferent features of each sensor [7-8]. In this paper we extendour previous design not only by using both multiple PIR sensors and ultrasonic sensors as a sensor group, but also byusing the MVM. Ultrasonic receivers and transmitters arelocated at opposite ends [9-10]. However, to reduce theinterference from other frequencies in ultrasonic signals, weuse a coding signal to enhance the ability to distinguish therandom interference [11]. To enhance system reliability in theexperiment, we focus on how to improve the shortcomings ofthe ultrasonic sensor. Some research explores the influence ofattenuation in air and crosstalk of ultrasonic signals by using acoding signal [12], while some provides improvement ofthe ultrasonic signal
ISBN NO : 378 - 26 - 13840 - 9
by using different coding signal types. Other research uses the application of a coding signalto increase resolution and contrast of images. Yet anotherapproach builds a 3D image with an ultrasonic sensor in the PNcode that solves the problem with time delay. To enhancethe reliability of the ultrasonic sensors group, we proposeadding to the number of bits with coding to reduce theprobability of code breaking. II. SYSTEM ARCHITECTURE Fig. 1 shows the home embedded surveillance system which has two groups of sensors, indoor and outdoor. The outdoor sensor group contains a number of PIR and pressure sensors placed near windows and doors of a home. When the outdoor sensors sense an intruder, the MCU is woken up and turns on the power for the indoor PIR and ultrasonic sensors for the Majority Voting Mechanism. When this is completed, the decision signal passes to the embedded board GPIO (General purpose input and output). The software module of the power embedded board turns on the Web camera to capture images and user can view the images captured by the home surveillance system.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 84
Proceedings of International Conference on Developments in Engineering Research
///
www.iaetsd.in
span over several tens of degree width. Thus total configuration improves immunity to changes in background temperature, noise or humidity and causes a shorter settling time of the output after a body moved in or out the FOV. Along with Pyroelectric sensor, a chip named Micro Power PIR Motion Detector IC has been used. This chip takes the output of the sensor and does some minor processing on it to emit a digital output pulse from the analog sensor. Schematic of PIR sensor output waveform is shown in Fig. 2.
Fig.1.The home embedded surveillance system with ultra-low alert & GSM modem
Fig.2.PIR sensor output waveform
A. PIR Sensor
B. Ultrasonic Sensor
PIR is basically made of Pyroelectric sensors to develop an electric signal in response to a change in the incident thermal radiation. Every living body emits some low level radiations and the hotter the body, the more is emitted radiation. Commercial PIR sensors typically include two IR-sensitive elements with opposite polarization housed in a hermetically sealed metal with a window made of IR-transmissive material (typically coated silicon to protect the sensing element). When the sensor is idle, both slots detect the same amount of IR, the ambient amount radiated from the room or walls or outdoors. When a warm body like a human or an animal passes by, it first intercepts one half of the PIR sensor which causes a positive differential change between the two halves. When the warm body leaves the sensing area, the reverse happens, whereby the sensor generates a negative differential change. These change pulses are what is detected. In order to shape the FOV, i.e. Field Of View of the sensor, the detector is equipped with lenses in front of it. The lens used here is inexpensive and lightweight plastic materials with transmission characteristics suited for the desired wavelength range. To cover much larger area, detection lens is split up into multiple sections, each section of which is a Fresnel lens. Fresnel lens condenses light. Providing a larger range of IR to the sensor it can
Ultrasonic sensor is non-contact distance measurement module, which is also compatible with electronic brick. Itâ&#x20AC;&#x2122;s designed for easy modular project usage with industrial performance. A short ultrasonic pulse is transmitted at the time 0, reflected by an object. The senor receives this signal and converts it to an electric signal. The next pulse can be transmitted when the echo is faded away. This time period is called cycle period. The recommend cycle period should be no less than 50ms. If a 10Îźs width trigger pulse is sent to the signal pin, the Ultrasonic module will output eight 40 kHz ultrasonic signal and detect the echo back. The measured distance is proportional to the echo pulse width and can be calculated by the formula above. If no obstacle is detected, the output pin will give a 38ms high level signal.
ISBN NO : 378 - 26 - 13840 - 9
C. GSM Modem Global System for Mobile communications (GSM: originally from Group Special Mobile) is the most popular standard for mobile phone in the world. GSM/GPRS Smart Modem is a multi-functional, ready to use, rugged and versatile modem that can be embedded or plugged into any application. The Smart Modem can be customized to various applications by
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 85
Proceedings of International Conference on Developments in Engineering Research
using the standard AT commands. The modem is fully type-approved and can directly be integrated into your projects with any or all the features of Voice, Data, Fax, SMS, and Internet etc. III. WORKING CIRCUIT The total system can be divided into three segments
www.iaetsd.in
kept unchanged on Ql and inverted on Q2. Inverted output images the voltage level set by MCU keeping the alarm on even when MCU is at SLEEP mode. Alarm: The alarm has two pins- VCC and GNO. The power pin is connected to Ql output pin of the DLatch IC. The alarm can be set to ring by MCU. MCU: Pin RC4 of PIC 16f876A is connected to the 0 1 input of 74LS75.
A. Sensor and signal processing segment: C. GSM Modem interfacing segment: This segment consists of five parts: - PIR sensor module: The PIR sensor module is fed from the output of fixed output voltage regulator IC LM7805. PIR positive input terminal is fed with a +5V supply and negative terminal is grounded. PIR sensor module output pin is connected to MCU pin. For retriggering purpose, a jumper (JP) is attached on the COMMON (C) pin and HIGH (H) pin. - LM7805: LM7805 is a fixed output voltage regulator IC. 7805 takes + 12V input and gives a fixed regulated output voltage of +5V. - LM35: This is temperature sensor IC rated for full -55° to + 150°C temperature range. This is a transducer IC that takes voltage input and gives a voltage output proportional to the ambient temperature. +VS pin is connected to the output pin of LM7805 and the VOUT pin is connected to one of the analog input channels available on MCU. - Switch: This is a mechanical switch which is of NO (Normally Open) type. One end of the switch is connected to the +5V supply and the other end is connected to one of the MCU input pins. For practical use, electronic remote controlled switch is a better option to secure the system operation. - MCU: For this system, PIC 16f876A is used as the MCU, i.e. Microcontroller unit. It has built-in USART module which is necessary for passing AT commands to the GSM modem. The PIR sensor module output is tied to the pin RB I. The output of temperature sensor IC and one end of the mechanical switch is connected to pins RA3 and RBO respectively. A LED is connected to the pin RC3. The MCLRI VPP is connected to +5 V supply. A 4 MHz crystal is connected between OSCI an OSC2 pins. This crystal determines the clock speed of the MCU operation. B. Alarm segment: This segment consists of three parts: - 74LS75: This is a D-Latch IC. The input voltage level on D 1 is
ISBN NO : 378 - 26 - 13840 - 9
As GSM modem uses serial communication to interface with other peripherals, an interface is needed between MCU and GSM modem. This segment consists of four parts: -DB9 male connector: The serial port used here is a 9 pin DB9 male connector as the GSM modem side uses a female connector. Pin 14 and 13 of MAX232 are connected to pin 2 and 3 ofOB9 respectively. Pin 5 ofOB9 is grounded. – MAX232: This particular IC is necessary for increasing the voltage swing at the outputs. It takes OV and +5V inputs and makes it a + 12V and - 12V output voltages. This increased voltage swing is a requirement for serial communications. Two 1 μF capacitors are connected between pins 4, 5 and 1, 3 of MAX232. V+ and V- pins are fed from VCC and GNO, i.e. G round through two 1 μIF capacitors. Between VCC and GNO pins, one 10 μF capacitor is placed. - GSM modem: GSM modem is connected through a DB9 female connector to the interfacing circuit. - MCU: The VCC, i.e. power pin, TTL input and TTL output pins of MAX232 are connected to the pins RCO, RCI and RC2 of MCU respectively. 3.1CIRCUIT OPERATION A. Sensor and signal processing segment: As the jumper of PIR sensor module is placed between C and H, the output will stay on the entire time something is moving. The regulator IC serves regulated +5V to the LM35 and PIR sensor module. Prior to any operation, external interrupt is disabled in software of MCU. When the mechanical switch is closed, pin RBO gets an input voltage. This sets the system to run. The analog voltage output fromLM35 is taken and converted to an equivalent binary value which represents the ambient temperature. As PIR sensor module does not perform satisfactorily below 15°C temperature, MCU monitors the temperature
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 86
Proceedings of International Conference on Developments in Engineering Research
and light LED on pin RC3 when the temperature is equal to or greater than the critical temperature. After the LED is on, the MCU waits a pre-defined time for the place to be fully evacuated. After that time is over, the system is online. After activation of the system, if there is any movement on that place within the coverage region of the PIR sensor module, it outputs a pulse which is taken as input by MCU. MCU then waits a pre defined time and checks for that signal again. This is done for avoiding false triggering. B. Alarm segment: RC4 remains HIGH right from the beginning. Thus, the output pin IQ of 74LS75 stays LOW and the alarm does not ring. If the signal is still present during the second check, MCU makes pin RC4 LOW. This makes a HIGH on IQ of 74LS75 and the alarm rings. C. GSM Modem interfacing segment:
www.iaetsd.in
When the signal gets HIGH, MCU converts the analog signal from the temperature sensor to the binary equivalent and checks repeatedly if the temperature of the surrounding is greater or equal to 15째 Celsius. When the temperature rises to 15째 Celsius or more, MCU waits for a pre-defined time before executing any instruction. This wait state is introduced to ensure proper evacuation of the place where the system is to run. After the wait state is over, MCU starts checking for any signal from the PIR sensor module. When there is no signal from the sensor, MCU checks the status of the switch. If the switch is still closed, it continues to check for sensor signal.Wait state allows the call to be completed successfully. After that, MCU enables the external interrupt and goes to SLEEP mode. Enabling the external interrupt prior to SLEEP mode ensures that MCU will wake from the SLEEP mode whenever there is a HIGH to LOW transition on RBO, i.e. when the switch gets opened. When an external interruption occurs, MCU wakes up from sleep mood and disable the external interrupt and the program goes to the beginning of the algorithm.
MCU makes HIG H on RCO which in turn, activates MAX232 IC. Then MCU starts sending AT commands to the GSM modem through the pins RCI and RC2. The commands are sent through the interface to the modem. The modem receives the commands and sets up a call to a pre-defined number. The call is not disconnected until the call time - up or the recipient disconnects the call. After the call is disconnected, MCU goes to SLEEP, i.e. low power consuming mode. Before going into SLEEP, MCU enables the external interrupt in software. When the mechanical switch is open, an interrupt occurs and MCU is brought out of SLEEP mode. 3.2. SOFTWARE The whole system is built around a MCU. MCU requires to be burned with software written for specific applications. The code is written using ASSEMBLY language and compiled using MPLAB. MPLAB generated a hex file which is burned using a burner into the IC. This section demonstrates the flowchart of the software which helps to visualize the coding steps as shown in fig.3. At the beginning of the program, external interrupt of MCU is disabled in software. Therefore, any signal input on the pin RBO cannot generate interrupt. Then, MCU looks for the switch whether it is closed or open. When the switch is open the signal is LOW and when the switch is closed, the signal is HIGH on pin RBO. If the signal is LOW, MCU repeatedly checks for the switch status.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 87
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
sure the intruder passes through if the outside grouphas detected an individual. Fig.4 shows an ultrasonic signal being interfered with byanother ultrasonic frequency. As ultrasonic signal interferencecauses difficulties in setting up the amplitude of the referencevoltage of the circuit, therefore we use the coding signal toreduce ultrasonic signal interference based on thecharacteristics of the ultrasonic signal.
Fig.4. Relationship between reliability number of bits of the ultrasonic signal code
and
Fig.5. Ultrasonic signal interfered by another frequency Fig.3.Software flowchart IV. IMPLEMENTATION AND EXPERIMENT RESULTS
In Fig.6 we see that the signal has been coded by oursystem. The coded signal is not affected by another frequencybecause our design receives a signal through the code insteadof through its signalâ&#x20AC;&#x2122;s frequency.
In the experiment results we found that an ultrasonic signalwould be affected by environment sounds and the amplitude ofthe reference voltage. Those factors affect the transmissiondistance and the error rate of detecting. We therefore put thetransmitter and the receiver on both ends of the sensing areaand make
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 88
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig.6. Our ultrasonic coding signal in the scope Fig.7 shows the signal judgment of our experiment. Whenjudging a coding signal, one method of our design counts therising edge number. If the rising edge number is equal to two, itmeans the signal is correct. Lines A and B of Fig.7 show acoding interval.
Fig.8. Software flowchart surveillance system [12]
of
embedded
After the intruder has been detected outdoors, the MCUsoftware submodule of the ultrasonic coding signal is executedas shown in Fig.9. Our design transmits the ultrasoniccoding signal, and the receiver checks whether the codingsignal is correct or not. If the coding signal is correct, theultrasonic sensor group continues for some seconds to makesure that there is no intruder. If the coding signal is not correct,our design also starts the majority voting mechanism (MVM)to make sure whether there is any detection.
Fig.7. Judgment from received ultrasonic coding signal Fig.8 shows our design that consists of the internalsoftware module and the home embedded system softwaremodule. When an intruder has been detected, the MCU wakesup the majority decision to test the threshold and then turn onthe power supply for the indoor sensors. If the indoor sensorsdetect no intruder when the outdoor sensors are misjudging,the MCU turns off the power of the indoor sensors and goesback to the alert state. If the indoor sensors detect an intruder,the MCU turns on the Web camera to capture images inkeeping with the decision of the MVM.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 89
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig.10 shows the arrangement of our experimentalenvironment that detect intruders in a suitable place. We placethe PIR sensor on the ceiling or above the detection area.Transmitter and receiver of the ultrasonic sensor module areplaced in a line direction. When an intruder enters thedetection area, the ultrasonic coding signal will be blockedand the PIR sensors will detect temperature changes.
Fig.9. Flowchart of detection by ultrasonic coding signal In the experiment the ultrasonic signal of the receiver hasthe same direction as the transmitter, and we find that if theamplitude of the voltage waveform has been reduced toapproximately the same as the comparator's reference voltageour design can work normally. The scattering causes theamplitude of the voltage waveform to become gradually lowerthan the reference voltage. To increase the amplitude of thevoltage waveform we place a PET bottle at the front end forfocusing. Fig.9 shows the distribution of scattering afteradding a PET bottle focus. We have found that the ultrasonicsignal increases at the central point and achieves the focusingeffect.
Fig.11. Arrangement of our experimental environment Table I compares our coding signal and non-coding signal.The coding signal is not interfered with by other frequenciesunless their patterns are similar. It is easier to break anon-coding signal than a coding signal. When we add to thebits of the ultrasonic coding signal, the message type rise with2N A noncoding signal transmits just two types of messages,0 and 1. N means number of bits. With more message types,more codes can be used in the same design to decrease theprobability of breaking a signal. In our design, when N isequal to 8, the message combination of the ultrasonic codingsignal is 128 times better than that of non-coding signal, andthe reliability is enhanced from 0.5 to 0.996. TABLE I COMPARISON OF OUR CODING SIGNAL AND NONCODING SIGNAL
V. CONCLUSION
Fig.10. Curves showing distribution of scattering after adding PET bottle focus
ISBN NO : 378 - 26 - 13840 - 9
Our experiment shows two different types of sensors whichare enhancing the overall sensing probability by using theMVM to reduce the shortcomings of both the ultrasonicsensors and the PIR sensors. By adding an ultrasonic codingsignal our design reduces the
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 90
Proceedings of International Conference on Developments in Engineering Research
miss rate of the receiver withultrasonic sensors by different patterns, improving thereliability of the overall system. VI. REFERENCES [1] Ying-Wen Bai, Chen-Chien Cheng and Zi-Li Xie, “Use of Ultrasonic Signal Coding and PIR Sensors toEnhance the Sensing Reliability of an EmbeddedSurveillance System”, 2013 IEEE. [2] Jun Hou, Chengdong Wu, Zhongjia Yuan, Jiyuan Tan, Qiaoqiao Wangand Yun Zhou, “Research of Intelligent Home Security SurveillanceSystem Based on ZigBee,” International Symposium on IntelligentInformation Technology Application Workshops, Shanghai, 21-22 Dec. 2008, pp. 554-57. [3] Xiangjun Zhu, Shaodong Ying and Le Ling, “Multimedia sensornetworks design for smart home surveillance,” Control and DecisionConference, 2008, Chinese, 2-4 July 2008, pp. 431-435. [4] L. Lo Presti, M. La Cascia, “Real-Time Object Detection in EmbeddedVideo Surveillance Systems,” Ninth International Workshop on ImageAnalysis for Multimedia Interactive Services, 7-9 May 2008, pp.151-154.
www.iaetsd.in
[9] Hai-Wen Zhao, Hong Yue, and He-Gao Cai, “Design of a DistributedUltrasonic Detecting System Based on Multiprocessor for AutonomousMobile Robot,” Proceedings of tbe 2007 WSEAS Int. Conference onCircuits, Systems, Signal and Telecommunications, Gold Coast,Australia, January 17-19, 2007, pp. 59-64. [10] Zi-LI Xie and Zong-Han Li, “Design and implementation of a homeembedded surveillance system with ultra-low alert power,” IEEETransactions on Consumer Electronics, Feb. 2011, pp. 153-159. [11] Francesco Alonge, Marco Brancifortem and Francesco Motta, “A novelmethod of distance measurement based on pulse position modulationand synchronization of chaotic signals using ultrasonic radar systems,”IEEE Transactions on Instrumentation and Measurement, Feb.2009, pp.318-329. [12] ShragaShoval and Johann Borenstein, “Using Coded Sognals to Benefitfrom Ultrasonic Sensor Crosstalk in Mobile Robot Obstacle Avoidance,”IEEE International Conference on Robotics and Automation, Seoul,Korea, 21-26 May, 2001, vol.3, pp. 2879-2884.
[5] Wen-Tsuen Chen, Po-Yu Chen, Wei-Shun Lee and Chi-Fu Huang,“Design and Implementation of a Real Time Video Surveillance Systemwith Wireless Sensor Networks,” VTC Spring 2008. IEEE VehicularTechnology Conference, 11-14 May 2008, pp. 218-222. [6] MikkoNieminen, TomiRaty, and MikkoLindholm, “Multi-SensorLogical Decision Making in the Single Location Surveillance PointSystem,” Fourth International Conference on Systems, France, 1-6March 2009, pp. 86-90. [7] Ying-Wen Bai, Li-Sih Shen and Zong-Han Li, “Design andImplementation of an Embedded Surveillance System by Use ofMultiple Ultrasonic Sensors”, The 28th IEEE International Conferenceon Consumer Electronics, Las Vegas, Nevada, USA, 1113 Jan. 2010,11.1-3, pp. 501-502. [8] Yang Cao, Huijie Zhao, Na Li and Hong Wei “Land-CoverClassification by Airborne LIDAR Data Fused with Aerial OpticalImages,” International Workshop on Multi-Platform/Multi-SensorRemote Sensing and Mapping (M2RSM), Jan. 2011, pp. 1-6.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 91
Proceedings of International Conference on Developments in Engineering Research
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 92
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Modified Artificial Potential Fields Algorithm for Mobile Robot Path Planning Amrita.K, Keerthana.S.T and Vasanthkumar.B Department of Robotics and Automation, PSG College of Technology, Coimbatore, India. amritakrishnaraj@gmail.com keerthanaast@gmail.com
vasanthbe@gmail.com
Abstract— The problem of path planning is studied for the case of a mobile robot moving in an environment filled with obstacles whose shape and positions are not known. Path planning is dynamic when the path is continually recomputed as more information becomes available. A computational framework for dynamic path planning is proposed which has the ability to provide navigational directions during the computation of the plan. Path planning is performed using a potential field approach. This paper introduces a new algorithm, bidirectional artificial potential fields, capable of planning paths in unknown, partially known, and changing environments in an efficient, optimal, and complete manner. The algorithm uses the potential field values iteratively to find the optimum points in the workspace in order to form the path from start to destination. In the algorithm, motion planning is done continuously (dynamically), based on the system’s current position and on its feedback.
The abominable and potentially dangerous objects encountered by the robot in its route to destination are the obstacles and must be avoided. The capacity to move without collision in the uncertain environment taking into consideration the perception of the system is a fundamental problem to be solved in the autonomous mobile robot field. We study the problem of robust navigation for indoor mobile robots. Numerous algorithms and methods have been proposed for path planning of mobile robots. Artificial potential fields (APF) are the method most widely used due to its mathematical simplicity and ease of implementation and high efficiency. The Artificial potential field method was proposed by Khatib, which is a virtual force field method [10]. The basics of artificial field method are finding a function that represents the energy of the system and forces the robot to move towards the destination which posses the minimum energy value [12]. The robot is made to travel from highpotential to low-potential state. Moving the robot from source point to the destination in a ‘downhill approach’ is mathematically termed as the gradient Keywords— Mobile robot, Path Planning, descent (i.e.) Artificial Potential Field, Collision Free Path. x xk f ( xk ) (1) The motion terminates as the gradient vanishes. Although these methods are fast and efficient, they I. INTRODUCTION have the following drawbacks and limitations as Obstacle avoidance is one of the key issues to discussed in [7]: successful application of mobile robot systems. a) Trap situations due to local minima. Based on the configuration space and the b) No passage between closely spaced destination generating a path (finding a continuous obstacles. route) in a 2D environment with unknown obstacles c) Oscillations in the presence of obstacles. represent still a fundamental problem to be solved. d) Oscillations in narrow passages.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 93
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
e) Incapability to avoid dynamic obstacles. While trying to solve the above stated problems, researchers have introduced an additional potential so that the destination becomes the global minima. Still others have made changes to the path and tried to avoid the minimum points. A few researchers have tried to incorporate artificial intelligence methods into the potential methods. The aim of this paper is to use the artificial potential field method to detect and avoid the static as well as the dynamic obstacles. The navigation architecture is made of a global planner, local planner and a cost map. Localization of the robot is done with the visualization obtained from amcl. The proposed algorithm was implemented in the robot operating system (ROS), a LINUX based framework used for controlling robots. The reminder of the paper is arranged as follows: section1 discusses the algorithm for sensing and avoiding the static obstacles. Section2 examines the performance of this method through simulations. In section 3, new APF based method for dynamic path planning is explained. Finally section 4 discusses the conclusions and further work.
Where, (x, a) is the Euclidean distance between the locations of robot and the cell. (x, b) is the Euclidean distance between the destination of robot and the cell. (x, c). is the minimum distance between the affected areas of obstacle and the location of the robot. ƍ and Ɣ, the gain’s coefficients of attraction and repulsion functions respectively are positive constants. Using the expression (2) the potential for each cell in the workspace is determined. These determined potentials are then sorted in descending order. The cells with high potential are found to occupy positions close to the source point and destination point. Of the sorted potentials, the top 60% is considered to determine the threshold value, a notion to be continued for the rest of the paper. A threshold value is picked from the sorted list and all the cells with potential values greater than the threshold is studied. If the set threshold value is large, two distinct clusters of marked cells around start and destination points is obtained. As the threshold value is gradually decreased, these two clusters get bigger and bigger until they run into each other The threshold values are altered until a particular value is reached that guarantees that there is one II. STATIC OBSTACLE AVOIDANCE and only one cell that connects the source point Since the model under study is a mobile robot cluster to the destination point cluster. Values possessing freedom along the X and Y and rot smaller than this threshold value would make a (rotation about Z) motion, the gross workspace is a connection between two clusters, but this threshold 2D space. Each point of the workspace is value is the biggest value that by using all cells with considered and the potential at each point is potential value bigger than this value makes it determined. The notion of cell is introduced for possible for having a path between start and end further use in this paper which represents each pixel points. The point that connects these clusters is point in the camera transformation matrix. The called the midpoint. It is observed that, there is a workspace is considered to be a discrete space of N path from the start point to the midpoint, and there coordinate cells X=(x, y) and each cell is either is a path from midpoint to the destination within the range of the clusters. This is implementing by using obstacle or free space. the breadth first search algorithm. The potential is given as The next step is to find the midpoint between the otal(x)= source (x)+ destination (x)− tacle(x) [1] source and the current midpoint assuming that only (2) a few cells of the workspace are available. This process is iteratively repeated until a straight line The individual terms are expressed as path exists between consecutive points. (3) source (x) = ƍ/ (x, a) III. SIMULATION RESULTS (x) = ƍ/ (x, b) (4) destination (5) Two examples with two different environments are tacle (x) = Ɣ/ (x, c) studied. If a collision is detected, the robot due to
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 94
Proceedings of International Conference on Developments in Engineering Research
the prime repulsive force is deviated. Using this method, a smooth path with reasonable distance form obstacles is identified while keeping the path as short as possible. The algorithm has been simulated in the ros4mat an open source client/server library which supports the Kinect interface. Ros4mat allows a platform independent connection between the robot and multiple clients.
www.iaetsd.in
First step in the analysis of behaviour of dynamic objects is learning from the data obtained about their previous states. The most basic conceptualization of a moving object’s space–time behaviour is a geo-spatial lifeline [5] – also referred to as a movement path or trajectory, which describes a sequence of visited locations in space, at regular or irregular temporal intervals [9]. This process is known as spatiotemporal data mining a field still under persistent research and only limited researchers are working on the human – interface spatial data analysis. Several prediction techniques are used to predetermine the location of moving entities: neural networks ([2][12]), Markov models [6] Ehrenfest chains or Kalman filter ([8][3][4]). A.Markov Models Though the motion of obstacle is continuous, the imaging and other communication technologies support sampling only at discrete time. These sample datum obtained are analyzed to determine the behaviour of moving entities. A Markov Chain consists of a countable (possibly finite) set S (called the state space) together with a countable family of random variables X0, X1, X2, ... with values in S such that
P[Xl+1 = s | Xl = sl, Xl−1 = sl−1, · · · , X0 = s0] = P[Xl+1 = s | Xl = s
Fig. 1 Simulation Results of Static Obstacle Sensing IV. DYNAMIC
OBSTACLE AVOIDANCE
The proposed algorithm is based on the principles of APF path planning and our proposed ideas. These ideas are suggested to avoid collision with moving entities and to ensure a reliable on-line path planning where the robot employs continuous sensing and acting. The proposed algorithm is presented as a sequence of following steps.
A.Analysis of Obstacle Behaviour
This fundamental equation is referred to as the Markov property. Consider a autonomous mobile system that may be described at any time as being in one of the N countable states S0 S1… Sn. At regularly spaced discrete times, the system undergoes a change of state within the workspace. The time instants associated with the change of state is denoted as t = 0,1,... , and the actual state at time t is given as St [19]. X is a set of stochastic variables {X t, t ϵT}. The random variables X0, X1, X2, ... are dependent. Markov property expresses the assumption that the knowledge of the present (i.e., Xl = sl) is relevant to predictions about the future of the system, however additional information about the past (Xj = sj , j ≤ l − 1) is irrelevant. For the following consideration it is assumed that the chains are time-homogeneous: Aij =P (Xt+1 =i|Xt = j) = P(Xt = i|Xt-1 = j), t ϵT,
ISBN NO : 378 - 26 - 13840 - 9
(6)
i, j ϵS .
(7)
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 95
Proceedings of International Conference on Developments in Engineering Research
Unless stated to the contrary, all Markov chains considered in this paper are time homogeneous and we simply represent the matrix of transition probabilities as P = (Pij ). P is called the transition matrix. The non-homogeneous case is generally called time inhomogeneous or non-stationary in time. For a homogeneous Markov chain the transition probabilities can then be noted in a time independent stochastical matrix A: A=[aij ], aij ≥ 0 i, j ϵ S, Σ aij =1 (8)
www.iaetsd.in
or decreased with an allowable scaled factor as measure. This step depends on the velocities of the robot and the obstacles. An alternative proposed method utilizes the earlier calculated potentials. The map at the location of the expected collision point is retrieved from the global map and is further analyzed by applying the iterative midpoint method. Size of the retrieved map depends on the direction of motion and the dimensions of obstacles.
V. CONCLUSION Where A is called the transition matrix .It can be In this paper, modified artificial potential field shown that the probability of getting in m steps to algorithm is used to solve the path planning in an state j, starting from state i unknown and dynamic environment. The aij = P(Xt+m= j | X t =i) (9) simulation results show that the proposed algorithm can be computed as the m-th power of the transition is fast and efficient. In addition, it overcomes the matrix aij=A[i,j]. Recapitulating, a first-order time- drawbacks and limitations of traditional artificial homogeneous Markov Chain can be defined as a 3- potential field. The proposed algorithm has the tuple, consisting of the set of states S, the transition capabilities like escapes from dynamic obstacles. matrix A and the initial distribution vector The variations in the velocity of the obstacle might θ = (S, A,π ) (10) lead to changes in behaviour and result in collision. The embedded-renewal process is considered to As a further work the dynamic path planning in the determine the time at which the moving entity ROS platform and real time implementation is intrudes the path if the mobile robot. The time of under study. the nth visit to point y is denoted by . Ty,n=min{k∈N+:Ny,k=n} (11) REFERENCES The function is expressed as
Ny,n=∑i=1n1(Xi=y)
(12)
[1] H.Adeli, M.H.N.Tabrizi and A. Mazloomian, “Path
Planning for Mobile Robots using Iterative Where Ny,n is the number of visits ArtificialPotential Field Method”, IJCSI International The behaviour of the system thus analyzed from the Journal of Computer Science, vol. 4,no. 2, pp. 28-32 , July sampled data and is simultaneously expounded to 2011. obtain the accurate time and point of collision. The [2] S.Akoush and A.Sameh, “Mobile User Movement feedback data obtained from the odometers paired Prediction Using Bayesian Learning for Neural Network”, International conference on Wireless communications and with the mobile robot wheel helps in the mobile computing, 2007. localization of the robot A. Collision Avoidance
It is ensured that the mobile robot avoids the obstacles by passing them with safe potential distance. The analyzed behaviour determines the exact point of collision of the robot and the obstacle, henceforth providing a time frame in advance to take the necessary precautions. The velocity of the robot can be altered (i.e.) increased
ISBN NO : 378 - 26 - 13840 - 9
[3] J.M.François, G.Leduc and S.Martin, “Learning movement patterns in mobile networks: a generic approach', Proceedings of European Wireless, pp. 128-134, 2004. [4] J.Froehlich and J.Krumm, “Route Prediction from Trip Observations”, Society of Automotive Engineers (SAE) World Congress, 2008. [5] K.Hornsby and M.J.Egenhofer, “Modeling moving objects over multiple granularities”, Annals of Mathematics and Artificial Intelligence, vol. 36,pp. 177–194, 2002. [6] Y.Ishikawa,” Data Mining for Moving Object Databases, Mobile Computing and Computational Intelligence”, John Wiley & Sons , Chapter 12 (to appear), 2008.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 96
Proceedings of International Conference on Developments in Engineering Research
[7] Y.Koren and J.Borenstein, “Potential Field Methods and Their Inherent Limitations for Mobile Robot Navigation”, Proceedings of the IEEE Conference on Robotics and Automation, pp. 1398-1404, April 1991. [8] P.Kontkanen, P.Myllymäki, T.Roos, H.Tirri, K.Valtonen and H.Wettig, “Probabilistic Methods for Location Estimation in Wireless Networks”, In Emerging Location Aware Broadband Wireless Ad Hoc Networks, Springer US, pp. 173-188, 2005. [9] P.Laube, S.Imfeld and R.Weibel,”Discovering relative motion patterns in groups of moving point objects”, International Journal of Geographical Information Science, vol. 19,pp. 639–668,2005.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
[10]M.Mohamed and M.Abbas, “Obstacles Avoidance for Mobile Robot Using Enhanced Artificial Potential Field”, Al-Khwarizmi Engineering Journal, vol. 9, pp. 80-91,2012. [11]L.R.Rabiner, “A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition”, Proceedings of the IEEE, vol. 77,1989. [12] K.Uyanik, “A study on Artificial Potential Fields”, KOVAN Research Lab, 2004. [13] L.Vintan, A.Gellert, J.Petzold and T.Ungerer,”Person Movement Prediction Using Neural Networks”, Workshop on Modelling and Retrieval of Context, 2004.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 97
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Protecting Privacy Preserving For Cost Effective Adaptive Actions Using Memitic Algorithm Ulsa Suresh Babu M.Tech, Dept. of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India.
Abstract â&#x20AC;&#x201C; We present feasible algorithms to solve this difficult optimization problem, and point
SLO's of service level agreements #
SLO Name
an end-to-end system based on our prior work on the PREvent (prediction and prevention based on
Time between Receiving the 1 Time to Offer
event monitoring) framework, which noticeably
RFQ and responding with an offer(in working days)
indicates the usefulness of our model.
Order
Index term â&#x20AC;&#x201C; Memetic Algorithm, Local Search,
Description
2 Fulfillment Time
Branch and Bound.
Time between receiving the order and finishing the process (in working days) Time between initializing
I. INTRODUCTION I propose a generic Cloud computing stack that
3
Process Lead Time
days.
different layers. In this paper, we contribute to the 4
optimization problem, with the goal of minimizing the total costs for the service provider. We argue that this formulation better captures the real goals
(excluding activities at customer side) in working
classifies Cloud technologies and services into
state of the art by formalizing this tradeoff as an
the process and finishing it
5
Cost
Cost overrun with regard to
Compliance
the offer in % of the offer;
Product as
Product is exactly as
Specified
specified.
of service providers. Additionally, we present possible algorithms to solve this optimization
II. PROBLEM FORMULATION
problem efficiently enough to be applied at
In this section, we formalize the problem of
composition
these
selecting the most cost-effective adaptation actions
algorithms within our PREVENT (prediction and
to prevent one or more predicted SLA violations.
prevention based on event monitoring) framework.
We consider an interaction of the service
The below table show the SLO's of service level
composition with a given client, who has a given
agreements.
SLA with the composition provider. Let I be the
runtime.
We
evaluate
Table 1
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 98
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
set of all possible composition instances of this
improvements do not further reduce his costs (they
client, and let i ∈ I be concrete instances that we
are already 0 for this SLO).
can monitor. As part of the SLO definition, a penalty function is associated with all SLOs using the PREVENT tooling.
To prevent violations, we are able to apply a number of possible adaptations to an instance. Selecting the most cost-effective adaptation
Penalty functions for SLOs can take many
actions means finding the adaptation actions that
different shapes. The most important ones are:
minimize the total costs for the service provider. The total costs TC are defined in Equation 1 as the
1. constant penalty (a constant payment needs to be made if a certain SLO threshold value is
sum of the costs of SLA violations after adaptation (VC) and the costs of adaptation (AC).
surpassed), AC is the sum of the costs of all applied adaptation 2. staged penalty (similar to a constant penalty, but with different levels of penalty),
actions and VC is defined as the sum of all penalty functions applied to an instance. However, the
3. linear penalty (the penalty is linearly increasing
SLO Predictor provides estimations for SLOs
with the degree of violation), and
based on instance data. However, not all combinations of adaptation actions are legal. Some
4. linear penalty with cap (the penalty is linearly increasing up to a maximum value).
adaptation actions are mutually exclusive while others depend on each other.
Even though these functions span many different
III. ALGORITHMS
types of mathematical functions, they share two Now discuss different approaches for finding
essential characteristics.
solutions to this problem. These algorithms are i.
First, SLA penalty functions are always monotonically increasing.
point discontinuity in a special violation threshold point.
the
Cost-Based
Optimizer
predicted violation of at least one SLO, and receives as input a list of monitored facts and estimates of the current instance.
This also means that penalty functions are discontinuous.
in
component. Optimization is always triggered by a
ii. Second, SLO penalty functions always have a
generally
implemented
Furthermore,
1. Branch-and-Bound
this
property signifies that there is no incentive for the
Branch-and-bound is a very general deterministic
service provider to apply further adaptation and
algorithm for solving optimization problems. The
improve an SLO value below because all further
high-level idea of this approach is to enumerate the solution space in a “smart” way so that at least
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 99
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
some suboptimal solutions can be identified and
solution, the order may have an impact on the
discarded prematurely.
number of solutions we are able to prune. Therefore, we can conclude that it is beneficial to investigate actions in a specific order that maximizes the number of solutions that can be pruned. We specify two possible criteria for this ordering: 1) the impact of an action on the SLOs, and 2) the utility of an action. We will now define those two orderings. 2. Local Search While the Branch-and-Bound algorithm discussed above has the advantage of always finding the optimal set of actions for any composition instance, the execution time of the algorithm increases exponentially with the number of available actions. Even though we can reduce the runtime using impact- or utility-based sorting of
Fig. 1 General Branch-and-Bound Algorithm
actions, the complexity still remains exponential. Hence, there is an evident need to find strong
We describe our general Branch-and-Bound
heuristics.
algorithm in Fig. 1. The algorithm is easy to understand. What is the most important is the
A simple heuristic that is often used to very good
implementation of Line 13, the rules for pruning
ends is Local Search. Local Search is a
the search tree.
metaheuristic. The general idea is that in each iteration the algorithm searches a specified
In our Branch-and-Bound approach, we prune a partial solution in two cases: 1) if the partial solution already contains at least one conflict or 2) if the partial solution already prevents all SLA violations. In general, this approach is suboptimal. Even though the order in which we investigate actions has no impact on the quality of our
neighborhood for better solutions than the current one. If at least one such solution is found, the algorithm progresses to the next iteration with one of the better solutions. If no better solution can be found in the neighborhood, the algorithm has converged to a local optimum and is terminated. Usually, this algorithm is repeated multiple times with different starting solutions.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 100
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
This kind of algorithm typically depends on the
heuristic is based on similar concepts that we have
definition of: 1) a suitable neighborhood and 2) a
already used in our
senseful selection of starting solutions. We use the following neighborhood definition: A complete solution vector is in the neighborhood of an original solution if the two solutions represented as binary vectors have a Hamming distance of 1.
discussion of Branch-and-Bound: The idea is to stop adding actions if either no more SLOs are violated or no senseful actions are available anymore (the RCS is empty), and to prefer adding actions which have a high utility value. Hence, in every step, the RCS consists of the r (maximum size of the RCS) actions with highest nonnegative, which have not yet been added and which do not lead to a conflict. 3. Genetic Algorithm As an alternative to locality-based heuristics (local search, GRASP) we also present a solution based on the concept of evolutionary computation. More precisely, we use genetic algorithms (GAs) [23] as a more complex, but potentially also more
Fig. 2 GRASP Construction
powerful heuristic to generate good solutions to
For selecting the start solutions, we use two
the cost-based optimization problem. The overall
different approaches. The first and primitive one is
idea of GA is to mimic the processes of evolution
to select n start solutions with m bits set to 1 at
in biology, specifically natural selection of the
random. Alternatively, we propose to use an
fittest individuals, crossover, and mutation.
algorithm commonly referred to as GRASP (greedy randomized adaptive search procedure). GRASP is essentially a variation of local search, in which the start solutions are constructed using a greedy heuristic. The idea is that GRASP can converge to a better solution than a simple local
We use the term “fit” to describe solutions with a good (low) target function value. First, we generate a random start population. For this, we use the same primitive construction scheme stated in local search. Every following iteration of the algorithm essentially follows a three-step pattern.
search because the start solutions are already better than random start solutions. We have
First, we select a set of solutions from the
sketched the construction heuristic that we have
population to “survive” into the next generation. In
used in our implementation of GRASP. The
our genetic algorithm implementation, the fittest
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 101
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
solution is selected deterministically (elitism),
solution in the generation, basically reducing the
while all remaining slots in the next generation
population to a set of locally optimal solutions.
population are selected using a process called
Second, we remove the mutation operator from the
tournament selection. Second, crossover is used to
algorithm. The main reason is that given that all
produce new solutions based on the selected ones
solutions in the population are already locally
from the last generation. The main challenge of
optimal, randomly mutating one bit in a solution
implementing a strong crossover mechanism is to
can only lead to a worse solution.
ensure that the crossover product of two fit solutions is also likely to be fit. This simple procedure ensures that characteristics of both original solutions are preserved. Third, we use mutation to introduce entirely new features into the population. The need for mutation can be illustrated easily: Assume that a given action a is not applied in any solution in the population. After crossover, we may randomly flip every bit in every solution in the population with a very small probability. This means that most
Fig. 3 Memetic Algorithm
solutions in the population are not mutated, but
Furthermore, the main motivation for having
sometimes new actions are applied, which are not
mutation in the first place was that it is the only
the product of crossover. Unfortunately, this
way of introducing new actions in the canonical
â&#x20AC;&#x153;canonicalâ&#x20AC;? GA implementation takes a significant
GA. This is no longer the case, because local
amount of time to converge against a solution,
search can do the same thing. Generally, MA is
because the solution space is searched solely
slower than GA, because more solutions are
through the means of crossover and mutation.
evaluated in each generation. However, the
This leads us to an adapted algorithm, which we
algorithm potentially converges against a very
have sketched in Fig. 3. In such combinations of
good solution in a low number of generations.
GA and local search are often referred to as
Hence, we argue that in practice MA improves on
memetic algorithms (MA). The main changes of
the canonical form most of the time for our
MA (as compared to GA) are as follows. First, a
problem.
new Local Optimization operator is introduced after crossover. Local optimization applies the local search algorithm as discussed above to each
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 102
Proceedings of International Conference on Developments in Engineering Research
CONCLUSION
www.iaetsd.in
Management,” IBM Systems J., vol. 43, no. 1, pp. 136-158, Jan. 2004.
In this paper, we have modeled this problem as a one-dimensional discrete optimization problem. Furthermore,
we
have
presented
both,
deterministic and heuristic solution algorithms. We have evaluated these algorithms based on a manufacturing case study and have shown which types of algorithms are better suited for which scenarios. The main current limitation is that adaptation is only considered on instance level, which is, for each composition instance separately. We believe that the PREVENT adaptation model can be extended to this kind of SLOs and actions, but new approaches to predict violations and impact models are needed to this end. This need is taken as the future work.
[4] L. Bodenstaff, A. Wombacher, M. Reichert, and M.C. Jaeger, “Analyzing Impact Factors on Composite Services,” Proc. IEEE Int’l Conf. Services Computing (SCC ’09), pp. 218-226, 2009. [5] B. Wetzstein, P. Leitner, F. Rosenberg, S. Dustdar, Influential
F.
Leymann,
Factors
of
“Identifying
Business
Process
Performance Using Dependency Analysis,” Enterprise Information Systems, vol. 4, no. 3, pp. 1-8, July 2010. [6] P. Leitner, B. Wetzstein, F. Rosenberg, A. Michlmayr, S. Dustdar, and F. Leymann, “Runtime Agreement Services,”
REFERENCES
and
Prediction
of
Violations Proc.
Third
Service for
Level
Composite
Workshop
Non-
Functional Properties and SLA Management [1] M.P. Papazoglou, P. Traverso, S. Dustdar, and F. Leymann, “Service-Oriented Computing:
in Service-Oriented Computing (NFPSLAMSOC ’09), pp. 176-186, 2009.
State of the Art and Research Challenges,” AUTHORS
Computer, vol. 40, no. 11, pp. 38-45, Nov. 2007.
Ulsa Suresh Babu has received
[2] A. Lenk, M. Klems, J. Nimis, S. Tai, and T.
his B.Tech degree in Computer
Sandholm, “What’s Inside the Cloud? An
Science and Engineering from
Architectural Map of the Cloud Landscape,”
Brahmaiah
Proc.
ICSE
Workshop
Software
Eng.
Engineering,
Nellore
College affiliated
to
of JNTU,
Challenges of Cloud Computing (CLOUD
Anantapur in 2012 and pursuing M.Tech degree in
’09) pp. 23-31, 2009.
Computer Science & Engineering at Audisankara
[3] A. Dan, D. Davis, R. Kearney, A. Keller, R. King, D. Kuebler, H. Ludwig, M. Polan, M.
College of Engineering & Technology, Gudur affiliated to JNTU, Anantapur in (2012-2014).
Spreitzer, and A. Youssef, “Web Services on Demand:
ISBN NO : 378 - 26 - 13840 - 9
WSLA-Driven
Automated
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 103
Proceedings of International Conference on Developments in Engineering Research
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 104
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Use of Mobile Communication in Data-Intensive Wireless Networks Nuthanapati Jyothsna1
Reddy Sagar A C 2
Ravi Kumar G3
1
PG Student, Siddartha Institute of Science and Technology, Puttur chittoor(d) A.P, jyothsna_oct30@yahoo.com
2 3
Assistant Professor, Siddartha Institute of Science and Technology, Puttur chittoor(d) A.P, sagar.akkaru@gmail.com
Assistant Professor, Siddartha Institute of Science and Technology, Puttur chittoor(d) A.P, ravikumar.gunduru@gmail.com
Abstract: Wireless Networks are increasingly used in different types of data-intensive applications scenarios such as micro-weather monitoring, meticulousness agriculture, and audio/video observation. The sensor nodes are minute and limited in power. Sensor types vary according to the application of WN. Whatever be the application the key challenge is to broadcast all the data generated within an applicationâ&#x20AC;&#x2122;s life span to the base station in the face of the fact that sensor nodes have limited power supplies. The concept of mobile communication is that the mobile nodes change their locations so as to minimize the total energy consumed by both wireless transmission and locomotion. The predictable methods, however, do not take into account the energy level, and as a result they do not always extend the network lifetime. Keywords: Data-intensive; Energy; communication; routing tree; WN
networks are bi-directional, also enabling control of sensor activity. The development of wireless sensor networks was motivated by military applications such as battlefield surveillance; today such networks are used in many industrial and consumer applications, such as industrial process monitoring and control, machine health monitoring, and so on. Figure 1 shows an example of a wireless sensor network.
1. Introduction Sensors have the capabilities of doing sensing, data processing, and wirelessly transmitting collected data back to base stations by way of multiple-hop relay. Sensor itself supplies necessary operation with limited battery energy. Those operations that consume energy are transmitting and receiving data, running applications, measuring power, and even staying in standby mode. Among others, data transmission consumes the most energy. A wireless sensor network (WSN) consists of spatially distributed autonomous sensors to monitor physical or environmental conditions, such as temperature, sound , vibration , pressure humidity, motion or pollutants and to cooperatively pass their data through the network to a main location[1,2]. The more modern
ISBN NO : 378 - 26 - 13840 - 9
Figure 1: An example of Wireless Sensor Network
Recent advancement in mobile sensor platform technology has been taken into attention that mobile elements are utilized to improve the WSNâ&#x20AC;&#x2122;s performances such as coverage, connectivity, reliability and energy efficiency. The concept of mobile relay is that the mobile nodes change their locations so as to minimize the total energy consumed by both wireless transmission and locomotion. The conventional methods, however, do not take into account the energy level, and as a result they do not always prolong the network lifetime.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 105
Proceedings of International Conference on Developments in Engineering Research
2. Related Work Analyzing the three different approaches: Mobile base stations, data mules and mobile relays. All the three approaches use mobility to reduce energy consumption in wireless networks. A. Mobile Base Station A mobile base station is a sensor node collects the data by moving around the network from the nodes [4]. In some work, in order to balance the transmission load, all nodes are performing multiple hop transmissions to the base station. The goal is to rotate the nodes which are close to the base station. Before the nodes suffer buffer overflows, the base station computes the mobility path to collect data from the visited nodes. Several rendezvous based data collection algorithms are proposed, where the mobile base station only visits a selected set of nodes referred to as rendezvous points within a deadline and the rendezvous points buffer the data from sources. High data traffic towards the base station is always a threat to the networks life time [5]. The battery life of the base station gets depleted very quickly due to the sensor nodes which are located near to the base station relay data for large part of the network. The proposed solution includes the mobility of the base station such that nodes located near base station changes over time. All the above approaches incur high latency due to the low to moderate speed of mobile base stations. Figure 2 shows Mobile base station
www.iaetsd.in
B. Data Mules Data mules are another form of base stations. They gather data from the sensors and carry it to the sink. The data mule collects the data by visiting all the sources and then transmits it to the static base station through the network. In order to minimize the communication and mobility energy Consumption the mobility paths are determined. In paper [6] the author analyses an architecture based on mobility to address the energy efficient data collection problem in a sensor network. This approach utilizes the mobile nodes as forwarding agents. As a mobile node moves in close propinquity to sensors, data is transmitted to the mobile node for later dumps at the destination. In the MULE architecture sensors transmit data only over a short range that requires less transmission power. However, latency is increased because a sensor has to wait for a mule before its data can be Delivered. Figure 3. The three tiers of the MULE architecture. The Mule architecture has high latency and this limits its applicability to real time applications (although this can be mitigated by collapsing the MULE and access point tiers). The system requires sufficient mobility. For example, mules may not arrive at a sensor or after picking the data may not reach near an access-point to deliver it. Also, data may be lost because of radiocommunication errors or mules crashing. To improve data delivery, higher-level protocols need to be incorporated in the MULE architecture. Data mules also introduce large delays like base stations since sensors have to wait for a mule to pass by before initiating their transmission.
Figure 3: The three tiers of the MULE architecture Figure 2: Mobile base station
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 106
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
performs a single relocation unlike other approaches which require repeated relocations. Figure 4 shows Proposed Network.
C. Mobile Relay In this approach, the network consists of three nodes such as mobile relay nodes along with static base station and data sources. To reduce the transmission cost relay nodes do not transport data rather it will move to different locations. We use the mobile relay approach in this work. In [7] author showed that an iterative mobility algorithm where each relay node moves to the midpoint of its neighbors converges on the optimal solution for a single routing path This paper presents mobility control scheme for improving communication performance in WSN. The objectives of the paper are 1) Analyze when controlled mobility can improve fundamental networking performance metrics such as power efficiency and robustness of communications 2) Provide initial design for such networks. Mobile nodes move to midpoint of the neighbors only when movement is beneficial. Unlike mobile base stations and data mules, our approach reduces the energy consumption of both mobility and transmission. Our approach also relocates each mobile relay only once immediately after deployment. The paper study the energy optimization problem that accounts for energy costs associated with both communication and physical node movement. Unlike previous mobile relay schemes the proposed solution consider all possible locations as possible target locations for a mobile node instead of just the midpoint of its neighbors. 3. Proposed work We use low-cost disposable mobile relays to reduce the total energy consumption of dataintensive WSNs. Different from mobile base station or data mules, mobile relays do not transport data; instead, they move to different locations and then remain stationary to forward data along the paths from the sources to the base station. Thus, the communication delays can be significantly reduced compared with using mobile sinks or data mules. Moreover, each mobile node
ISBN NO : 378 - 26 - 13840 - 9
Figure4: Proposed Network
The network consists of mobile relay nodes along with static base station and data sources. Relay nodes do not transport data; instead, they move to different locations to decrease the transmission costs. We use the mobile relay approach in this work. Goldenberg et al. [13] showed that an iterative mobility algorithm where each relay node moves to the midpoint of its neighbors converges on the optimal solution for a single routing path. However, they do not account for the cost of moving the relay nodes. In mobile nodes decide to move only when moving is beneficial, but the only position considered is the midpoint of neighbors. The sink is the point of contact for users of the sensor network. Each time the sink receives a question from a user, it first translates the question into multiple queries and then disseminates the queries to the corresponding mobile relay, which process the queries based on their data and return the query results to the sink. The sink unifies the query results from multiple storage nodes into the final answer and sends it back to the user. The source nodes in our problem formulation serve as storage points which cache the data gathered by other nodes and periodically transmit to the sink, in response to user queries. Such a
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 107
Proceedings of International Conference on Developments in Engineering Research
network architecture is consistent with the design of storage centric sensor networks. Our problem formulation also considers the initial positions of nodes and the amount of data that needs to be transmitted from each storage node to the sink. We consider the sub problem of finding the optimal positions of relay nodes for a routing tree given that the topology is fixed. We assume the topology is a directed tree in which the leaves are sources and the root is the sink. We also assume that separate messages cannot be compressed or merged; that is, if two distinct messages of lengths m1 and m2 use the same link (si, sj ) on the path from a source to a sink, the total number of bits that must traverse link (si, sj ) is m1 + m2. a) Energy Optimization Framework In this section, we formulate the problem of Optimal Mobile Relay Configuration (OMRC) in data-intensive WSNs. Unlike mobile base stations and data mules, our OMRC problem considers the energy consumption of both mobility and transmission. The Optimal Mobile Relay Configuration (OMRC) problem is challenging because of the dependence of the solution on multiple factors such as the routing tree topology and the amount of data transferred through each link. For example, when transferring little data, the optimal configuration is to use only some relay nodes at their original positions. Assume the network consists of one source si −1, one mobile relay node si and one sink si +1. Let the original position of a node sj be oj = (pj , qj), and let uj = (xj , yj) its final position in configuration U. According to our energy models, the total transmission and movement energy cost incurred by the mobile relay node si is ci(U) = k║ui – oi ║ + am + b║ui+1 – ui ║2m Now We need to compute a position ui for si that minimizes Ci(U) assuming that ui−1 = oi−1 and ui+1 = oi+1; that is, node si’s neighbors remain at the same positions in the final configuration U. We calculate position ui = (xi, yi) for node si by finding the values for xi and yi where the partial derivatives of the cost function Ci(U) with respect to xi and yi become zero. Position Ui will be
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
toward the midpoint of positions ui−1 and ui+1. The partial derivatives at xi and yi, respectively are defined as follows: δ Ci(U) ---------- = −2bm(xi+1 − xi) + 2bm(xi − xi−1) δxi (xi − pi) + k -------------------------------------√ (xi - pi)2 + (yi - qi)2 δ Ci(U) ---------- = −2bm(yi +1 −yi) + 2bm(yi − yi−1) δ yi (yi − qi) + k -------------------------------------√ (xi - pi)2 + (yi - qi)2 and rms do not have to be defined. Do not use abbreviations in the title or heads unless they are unavoidable. b) Tree Optimization Algorithm In this section, we consider the sub problem of finding the optimal positions of relay nodes for a routing tree given that the topology is fixed. We assume the topology is a directed tree in which the leaves are sources and the root is the sink. We also assume that separate messages cannot be compressed or merged; that is, if two distinct messages of lengths m1 and m2 use the same link (si, sj ) on the path from a source to a sink, the total number of bits that must traverse link(si, sj ) is m1 + m2. Let the network consists of multiple sources, one relay node and one sink such that data is transmitted from each source to the relay node and then to the sink. We modify our solution as follows. Let si be the mobile relay node, S(si) the set of source nodes transmitting to si and sdi the sink collecting nodes from si. The cost incurred by si in this configuration U is: ci(U) = k║ui – oi ║ + ami + bmi║ui+1 – ui ║2 Algorithm 1 procedure OPTIMALPOSITIONS(U0) converged ← false; j ← 0; repeat anymove ← false;
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 108
Proceedings of International Conference on Developments in Engineering Research
j ← j + 1; ⊳ Start an even iteration followed by an odd iteration for idx = 2 to 3 do for i = idx to n by 2 do j (u i ,moved) ← LOCALPOS(oi, S(si), sdi ); anymove ← anymove OR moved end for end for converged ← NOT anymove until converged end procedure Our algorithm starts by an odd/even labeling step followed by a weighting step. To obtain consistent labels for nodes, we start the labeling process from the root using a breadth first traversal of the tree. The root gets labeled as even. Each of its children gets labeled as odd. Each subsequent child is then given the opposite label of its parent. We define mi, the weight of a node si, to be the sum of message lengths over all paths passing through si. This computation starts from the sources or leaves of our routing tree. Initially, we know mi = Mi for each source leaf node si. For each intermediate node si, we compute its weight as the sum of the weights of its children. Once each node gets a weight and a label, we start our iterative scheme. In odd iterations j, the algorithm j computes a position u i for each odd-labeled node si that minimizes Ci(Uj) assuming that uj i −1 =uj−1i −1 and uj i+1 = uj−1 i+1 that is, node si’s even numbered neighboring nodes remain in place in configuration Uj . In even-numbered iterations, the controller does the same for evenlabeled nodes. The algorithm behaves this way j because the optimization of u i requires a fixed location for the child nodes and the parent of si. By alternating between optimizing for odd and even labeled nodes, the algorithm guarantees that the node si is always making progress towards the optimal position ui. Our iterative algorithm is shown in algorithm1. 4. Conclusion The main objective of this paper is energy conservation which is holistic in that the total energy consumed by both mobility of relays and
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
wireless transmissions is minimized, which is in contrast to existing mobility approaches that only minimize the transmission energy consumption. The tradeoff in energy consumption between mobility and transmission is exploited by configuring the positions of mobile relays. We develop two algorithms that iteratively refine the configuration of mobile relays. The first improves the tree topology by adding new nodes. It is not guaranteed to find the optimal topology. The second improves the routing tree by relocating nodes without changing the tree topology. It converges to the optimal node positions for the given topology. Our algorithms have efficient distributed implementations that require only limited, localized synchronization. References [1] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E.Cayirci, “A Survey on Sensor Networks,” IEEE Comm.Magazine, vol. 40, no. 8, pp. 102114, Aug. 2002. [2] Shio Kumar Singh, M.P. Singh, and D.K. Singh,“Applications, Classifications, and Selections of Routing Protocols for Wireless Sensor Networks” International Journal of Advanced Engineering Sciences and Technologies (IJAEST), November 2010, vol. 1, issue no. 2, pp. 85-95. [3] Ganesan, B. Greenstein, D. Perelyubskiy, D. Estrin, and J.Heidemann, “An Evaluation of Multi-Resolution Storage for Sensor Networks,” Proc. First Int’l Conf. Embedded Networked Sensor Systems (SenSys), 2003. [4] Fatme El-Moukaddem, Eric Torng, Guoliang Xing" Mobile Relay Configuration in Dataintensive Wireless Sensor Networks" in IEEE Transactions on Mobile computing, 2013. [5] J. Luo and J.-P. Hubaux, Joint Mobility and Routing for Lifetime Elongation in Wireless Sensor Networks, in INFOCOM, 2005. [6] S. Jain, R. Shah, W. Brunette, G. Borriello, and S. Roy,Exploiting Mobility for energy Cient Data Collection in Wireless Sensor Networks, MONET,vol. 11, pp. 327339, 2006. [7] K. Goldenberg, J. Lin, and A. S. Morse, Towards Mobility as a Network Control Primitive, in MobiHoc, 2004, pp. 163174. [8] Tang and P. K. McKinley, Energy Optimization Under Informed Mobility, IEEE
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 109
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Trans. Parallel Distrib. Syst., vol. 17,pp. 947962, 2006. [9] Fatme El-Moukaddem, Eric Torng, and Guoliang Xing, Member, IEEE "Mobile Relay Configuration in Data- Intensive Wireless Sensor Networks".
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 110
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Confidential Multiparty Computation with Anonymous ID Assignment using Central Authority P Babitha1 Assist Professor SISTK
P Yogendra Prasad2 Assist Professor SISTK
ABSTRACT- For giving out private data securely between several parties an algorithm has been used. An anonymous ID assignment technique is used iteratively to assign the nodes with ID numbers ranging from 1 to N. This technique enhances data that are more complex to be shared securely. The nodes are assigned with the anonymous ID with the help of a central authority. The algorithm has been compared with the existing algorithm. In this paper we propose an algorithm has been developed based on Sturm’s theorem and Newton’s identities. The numbers of iterations are found out with the help of Markov chain. KEYWORDS: Anonymous Id, AIDA-Anonymous Id Assignment. 1. INTRODUCTION The anonymous message plays a very significant role in internet’s popularity for both individual and business purposes. Cloud based website management tools enable the servers to analyze the user’s behavior. The disadvantages of giving out private data are being studied in detail. Other applications for anonymity are various viz Doctor medical records, social networking, electronic voting and many more. In secure multiparty computation which is a new form of anonymity allows several multiple parties to share data that remains unknown? A secure computation function enables multi parties to compute the sum of their inputs rather than revealing the data. This method is very much
ISBN NO : 378 - 26 - 13840 - 9
A C Reddy Sagar3 Assist Professor SISTK
well-liked in data mining operations and enables classifying the complexities of secure multiparty computation. Our main algorithm is built on top of a method that shares simple data anonymously and yields a technique that enables sharing of complex data anonymously. With the help of permutation methods the assigned ID are known only to the nodes which are being assigned IDs. There are several applications where network nodes needs self-motivated unique IDs. One such application is grid computing where the services are requested without disclosing the identities of the service requestor. To differentiate between anonymous communication and anonymous ID assignment, think about a situation where N parties wish to exhibit their data in total, in N slots on a third party site, anonymous ID assignment method assigns N slots to the users whereas anonymous communication allows the users to conceal their identities In our network the identities of the parties are known but not the true identity. In this project we use an algorithm for sharing simple integer data which is based on secure sum. This algorithm is used in every iteration of anonymous ID assignment. Here we consider all the nodes to be half direct. Even though they follow a set of rules for message if they happen to see information they might intrude.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 111
Proceedings of International Conference on Developments in Engineering Research
In existing system, the information about each node will be shared along with the data. It is usually encrypted along with the data. The usage of Newton’s polynomial cannot be avoided as it increases the number of rounds of iterations that are used to compute the secure sum and power sum. Hence the performance of the system also decreases. It only focuses on the sum inputs whereas our project deals with the number of rounds. The usage of secure multiparty computation is being avoided with the usage of Sturm’s theorem to make sure that the information about the nodes are not revealed. In the current system the main goal is to provide anonymous id for each node. Each node will have a secure communication of simple and complex data. Those data’s may be from static data or dynamic data. By implementing secure sum hides permutations method and anonymous id assignment (AIDA) method the permutation methods are kept anonymous to each other. Hence here encoding technology is used to create anonymous ID and the ID is being assigned to the user by the central authority and the receiver receives the data and decodes it with the key that is known only to the sender and the receiver which might not be known to the other semi honest node that might intrude. II MODULES DESCRIPTIONS AUTHENTICATION The process of identifying an individual usually based on a username and password. In security systems, Authentication merely ensures that the individual is who he or she claims to be, but says nothing about the access rights of the individual. 1 LOGIN: In User and Admin login we are going to check whether the system is trusted machine or distrust machine. If the machine is trusted then the user or admin is allowed with n attempts. If
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
the machine is distrusted machine then the user is allowed with single attempt. Process Involved is to Check the login name and password Then allows the authorized user to use these pages. If the unauthorized user attempts to access user login then restrict that user and give the information. A FORGET PASSWORD: When the users forget their password then the user can access this forget password. It is used to create a new password. To ensure that user accessing forget password is a legitimate user, the user will be asked a question. These questions and their answers are created, while the user is registering to the site. If the user enters the answer then the entered text will be matched with the database. If the result is true, then the user will be allowed to enter the new password to access the site. If the result is false, user will not be allowed to enter the new password to access the site. B REGISTRATION: When a new user is creating an account he needs to register here by giving the sufficient information. Registration might also contain some private data that will be kept confidential so that the information about username and password is retrieved when it is forgotten. 2 ADMIN: In this module when the admin attempts to login we need to find whether the machine is trusted or it is distrusted machine. It is found by user id and pass word. Admin can provide AID to all nodes. With the help of that AID each node can share the data in internet. Admin can generate that AID for individual nodes. So the sharable data can be kept in a sharable database. As a result their own private data will be maintained secret. A GENERATES AID: In this module admin wants to create the AID for individual nodes. Admin can make this unique
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 112
Proceedings of International Conference on Developments in Engineering Research
AID for each user presented in network. With the help of this AID user can share his data and also he can keep his own private data as secret. B ASSIGN AID: Admin can provide unique AID to all nodes. Nodes presented in network will be communicating by using AID. AID can not contain any private information.AID helps to keep personal information as more secret. 3. USERS Users can login by entering the given username and password. Then he/she may go for corresponding page. User can keep his own information in a sharable data base. And also he can retrieve data shared by other users. User has to use his AID for sharing information. A SHARE DATA: New user has to get AID from admin. Admin will assign that AID for every node. So data shared by user can be kept in a sharable database and it can be shared by all users. Each node will have unique AID with the help of that unique AID any user can store and retrieve from sharable database. B RETRIEVE SHARED DATA: In this module user can retrieve the shared data. Shared data may be stored by him or by any other node. So it will be easy to make own private data as secret by implementing anonymous id algorithm. A SHARE DATA: New user has to get AID from admin. Admin will assign that AID for every node. So data shared by user can be kept in a sharable database and it can be shared by all users. Each node will have unique AID with the help of that unique AID any user can store and retrieve from sharable database. B RETRIEVE SHARED DATA:
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
In this module user can retrieve the shared data. Shared data may be stored by him or by any other node. So it will be easy to make own private data as secret by implementing anonymous id algorithm. Secure sum Algorithm Given nodes each holding an data item from a finitely represent able abelian group, share the value among the nodes without revealing the values . 1) Each node , chooses random values such that ri,1+…ri,N = di 2) Each “random” value is transmitted from node to node . The sum of all these random numbers is, ofcourse, the desired total . 3) Each node totals all the random values received as: si,j = r1,j+…+ rN,j 4) Now each node simply broadcasts to all other nodes so that each node can compute: T = s1 +…sN AIDA Algorithm Implementation: Given nodes n1,…,nN, use distributed computation (without central authority) to find an anonymous indexing permutation s : {1,…,N} {1,…,N}. 1) Set the number of assigned nodes A = 0. 2) Each unassigned node chooses a random number in the range 1 to S. A node assigned in a previous round chooses r1 = 0. 3) The random numbers are shared anonymously. Denote the shared values by q1,…,qN. 4) Let q1,…,qk denote a revised list of shared values with duplicated and zero values entirely removed where k is the number of unique random values. The nodes ni which drew unique random numbers then determine their index si from the position of their random number in the
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 113
Proceedings of International Conference on Developments in Engineering Research
revised list as it would appear after being sorted: si = A + Card{qj : qj <=ri} 5) Update the number of nodes assigned: . 6) If A < N then return to step (2). Basic formula P (collision) =1-e^(-N^2/(2*H))=alpha N=sqrt (2*ln (1/1-alpha))*sqrt (H) N: number of evenly distributed hashes to compare H: the Size of the element count of all possible hashes H=2^ (IDlength *ln (36)/ln (2)) Hashes calculated with SHAI (IDlength<=30). K.Communications Requirements of AIDA Methods Consider the required number of data bits for each of the three variant methods just described. This is the number of data bits that would be transmitted in each packet by the secure sum algorithm introduced earlier. The required numbers of data bits B are slightly overestimated by the formulae: = N. [ 2(P+1)] = N. [ 2(N)] /2.[ 2(S)] = S. [ 2(N+1)] (2) The computational requirements of the “slot selection” appear, at first, to be trivial. However, for every root that the “prime modulus” method must check, “slot selection”. L. The Completion Rate after R Rounds Two nodes might make identical choices of random numbers, or slots as they will be termed in this section. One can only guarantee that a complete assignment of nodes using possibilities for slots or random number choices and rounds will occur with at least a desired probability. The formulae are derived by assuming that N -1 node have chosen slots and
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
looking at the next choice. The Nth node into choose a slot resulting in assignments and conflicts. The slot it chooses could be unassigned, already in conflict with multiple occupants, or already assigned with exactly one occupant. III. EXPERIMENTAL RESULTS SI.NO N H P 1 189 45 390 2 210 55 400 3 290 95 442 4 549 225 667 5 999 455 1109 Table 1: Anonymous ID creation Algorithm Results are given in terms of identities. N represents the number of evenly distributed hashes to compare. H indicates the size of the element count of all possible hashes. P Indicates the value of anonymous ID. This is unique to every members of the group. Anonymous ID received is unknown to other members of the group. Anonymous id is examined between communication and computational requirement.
IV. CONCLUSION Proposed paper greatly decreases communication overhead. By using private communication channel that is anonymity router to transmit the data more securely. To overcome the problem of identifying details and changing information anonymous id was utilized. Random serial number is used to identify whether the data requesting person is a correct authorized person or hackers. The use of the Newton identities greatly decreases communication overhead. This can enable the use of a larger number of “slots” with a consequent reduction in the number of rounds required.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 114
Proceedings of International Conference on Developments in Engineering Research
The solution of a polynomial can be avoided at some expense by using Sturm’s theorem. The development of a result similar to the Sturm’s method over a finite field is an enticing possibility. With private communication channels, the algorithms are secure in an information theoretic sense. Apparently, this property is very fragile. The very similar problem of mental poker was shown to have no such solution with two players and three cards. The argument can easily be extended to, e.g., two sets each of N colluding players with a deck of 2N+1 cards rather than our deck of 2N cards. In contrast to bounds on completion time developed in previous works, our formulae give the expected completion time exactly. All of the no cryptographic algorithms have been extensively simulated, and the present work does offer a basis upon which implementations can be constructed. The communications requirements of the algorithms depend heavily on the underlying implementation of the chosen secure sum algorithm. In some cases, merging the two layers could result in reduced overhead. REFERENCES [1] Sarbanes–Oxley Act of 2002, Title 29, Code of Federal Regulations,Part 1980, 2003. [2] Y.Zhang, W.Liu, and W.Lou .Anonymous communications in mobile ad hoc networks. In INFOCOM 2005, 24th annual joint conference the IEEE Computer Societies. Proceeding IEEE, volume 3, pages 1940– 1951, March 2005.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
[3] D. Chaum, “Untraceable electronic mail, return address and digital pseudonyms,” Commun. ACM, vol. 24, no. 2, pp. 84–88, Feb.1981. [4] C. Clifton, M. Kantarcioglu, J. Vaidya, X. Lin, and M. Y. Zhu, “Tools for privacy preserving distributed data mining,” ACM SIGKDD Explorations Newsletter, vol. 4, no. 2, pp. 28–34, Dec. 2002. [5] J. Wang, T. Fukasama, S. Urabe, and T.Takata, “A collusion-resistant approach to privacy-preserving distributed data mining,” IEICE Trans.Inf.Syst. (Inst. Electron. Inf. Commun. Eng.), vol. E89-D, no. 11, pp.2739– 2747, 2006. [6] F. Baiardi, A. Falleni, R. Granchi, F. Martinelli, M. Petrocchi, and A.Vaccarelli, “Seas, a secure e-voting protocol: Design and implementation,” Comput. Security, vol. 24, no. 8, pp. 642–652, Nov. 2005. [7] Sanil, A. P., Karr, A. F., Lin, X., and Reiter, J. P. (2007). Privacy preserving analysis of vertically partitioned data using secure matrix products. [8] J. Kong and X. Hong. Anodr: anonymous on demand routing with untraceable routes for mobile ad-hoc networks. InMobihoc’03: Proceedings of the 4th ACM international symposium on Mobile ad hoc networking ages 291–302, New York, NY, USA,2003. [9] J. Yoon and H. Kim, “A new collision-free pseudonym scheme in mobile ad hoc networks,”5th workshop on Resource allocation, Cooperation and competition in wireless networks. June 2009.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 115
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
[10] A. Friedman, R. Wolff, and A. Schuster, “Providing k-anonymity in data mining,” VLDB Journal, vol. 17, no. 4, pp. 789–804, Jul. 2008. [11] J. Domingo-Ferrer, “A new privacy homomorphism and applications,” Information Processing Letters, vol. 60, no. 5, pp. 277–282, December1996.[Online]. Available: citeseer.nj.nec.com/290190.html [12] J. Domingo-Ferrer, “A provably secure additive and multiplicative privacy homo morphishm,” in Information Security, ser. Lecture Notes in Computer Science, A. Chan and V. Gligor, Eds., vol. 2433. Springer Verlag, 2002, pp. pp. 471–483. [13] D. M. Goldschlag,M. G. Reed, and P. F. Syverson, “Hiding routing information,” inProc. Information Hiding, 1996, pp. 137150,Springer Verlag [14] A. Karr, “Secure statistical analysis of distributed databases, emphasizing what we don’t know,” J. Privacy Confidentiality, vol. 1, no.2,pp. 197–211, 2009. [15] Q. Xie and U. Hengartner, “Privacypreserving matchmaking for mobile social networking secure against malicious users,” in Proc. 9thAnn. IEEE Conf. Privacy, Security and Trust, Jul. 2011, pp. 252–259.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 116
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
DESIGN OF RF BASED VOICE-CONTROLLED MULTI-TERRAIN ROBOT TO TRAVEL ON UNEVEN SURFACES C.Venkatesh
T.Venkata Subbaiah
Assistant Professor, Department of ECE, AITS, Rajampet, Andhra Pradesh, India E-mail: venky.cc@gmail.com
Department of ECE, AITS, Rajampet, Andhra Pradesh, India E-mail: t.v.subbaiah1994@gmail.com
G.Sarath Kumar Department of ECE, AITS, Rajampet, Andhra Pradesh, India E-mail: sarath.g66a@gmail.com
Abstractâ&#x20AC;&#x201D; Ambulation in rough and unstructured terrain is naturally very difficult for small animals or robots. The ability to climb is found widely among small animals .Mobile robots are more and more leaving the protected lab environment and entering the unstructured and complex outside world, e.g. for applications such as environmental monitoring. A substantial portion of the Earth is inaccessible to any sort of wheeled mechanism natural obstacles like large rocks, loose soil, deep ravines, and steep slopes conspire to render rolling locomotion ineffective. Hills, mountains, shores, seabeds, as well as the moon and other planets present similar terrain challenges. In many of these natural terrains, the terrain robot is wellsuited to travel in uneven surfaces In robotics, various climbing systems have been presented so far using different approaches for climbing but only very few are capable of up righting themselves after landing. Inspired by the terrain climbing capability, a small terrain robot has been developed. In this paper the working and design of terrain robot is presented. This robot utilizes four wheels to drive over just about any terrain. It works on any indoor surface and outdoor surfaces. This is an intelligent robot that can easily move on uneven surfaces too. This robot uses RF technology controlled by RF remote. This can be moved forward and reverse direction using geared motors of 60RPM. Also this robot can take sharp turnings towards left and right directions. Keywordsâ&#x20AC;&#x201D; Ambulation, Seabeds, Terrain, Locomotion. I.
II.
DESIGN STUDY
The block diagram of the robotic system consists of Transmitter and Receiver sections. The Transmitter section consists of PC with commands, RF Module and the Receiver section consists of RF module, ARM processor (LPC2148), H- Bridge and DC Motor. A. Transmitter: The Transmitter section having the four switches placed around the neck and RF Module. Initially the switches are at logic1. When the switch is pressed the concerned switch level goes to logic0.The switches are the inputs to RF transmitter through the RF encoder is shown in fig.1
INTRODUCTION
In Recent technology advances mechanical, electronics, communication, and networking make it possible to build small robots and organize many of them to form a mobile sensor network. This mobile sensor network has various applications such as search and rescue, surveillance, and environmental monitoring. The small terrain robot are developed for such applications. Compared with the traditional wheeled sensors. This ability makes jumping an ideal locomotion method in the areas of rugged terrain and natural obstacles. To use jumping as a locomotion method, the
ISBN NO : 378 - 26 - 13840 - 9
robot must be able to jump continuously; therefore, the following two capabilities are needed: (1)The robot can jump with a certain takeoff angle; (2) After landing on the ground, the robot can self-right from any possible landing posture. Besides these two capabilities, it would be beneficial that the robot can also change the jumping direction or the takeoff angle.In nature, millimeter-sized insects often address both obstacles and efficiency through jumping. While jumping is often seen as an energetically costly escape mechanism, has shown that as insect size shrinks, jumping becomes more advantageous due to the higher takeoff velocities allowed. Because small jumpers are more mechanically efficient than their larger counterparts, they require less muscle tissue to make them more energy efficient as well. This paper presents the design and working of terrain robot with experimental results
Fig. 1.
Transmitter Section
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 117
Proceedings of International Conference on Developments in Engineering Research
B. Receiver The received signal from the transmitter is fed to the RF decoder (Serial input and parallel output) .The output of the decoder is given to H-Bridge through the ARM processer shown in fig2.The output of H-Bridge drives the DC motors.
www.iaetsd.in
employs a SAW-stabilized oscillator, ensuring accurate frequency control for best range performance. The manufacturing friendly SIP style package and lowâ&#x20AC;&#x201C; cost make the STT-433 suitable for high volume applications.
Fig. 3.
Fig. 2.
III.
Receiver section
RF TECHNOLOGY
Radio frequency [RF] has a frequency range about 3Hz to 300GHz. This range corresponds to frequency of alternating current electrical signals used to produce and detect radio waves. Since most of this range is beyond the vibration rate that most mechanical systems can respond to RF usually refers to oscillations in electrical circuits or electromagnetic radiation. When an RF current is supplied to an antenna, it gives rise to an electromagnetic field that propagates through space. Electrical currents that oscillate at RF have special properties not shared by direct current signals. One such property is the ease with which it can ionize air creates a conductive path through air. Another property is the ability to appear to flow through paths that contain insulating material, like the dielectric insulator of a capacitor. The degree of effect of this property depends on the frequency of the signals. RF is a radio frequency technology which uses frequencies in the range of 3MHZ to 300 MHZ in general. Here in this RF system, we are using the frequency of 433MHZ which is in the Frequency range. The distance of this radio frequency range is up to 100m in general. In this project, the distance is up to 100m. A. RF TRANSMITTER: The STT--433 is ideal for remote control applications as shown in Fig.1 where low cost and longer range is required. The transmitter operates from a 1.5-12v supply, making it ideal for battery powered applications. The transmitter
ISBN NO : 378 - 26 - 13840 - 9
RF Transmitter STT-433MHZ
Features: 433.92Hz Frequency, Low cost, 1.5-12V operation, small size. PIN OUT: GND: Transmitter ground. Connect to ground plane. DATE: Digital date input. This input is CMOS compatible and should be driven with CMOS level inputs. VCC: Operating voltage for the transmitter.VCC should be bypassed with a .01uF ceramic capacitor and filtered with a 4.7uF tantalum capacitor . Noise on the power supply will degrade transmitter noise performance. ANT: 50ohm antenna output. The antenna port impedance affects output power and harmonic emissions. Antenna can be single core wire of approximately 17cm length or PCB trace antenna. B. RF RECEIVER: The data is received by the RF receiver from the antenna pin and this data is available on the data pins. Two data pins are provided in the receiver module. Thus data can be used for further application. This output is capable of driving one TTL or CMOS load .It is a CMOS compatible output. The data transmitted in to the air is received by the receiver. The received data is taken from the data line of the receiver and is fed to the decoder.
Fig. 4.
RF Receiver STT-433MHZ.
Features: * Low current (max.100ma) * Low voltage (max.65v).
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 118
Proceedings of International Conference on Developments in Engineering Research
C. ENCODER AND DECODER: The encoder and decoder used here is HT12E and HT12D respectively from HOLTEK SEMICONDUCTORS INC. The HT12E encoder ICs are series of CMOS LSIs for remote control system applications. They are capable of encoding 12 bit of information which consists of N address bits and 12-N data bits.
www.iaetsd.in
navigated. A wheel requires a relatively flat surface on which to operate. Rocky or hilly terrain, which might be found in many applications as forestry, waste clean-up and planetary exploration, imposes high demands on a robot and precludes the use of wheels. A second approach to this problem would be to use tracked wheel robots. For many applications this is acceptable, especially in very controlled environments. However, in other instances the environment cannot be controlled or predicted and a robot must be able to adapt to its surroundings. V.
Fig. 5.
RF Encoder Connections
Encoder IC (HT12E) receives parallel data in the form of address bits and control bits. The control signals from remote switches along with 8 address bits constitute a set of 12 parallel signals. The encoder HT12E encodes these parallel signals into serial bits. Transmission is enabled by providing ground to pin14 which is active low. The control signals are given at pins 10-13 of HT12E.
Fig. 6.
RF reciver connections
Features: * Operating voltage 2.4V~12V. * Low power and high noise immunity CMOS Technology. * Low standby current. IV.
JUMPING MECHANISM:
Traditionally, most mobile robots have been equipped with wheels. The wheel is easy to control and direct. It provides a stable base on which a robot can maneuver and is easy to build. One of the major drawbacks of the wheel, however, is the limitation it imposes on the terrain that can be successfully
ISBN NO : 378 - 26 - 13840 - 9
TERRAIN ROBOT HARDWARE TOOLS
Robotic colonizer main board is the brain of mechanical robot which receives the commands from PC through ZigBee wireless connection and processes these commands to perform pattern motion control. The main part in the robot main board is ATmega168 microcontroller which generates two PWM signals for each wheel of DC motor. The two active wheels of the robot are actuated by two independent servo motors modified for continuous rotation. In particular, The robot is powered by 12 V battery. The hardware tools of the robotic system are: ZigBee, ATmega 168 Microcontroller, L293D Driver Circuit, DC Motors, PH Meter, Metal Detector, GPS The hardware components of the robotic system is as shown in Fig. 4. A) ARM Processor ARM7TDMI is a core processor module embedded in many ARM 7 microprocessors including LPC2148. The ARM7TDMI core is a 32-bit embedded RISC processor. Its simplicity results in a high instruction throughput and impressive real-time interrupt response from a small and costeffective processor core. Founded in November 1990, it is spun out of a corn computers, it designs the ARM range of RISC processor cores. Licenses ARM core designs to semiconductor partners who fabricate and sell to their customers. ARM dose not fabricate and sell to their customers it also develop technologies to assist with the design in of the ARM architecture. Software tools, boards, debug hardware application software, bus architectures, peripherals etc. The ARM processor core originates within a British computer company called acorn. Instruction set. Essentially, the ARM 7TDMI-S processor has two instruction sets: *The standard 32 â&#x20AC;&#x201C;bit ARM set. *A16-bit thumb set. ARM stands for advanced RISC machines. It is a 32 bit processor core used for high end applications. The LPC 2148 micro controller are based on a 16â&#x20AC;&#x201C; bit/32-bit ARM7 TDMI-S CPU with real time emulation and embedded trace support that combined the micro controller with embedded high speed flash memory running from 32 KB to 512 KB. ARM (Advance RISC machine) T- The thumb 16 bit instruction set. D-ON chip debug support. M- Embedded multiplier. I- Embedded ICE hard ware. S-Synthesizable.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 119
Proceedings of International Conference on Developments in Engineering Research
The ARM has seven basic operating modes User: unprivileged mode under which most tasks run FIQ: entered when a high priority (fast) interrupt is raised. IRQ: entered when a low priority (normal) interrupt is raised. Supervisor: entered on reset and when a software interrupt instruction is executed. Abort: used to handle memory access violations. Undef: used to handle undefined instructions. System: privileged mode using the same registers as user mode. The ARM architecture provides a total of 37 register , all of which are 32 –bits long . however these are arranged in to several banks, with the accessible bank being governed by the current processor mode. In each mode, the core can access; A particular set of 13 general purpose registers (r0 –r12),A particular r13 -which is typically used as a stack pointer, r14 is used as a link register for branching. B) LPC2148 CONTROLLER: The LPC2141/42/44/46/48 microcontrollers are based on a 16-bit/32-bit ARM 7TDMI-CPU With real–time emulation and embedded trace support, that combine microcontroller and embedded high speed flash memory ranging from 32KB to 512KB. Serial communications interfaces ranging from a USB 2.0 full–speed device, Multiple UART’s, SPI,SSP to 12C- Bus and on-chip SRAM of 8KB up to 40KB, make these devices very well suited for communication gate ways and protocol converters, soft modems, voice recognition and low end imaging, providing both large buffer size and high processing power. Various 32 –bit timers, or dual 10- bit ADC, 10-bit DAC, PWM channels and 45fast GPIO lines with up to nine edge or level sensitive external interrupt pins make these microcontrollers suitable for industrial control and medical systems. C) L293D Driver Circuit: Motor driver is basically a current amplifier which takes a low-current signal from the microcontroller and gives out a proportionally higher current signal which can control and drive a motor. In most cases, a transistor can act as a switch and perform this task which drives the motor in a single direction[4]. D) DC Motors: DC motors are configured in many types and sizes, including brush less, servo, and gear motor types. A motor consists of a rotor and a permanent magnetic field stator. The magnetic field is maintained using either permanent magnets or electromagnetic windings. DC motors are most commonly used in variable speed and torque. Motion and controls cover a wide range of components that in some way are used to generate and/or control motion. E) Ultrasonic Sensor Motors: An ultrasonic transducer is a device that converts energy into ultrasound, or sound waves above the normal range of human hearing. While technically a dog whistle is an ultrasonic transducer that converts mechanical energy in the
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
form of air pressure into ultrasonic sound waves, the term is more apt to be used to refer to piezoelectric transducers or capacitive transducers that convert electrical energy into sound. Piezoelectric crystals have the property of changing size when a voltage is applied; applying an alternating current (AC) across them causes them to oscillate at very high frequencies, thus producing very high frequency sound waves. The location at which a transducer focuses the sound can be determined by the active transducer area and shape, the ultrasound frequency, and the sound velocity of the propagation medium. Some systems use separate transmitter and receiver components while others combine both in a single piezoelectric transceiver.Nonpiezoelectric principles are also used in construction of ultrasound transmitters. Magnetostrictive materials slightly change size when exposed to a magnetic field; such materials can be used to make transducers. A capacitor microphone uses a thin plate which moves in response to ultrasound waves; changes in the electric field around the plate convert sound signals to electric currents, which can be amplified. VI.
EXPERIMENTAL RESULTS
The autonomous manoeuvre sailing robot for oceanographic research is used to explore all the details on the surface of the water. This robot is used for locating the position of the system using GPS & GSM, detects metals present in the ocean, and measures the depth and boundaries, used for surveillance and rescue operation.
Fig. 7. Robot is moving aty sandy surface
Fig. 8. Robot is moving in sandy and rocky surface
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 120
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
[11] L. Fortuna, M. Frasca, M.G. Xibilia, A. A. A , M. T. R s , “M p pu A m s”, P gs T 1s International Conference on Energy, Power, and Control, College of Engineering, The University of Basrah, Basrah, Iraq , pp.1215, Nov. 30 to Dec. 02, 2010. [12] I. Hayas , N. I suk, “M m g s”, I P . O I . Symposium on Micromechatronics and Human Science, Nagoya, 1998, pp.41-50. [13] T. H. L , F.H. F. L u g, P.K.S. T m, “P s m s us g uzz g ,” IEEE international Conference on Industrial Electronics Socielty, vol. 3, no. 2, pp.525-528, 1999. [14] Embedded systems 2nd edition - RajKamal [15] Programming AVR microcontroller - Dhananjay v.Gadre [16] Robotics reference guide - Joseph A. Angelo [17] Data networks and interfacing - Hawkins [18] Texas - linear IC’s manual [19] Segnetics Digital IC’s manual ZigBee Specification
Author Profile Fig. 9. Robot is moving in rocky surface
VII. CONCLUSION In this paper, we introduce a successful working prototype model of terrain robot is designed to travel in uneven surfaces. An autonomous terrain robot offers major advantages compared to other vehicles. In this paper we also discuss the development process of a robotic system for environmental monitoring in search &rescue and demining applications. Further development is required to demonstrate the feasibility of a terrain robot for long term use in open area and helpful for the scientists.
C.Venkatesh received B.Tech Degree from J.N.T.University, Hyderabad and M.Tech Degree from J.N.T.U.A.,Anantapuramu. Presently he is with Annamacharya institute of Technology & Sciences, Rajampet, Andhra Pradesh, India, working as an Assistant Professor in Department of ECE. His research interests include Embedded Systems, Signal Processing and Digital Imaging. He presented many research papers in National & International Conferences. He is a member of professional societies like MISTE,MIE,IACSIT(Singapore), IAENG(Hong Kong), UACEE(India), ISOC(Switzerland), APCBEES(China) and SIE (Singapore).He acted as reviewer for many International conferences and journals
REFERENCES [1]
Alina Conduraru , Ionel Conduraru, Emanuel Puscalau, Geert De Cubber, Daniela Doroftei, Haris Balta” Development of an autonomous rough-terrain robot” [2] G. De Cubber and D. Doroftei and H. Sahli and Y. Baudoin, Outdoor Terrain Traversability Analysis for Robot Navigation using a Time-Of Flight Camera, RGB-D Workshop on 3D Perception in Robotics, 2011 [3] D. Doroftei and G. De Cubber and K. Chintanami, Towards collaborative human and robotic rescue workers, Human Friendly Robotics, 2012. [4] K. Richardson, Rescue robots - where were they in the Japanese quake relief efforts?, Engineering and Technology Magazine, vol.6, nr. 4, 2011 [5] M.P. Khorgade “Application of MEMS in Robotics and Bio MEMS”, Proceedings of the UK sim 13th international conference. [6] Mobile Robotic “ Navigation and control for large-scale wireless sensor network repair”, by Kyle lathy in north Carolina state university on may6, 2009 [7] S. Mehta, ET. “Al. CMOS Dual -Band Tri-Mode chipset for IEEE 802.11a/b/g wireless LAN”, IEEE RF IC Symposium, pp 427430,2003 [8] B. Razavi,” RF Transmitter Architectures and circuits”, IEEE CICC, pp.197-204, 1999. [9] M.P. Khorgade “Application of MEMS in Robotics and Bio MEMS”, Proceedings of the UK sim 13th international conference. [10] Mobile Robotic “ Navigation and control for large-scale wireless sensor network repair”, by Kyle lathy in north Carolina state university on may6, 2009
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 121
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Zigbee For Vehicular Communication Systems J.Ram Harish Yadav1, K.Dhanunjaya2 1
2
PG Student, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India. Head of the Deepartment, Department of Electronics & Communication Engineering, ASCET, Gudur, A.P, India 1
ramharish.j@gmail.com
ď&#x20AC; Abstract:Vehicular communication is a popular
topic in theacademia and the car industry. The aim of this growing interestis to develop an effective communication system for theIntelligent Transportation System (ITS). In this paper wepresented the model of wireless base station goodput evaluation.We used wireless access point model as a queuing system withvariable requests and the auto traffic model. The performance ofthe wireless networks can be impacted from a variety ofparameters, such as radio communication range, availablebandwidth and bit rate, the number of clients in wireless networkrange and vehicle speed. The basic parameters were analyzedand presented in this paper. Index Terms:Short Range Vehicle Network; 802.11n; wirelessnetwork; goodput; network performance; transport; mobilestations; auto traffic; vehicle speed; Markov chain. I. INTRODUCTION The needs to enhance road safety, traffic efficiency and to reduce environmental impact of road transport are seriouschange for both academics and industry. Researchers aregreatly interested to develop vehicular communication andnetworking technology in two realistic ways vehicle tovehicle (V2V) in ad hoc mode and vehicle to infrastructure(V2I) with fixed nodes along the road. The potency toexchange information wireless via V2X is a foundationstone for building powerful Intelligent Transport Systems(ITS). In Europe, USA and Japan are great efforts madefrom automakers and governments to reach single standardsthrough the several and common projects such as CAR 2CAR Communication Consortium, Vehicle Safety Communication Consortium, and EUCAR SGA etc. Result fromcommon effort is an international standard, IEEE802.11p[2], also known as Wireless Access for VehicularEnvironments (WAVE). This standard will be used as thegroundwork for Dedicated Short Range Communications(DSRC).
ISBN NO : 378 - 26 - 13840 - 9
This type of communication has potential toimprove safety on the road, traffic flow and provide comfortfor passengers and drivers with expedited applications suchas INTERNET, network games, automatic electronic tollcollection, drive-through payments, digital map update,wireless diagnostic and flashing etc. DSRC is the one stepin the future, because it lets inter-vehicle and vehicle toinfrastructure wireless communication. Wireless networking based on IEEE802.11 technology[3] has recently become popular and broadly available atlow-cost for home networking and free Wi-Fi orcommercial hotspots. The DSRC starting idea was to equipvehicular network nodes with off-the-shelf wirelesstechnology such as IEEE802.11a. This technology is costeffective and has potential to grow and new versions havebeen recently produced. The latest standard of wirelesslocal area network (WLAN) is IEEE802.11n [4]. The IEEE802.11n standard promises to improve and extend mostpopular WLAN standards by significantly increasingthroughput, reliability and reach. Nowadays dispositions of WLAN-based accesstechnology are predominantly to stationer indoor andoutdoor users who are most slowly moving and in rangelimited. Despite the fact that the standard has not beendeveloped for fast dynamic usage, nothing limits it to beevaluated for vehicular communication systems. Themotivation is to understand the interaction between thevehicle speed and goodput of WLAN-based network.Realizing field trials for goodput evaluation ofvehicular wireless communication systems is very difficultand costly because many vehicles and communicationequipments need to be purchased or rented, and also manyexperimenters need to be employed. Given such problems,it is highly desirable to obtain a mathematical descriptionof process with real data from small scale scenarios ofpractical measurement results and performance evaluationsprior to conducting field trials as it is made in this work.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 122
Proceedings of International Conference on Developments in Engineering Research
This paper constructs as follows: After introducing theproblem in Section 1, Section 2 provides some Vehicular Communication Systems. Then, in Section 3 SeVeCom Implementation and demonstrating the analysis results inSection 4. Section 5 summarize and concludes this paperwith a brief description on future works. II. VEHICULAR COMMUNICATION SYSTEMS There are significant differences between devices such asmobile phones or desktop computers connected to the Internetand devices in a VC system. Differences in development,production, and operation, determine VC-specific constraintsand conditions: 1) Vehicles have a long life span, lasting several yearsin most cases. This makes it hard to change onboardsystems as reaction to new upcoming risks to the vehiclesafety. 2) Owners have constant physical access to and full controlover vehicles. In spite of the involved safety risks,many users might try to modify or “enhance” theirvehicles. From a manufacturer’s point of view, the riskof hardware tampering cannot be neglected. 3) No technical expertise on vehicle electronics or VCsecurity aspects is expected from a user that runs avehicle. Hence, the vehicular security measures have tooperate autonomously with no need for intervention orfeedback from the user.
www.iaetsd.in
adding, replace, and reconfigurecomponents (for example, substitute cryptographic algorithms)throughout the life cycle of the vehicle. The large number and the variety of vehicles have tobe taken into account. Even for a single car type, differentproduction and equipment lines lead to many distinct versionsand variants. Nonetheless, it should be possible to integratea security system into all those platforms. In addition, thecommunication stack and security measures might be designedby different teams or vendors; a situation that clearly requireswell-defined but still flexible interfaces. These reasons ledto the development of the so called “hooking architecture”,which introduces special hooks at the interface between everylayer of the vehicular communication system. The hookingarchitecture introduces an event-callback mechanism into thecommunication stack which allows adding security measureswithout the need to change the entire communication system.The security system in a vehicle has to fulfill real-time ornear realtime requirements. For the underlying cryptographicprimitives, this implies optimized cryptographic hardware,in order to guarantee the near real-time performance. Thepotential trade-off between security and performance has tobe well balanced.To enable VC systems to withstand future, yet unknownattacks, besides the traditional prevention-oriented approach,functionalities to detect attacks, such as intrusion detectioncapabilities, and to recover after an attack, are needed. In thelong run, the goal is to enhance the resilience of the system. III. SEVECOM IMPLEMENTATION
4) Robustness requirements and time constraints are demanding.Functions necessary, for example, for drivingor alerts received via the VC system must be processedin real-time: delays or errors could lead to vehicle malfunctions,driving errors, and consequently to physicaldamages and injuries. 5) Liability and conformance require precise formulationof legal issues. Differing regulations and requirements invarious countries make it even more difficult to addressthese challenges. These observations have consequences on the implementationof a VC security system. Due to the long vehicle lifecycle, it cannot be ensured that all threats are thwarted at thetime of development. Therefore, the VC security mechanismsshould be flexible, adaptable, and extensible, to allow lateradjustments to changing security requirements. To address this need, we propose a component-based security architecture forVC systems, which allows
ISBN NO : 378 - 26 - 13840 - 9
The SeVeCom project defines baseline securityarchitecture for VC systems. Based on a set of designprinciples, SeVeCom defines an architecture that comprisesdifferent modules, each addressing certain security and privacyaspects. Modules contain components implementing one partof system functionality. The baseline specification providesone instantiation of the baseline architecture, building on wellestablishedmechanisms and cryptographic primitives, thusbeing easy to implement and to deploy in upcoming VCsystems. A. Baseline Architecture: Deployment View The SeVeCom baseline architecture addresses different aspects,such as secure communication protocols, privacy protection,and in-vehicle security. As the design and development ofVC protocols, system architectures, and security mechanismsis an ongoing process, only few parts of the overall system
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 123
Proceedings of International Conference on Developments in Engineering Research
areyet finished or standardized. As a result, a VC security systemcannot be based on a fixed platform but instead has to beflexible, with the possibility to adapt to future VC applicationsor new VC technologies. To achieve the required flexibility, the SeVeCom baselinearchitecture consists of modules, which are responsible fora certain system aspect, such as identity management. Themodules, in turn, are composed of multiple components eachhandling a specific task. For instance, the Secure CommunicationModule is responsible for implementing protocols forsecure communication and consists of several components,each of them implementing a single protocol. Components areinstantiated only when their use is required by certain applications,and they use well-defined interfaces to communicatewith other components. Thus, they can be exchanged by morerecent versions, without other modules being affected.
Fig.1. Baseline Architecture: Deployment View. As shown in Fig. 1, the Security Manager is the centralpart of the SeVeCom system architecture. It instantiates andconfigures the components of all other security modules andestablishes the connection to the Cryptographic Support Module.To cope with different situations, the Security Managermaintains different policy sets. Policies can enable or disablesome of the components or adjust their configuration, forexample, to enhance or relax the parameters for a pseudonymchange under certain circumstances.
www.iaetsd.in
inspired by similararchitectures such as the Linux Netfilter kernel subsystem.Inter Layer Proxies (ILPs) are inserted at several points in thecommunication stack. Every ILP maintains a list of callbackhandlers that are to be notified of certain events.During initialization, the SeVeCom components can registerat an ILP, subscribing for certain message types and direction(up or down the stack). Therefore, they have to implement anevent listener interface and use the registerHandler() methodto connect to an ILP. Some components may have to register atmultiple ILPs, subscribing for different kinds of packets. Whena message arrives at an ILP, an event callback is triggered forall components that have registered for this message type andtheir eventHander() method is called. The callback includes areference to the received message, and the component is thenable to inspect or modify it. By the return value the componentindicates if the message was modified, if it should be reinsertedinto the stack, or if it should be simply dropped by the ILP. The Secure Beaconing Component, for example, connects tothe ILP above the MAC layer and checks the signatures of allincoming beacon messages. Beacons with invalid signaturesare either discarded or tagged. Using this hooking architecture,it is possible to transparently integrate security functionalityinto an existing network stack with minimal modifications. Whereas events are triggered by the communication stack,the security system can also access the stack by means ofcommand calls using a well-defined API offered by stacklayers. Command calls could, e.g. instruct the MAC layer toset its MAC address to that of a new pseudonym.The hooking concept makes certain assumptions aboutthe network stack. It assumes a layered architecture, wherethe ILPs can be inserted in between, and the stack has toimplement a certain command API, e.g. for change of MACaddresses. To be able to port the SeVeCom architecture tomany different communication platforms, we also providean additional convergence layer: This defines an abstractioninterface that proxies call between the communication systemand the security components. Whenever the SeVeCom systemis ported to a new platform, besides adapting to differentpacket formats, only the ILPs and the convergence layer haveto be modified, while all other components remain unaffectedboth in terms of security and communication. C. Hardware Security Module
B. Communication Stack Integration To be independent of the actual communication stack, theintegration of the SeVeCom security system into the protocolstack is based on a hooking concept,
ISBN NO : 378 - 26 - 13840 - 9
As explained, the purpose of the Hardware SecurityModule (HSM) is to provide a physically protected environmentfor the storage of private keys
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 124
Proceedings of International Conference on Developments in Engineering Research
and for the execution ofcryptographic operations using them. Clearly, the full implementationof a HSM is beyond the scope of the SeVeComProject, but we can summarize the main requirements thatsuch an implementation should meet in order to be applicablefor securing vehicle communication systems.First of all, the HSM must be tamper resistant, to someextent. High-end tamper resistant modules (such as the IBM4758 Cryptographic Coprocessor) are too expensive to beadded to every vehicle. At the same time, we observe thatLow-end tamper resistant devices (such as smart cards) donot provide all the functionality that we need. In particular,commercially available low-end devices do not have built-inbatteries, and consequently, cannot provide a trusted internalclock. As pointed out, without a trusted source of time,such devices are not able to produce time-stamps that canbe trusted by other participants of the system. Therefore, weneed an HSM implementation somewhere between highendand low-end devices. A potential approach is to implementthe HSM as an Application-Specific Integrated Circuit(ASIC)with some special coating that provides a certain level oftamper resistance. Such a customized device can provide allthe necessary functionality by design and it can be producedin large quantity at sufficiently low costs. Second, the HSM must have an API, through which itcan provide services to the other modules of the securityarchitecture that run on the OBU. This API should supportthe digital signature and timestamping service, the decryptionservice, as well as the key and device management servicesdescribed. We specified such an API in the SeVeComProject, however, lacking the appropriate HSM hardware; weonly implemented it in the form of a software library runningon a general purpose computer. Nevertheless, besides beinguseful for demonstration purposes, our implementation canalso serve as a reference for future implementations on realHSM devices. In our implementation, we used ECDSA fordigital signature generation, ECIES with HMAC-SHA1 andAES-CBC for encryption, and we fully implemented the keymanagement services of the HSM described. Finally, we note that some examples published showthat physically secure modules can successfully be attackedthrough their weakly designed API. For this reason, we usedformal verification techniques to verify the SeVeCom HSMAPI. Our method is based on the applied pi-calculus and anautomated verification tool called Prove it. We proved that akey generated by an adversary cannot be implanted as a newroot key in the HSM through the API. Additionally, short-termand long-term private keys
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
are proved not to be revealed asthe result of possible series of function calls. D. In-Vehicle Security In order to achieve their full potential, VC systems need access to the in-car network and sensors that observe thecurrent status of the vehicle and the environment. This enablesthe VC system to process signals such as emergency braking,airbag activation, and slippery road detection, thus greatlycontributing to the avoidance of accidents and improvementof road safety.On-board system signals are transferred inside the carthrough different networks and domains. Usually, the networkarchitecture and the incar gateways restrict the signals tothe defined network segments and prevent information fromleaving its dedicated domains. This clear architecture andstrict separation is one measure that ensures the entire vehicle,especially its vital functions (brakes, engine or airbag control),always operate reliably and cannot be attacked from the outside. If this were to be changed into a more open architecture,for example, allowing reading out sensor information fromin-vehicle networks or displaying and reacting to warningmessages from external sources, it would be absolutely necessaryto ensure that in-vehicle systems are protected from anyexternal malicious influence.The In-vehicle Security Module protects the interface betweenthe incar networks and the wireless communicationsystem. It controls external access to the in-car networks,onboard control units and vehicle sensor data, but it alsoensures that data and services required by other V2V andV2I applications are provided correctly. Within the in-vehiclesecurity module, two main components are provided: (i) Afirewall that controls the data flow from external applicationsto the vehicle and backwards, and (ii) an Intrusion DetectionSystem (IDS) that constantly monitors the status of the incarsystems and provides real-time detection of attacks. The firewall realizes a packet or application based firewallapproach. Its rule-based table states which application isallowed to access each kind of data or service the IDS candynamically add rules to the firewall table, in order to denyaccess for a specific application or disable a service.The IDS is based on an anomaly detection approach,which implies that normal on-board system behavior is clearlydefined and specified. If an event results in an on-board systemstate that is not part of the standard specification, a potentiallydangerous situation is detected. Depending on the source andtype of the
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 125
Proceedings of International Conference on Developments in Engineering Research
event, appropriate reactions are taken to get thesystem back to a secure and safe state. IV. ANALYTICAL MODEL Realizing field trials for good put evaluation of vehicular wireless communication systems is very difficult and costly. Numerous vehicles and communication equipments need to be involved, and also many experimenters need to be employed. In this case, it is highly desirable to obtain theoretical analysis with real data from small scale scenarios of practical measurement results and perform an evolution prior to conducting field trials. In terms of analysis methods, were mapped previous approximations of vehicle mobility and good put into Markov M/M/1/N chain model. Use of Markov model is novel for evaluation of IEEE802.11n standard in a mobwith legacy standard (i.e. IEEE802.11g)
www.iaetsd.in
B. Results In the computation of the analytical model in the previous subsection was constructed a topology with an access point sending file data to all vehicles within coverage range of an access point. In the computation, each vehicle maintains its speed as it derives through the access point coverage range. The computation compares the results derived from trial field tests with analytical model for the single-lane in vehicle traffic. The range of good put that a vehicle can receive from the access point per pass shown in Fig.2. The results here are for the case where there are two types of vehicles, i.e, wireless-equipped and non-wirelessequipped vehicles. The type of vehicles can be interpreted as the penetration rate of wirelessequipped vehicles for use.
A. System model descript
From the Fig.2 can make the following observations:
For this model computation, was considering the case where the access point’s transmission data rate is variable through the access point coverage range.Primitive packets flow from finite wireless mobile users N and arrive to an infinite buffer of the system and are served by the server or wireless access point.
At low traffic density corresponding to high vehicle speed, there are few vehicles and as such there is a few connections using the access point resource and the value of good put is close to maximum. It is about two times less than plausible maximum good put.
In this case our system is expressed by the Kendall notation like M/M/1//N where first M-defines exponential inter arrival times between packet distribution (Poisson process), second M- defines exponential data packets transmission time distribution , next number defines the transmission channel and N- represents the number of packet sources.Queuing models for M/M/1/N systems are very elegant in analysis of wireless data networks in transmission channel with no packet loss and vehicles simultaneously under the coverage of the access point speed-N (v) (i.e, ρ (v). Based on this M/M/1/N queuing model the average good put by a vehicle can be computed as follows:
where π0 represent the probability of the idle system
where j = 1,2,3,….,N(v), π – the data packet transmission rate of the channel between vehicular and base station, λ is the packet arrival rate in the coverage range of the access point
ISBN NO : 378 - 26 - 13840 - 9
Fig.2. Average good put of a vehicle at different speed and WLAN penetration rate On low velocity increase value of vehicles and bandwidth connections increases leading to lower values of good put for the individual user. Despite reduction of maximum good put due to mobility at a velocity from 50 km/h to 100 km/h improves the good put value of a vehicle.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 126
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Penetration rates specify the possible optimal values of WLAN performance.
[3] IEEE 802.11, The Working Group for WLAN Standards,http://grouper.ieee.org/groups/802/11/ , April, 2006.
V. CONCLUSION AND FUTURE WORK
[4] ”Part 11: Wireless LAN Medium Access Control (MAC) and PhysicalLayer (PHY) Specifications Amendment 5: Enhancements for HigherThroughput”, http://ieeexplore.ieee.org/servlet/opac?punumber=53 0729, IEEE, 29.October, 2009.
In this article was presented field trial evaluationstogether with theoretical analyses of the IEEE802.11nstandard comparing with legacy standard in the vehicleenvironment. The trial field test was performed in thecontext of simple scenario of one vehicle and access point.At various velocities has been testing the performance ofWLAN. Wireless network link under fluent number ofvehicles respectively active users simultaneously realizingsuch field trials for goodput evaluation is very difficult andcostly. Therefore a simple mathematical model for goodputevaluation of vehicular communication systems in V2Iscenario was presented and analyzed for understanding thebasic processes in wireless data networks prior toconducting larger field trials. We mark that while numbers of necessary real time application of vehicular networks are the dissemination ofsafety and traffic condition messages, we can assume Wi-Fi for vehicle communication systems in the near future willalso be requested to provide different applications, for e.g.web browsing, video streaming, VoIP, downloading files,WiFi radio, etc. These types of applications have arequirement for high throughput during connections to theaccess point and existing mobile communication systemsexcept WLAN aren’t able to provide growing needs.And it is also important to note that the results wereshowing her serve as information for future analysis anddesign of vehicle networking systems.
[5] J. P.Singh, N. Bambos, B. Srinivassan and D. Clawin, Wireless LANperformance under varied stress conditions in vehicular trafficscenarios, proceedings of Vehicular Technology Conference, 2002,Vol. 2, pp. 24-28. [6] J. Ott, D. Kutscher, “Drive–thru Internet: IEEE 802.11b forAutomobile Users”, IEEE Infocom, Hong Kong, 2004. [7] R. Gass, J. Scott, C. Diot, “Measurements of In– Motion 802.11Networking”, WMCSA '06. Proceedings, 2006, pp. 69-74. [8] M. Wellens, B. Westphal, P. Mähönen “Performance Evaluation ofIEEE 802.11–based WLANs in Vehicular Scenarios”, Proc. VTCSpring, 2007. pp. 1167–1171. [9] M. Rubinstein, F. Ben Abdesslem, S. RodriguesCavalcanti, M. EliasMitreCampista, R. Alves dos Santos, L. Costa, M. Dias de Amorim,O. Duarte, “Measuring the capacity of in–car to in–car vehicularnetworks” IEEE Communications Magazine, Vol. 47., Iss. 11, 2009.,pp. 128–136. [10] P. Richards Shock waves on the highway. Operations Research 4,1956, 42–51.
VI. REFERENCES [1] Janis Jansons, Ernests Petersons, NikolajsBogdanovs, “WiFi for Vehicular Communication Systems”, 2013 27th International Conference on Advanced Information Networking and Applications Workshops.
[11] M. Lighthill and G. Whitham “On kinematic waves: II. A theory oftraffic on long crowded roads.” Proc. Roy. Soc. of London A 229,1955, pp 317–345. [12] B. Kerner and P. Konhauser, “Structure and parameters of clusters intraffic flow”, Physical Review E 50, 1994, pp 54–83.
[2] "Part 11: Wireless LAN Medium Access Control (MAC) and PhysicalLayer (PHY) Specifications Amendment 6: Wireless Access inVehicular Environments", http://ieeexplore.ieee.org/servlet/opac?punumber=55 14473, IEEE,15.July, 2010.
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 127
Proceedings of International Conference on Developments in Engineering Research
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in 7
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 128
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
An Efficient and Large data base using Subset Selection Algorithm for Multidimensional Data Extraction M.Nagendra M.Tech, Dept. of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India
Abstract- Feature subset selection is an effective way for reducing dimensionality, removing irrelevant data, increasing learning accuracy and improving results comprehensibility. This process improved by cluster based FAST Algorithm and Fuzzy Logic. FAST Algorithm can be used to Identify and removing the irrelevant data set. This algorithm process implements using two different steps that is graph theoretic clustering methods and representative feature cluster is selected. Feature subset selection research has focused on searching for relevant features. The proposed fuzzy logic has focused on minimized redundant data set and improves the feature subset accuracy.The FAST algorithm works in two steps. In the first step, features are divided into clusters by using graph-theoretic clustering methods. In the second step, the most representative feature that is strongly related to target classes is selected from each cluster to form a subset of features. Features in different clusters are relatively independent; the clustering-based strategy of FAST has a high probability of producing a subset of useful and independent features. To ensure the efficiency of FAST, we adopt the efficient minimum-spanning tree clustering method. Index Terms â&#x20AC;&#x201C; Feature Extraction, Minimum Spanning Tree, Clustering I. INTRODUCTION The performance, robustness, and usefulness of classification algorithms are improved when relatively few features are involved in the classification. Thus, selecting relevant features for the construction of classifiers has received a great deal of attention.With the aim of choosing a subset of good features with respect to the target concepts, feature subset selection is an effective way for reducing dimensionality, removing irrelevant data, increasing learning accuracy, and improving result comprehensibility. Many feature subset selection methods have been proposed and studied for machine learning applications. i. Existing System The embedded methods incorporate feature selection as a part of the training process and are usually specific to given learning algorithms, and therefore may be more efficient than the other three categories. Traditional machine learning algorithms like decision trees or artificial neural networks are examples of
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 129
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
embedded approaches. The wrapper methods use the predictive accuracy of a predetermined learning algorithm to determine the goodness of the selected subsets, the accuracy of the learning algorithms is usually high. However, the generality of the selected features is limited and the computational complexity is large. The filter methods are independent of learning algorithms, with good generality. Their computational complexity is low, but the accuracy of the learning algorithms is not guaranteed. The hybrid methods are a combination of filter and wrapper methods by using a filter method to reduce search space that will be considered by the subsequent wrapper. They mainly focus on combining filter and wrapper methods to achieve the best possible performance with a particular learning algorithm with similar time complexity of the filter methods. ii. Proposed System Feature subset selection can be viewed as the process of identifying and removing as many irrelevant and redundant features as possible. This is because irrelevant features do not contribute to the predictive accuracy and redundant features do not redound to getting a better predictor for that they provide mostly information which is already present in other feature(s). Of the many feature subset selection algorithms, some can effectively eliminate irrelevant features but fail to handle redundant features yet some of others can eliminate the irrelevant while taking care of the redundant features. Our proposed FAST algorithm falls into the second group. Traditionally, feature subset selection research has focused on searching for relevant features. A well-known example is Relief which weighs each feature according to its ability to discriminate instances under different targets based on distance-based criteria function. However, Relief is ineffective at removing redundant features as two predictive but highly correlated features are likely both to be highly weighted. Relief-F extends Relief, enabling this method to work with noisy and incomplete data sets and to deal with multiclass problems, but still cannot identify redundant features. iii. Framework of Feature Extraction
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 130
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
Fig: Framework of Feature Cluster-Based Extraction Algorithm II. FEATURE CLUSTER BASED EXTRACTION ALGORITHM Irrelevant features, along with redundant features, severely affect the accuracy of the learning machines. Thus, feature subset selection should be able to identify and remove as much of the irrelevant and redundant information as possible. Moreover, “good feature subsets contain features highly correlated with the class, yet uncorrelated with each other.” Keeping these in mind, we develop a novel algorithm which can efficiently and effectively deal with both irrelevant and redundant features, and obtain a good feature subset. We achieve this through a new feature selection framework which composed of the two connected components of irrelevant feature removal and redundant feature elimination. The irrelevant feature removal is straightforward once the right relevance measure is defined or selected, while the redundant feature elimination is a bit of sophisticated. In our proposed FAST algorithm, it involves (i) the construction of the minimum spanning tree (MST) from a weighted complete graph; (ii) the partitioning of the MST into a forest with each tree representing a cluster; and (iii) the selection of representative features from the clusters. Feature subset selection can be the process that identifies and retains the strong T-Relevance features and selects R-Features from feature clusters. The behind heuristics are that 1) Irrelevant features have no/weak correlation with target concept; 2) Redundant features are assembled in a cluster and a representative feature can be taken out of the cluster. III. ALGORITHM AND TIME COMPLEXITY ANALYSIS The proposed FAST algorithm logically consists of three steps: (i) removing irrelevant features, (ii) constructing a MST from relative ones, and (iii) partitioning the MST and selecting representative features. i. Algorithm: FAST Inputs: ( , , … , , ) - the given data set Output: − . 1 = 1 ( , ) 2 − = 3 − > ℎ 4 = ∪ { }; 5 = ; 6 ℎ , ⊂ 7
−
ISBN NO : 378 - 26 - 13840 - 9
=
- the
−
ℎ
ℎ
.
,
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 131
Proceedings of International Conference on Developments in Engineering Research
ℎ
8
ℎ 9 10 11
ℎ
ℎ
( );
= = ℎ
∈ ,
12 13 14 15
www.iaetsd.in
( , )∧
< =
= ∅ ℎ
,
<
,
ℎ
−
∈
16
=
17 18
=
∈
∪
( , )
;
ii. Time Complexity The major amount of work for Algorithm 1 involves the computation of SU values for TRelevance and F-Correlation, which has linear complexity in terms of the number of instances in a given data set. The first part of the algorithm has a linear time complexity O(m) in terms of the number of features m. Assuming k(1≤ k ≤ m) features are selected as relevant ones in the first part, when k = 1, only on feature is selected. The second part of the algorithm firstly constructs a complete graph from relevant features and the complexity is O(k2), and then generates a MST from the graph using Prim Algorithm whose time complexity is O(k2). The third part partitions the MST and Chooses the representative features with the complexity of O(k). Thus when 1<k≤ m, the complexity of the Algorithm is O(m) CONCLUSION This project presented a novel clustering – based feature extraction algorithm for high dimensional data. The algorithm involves 1) removing irrelevant features, 2) constructing a minimum spanning tree from relative ones, and 3) partitioning the MST and extracting representative features. The purpose of cluster analysis has been established to be more effective than feature extraction algorithms. Since high dimensionality and accuracy are the two major concerns of clustering, we have considered them together in this paper for the finer cluster for removing the irrelevant and redundant features. The proposed supervised clustering algorithm is processed for high dimensional data to improve the accuracy and check the probability of the patterns. Retrieval of relevant data should be faster and more accurate. This displays results based on the high probability density thereby reducing the dimensionality of the data. FUTURE ENHANCEMENT
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 132
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
In the near feature, we plan to analyze the distinct types of relationship measures and some formal properties of feature space.
REFERENCES [1] Almuallim H. and Dietterich T.G., Algorithms for Identifying Relevant Features, In Proceedings of the 9th Canadian Conference on AI, pp 38-45, 1992. [2] Arauzo-Azofra A., Benitez J.M. and Castro J.L., A feature set measure based on relief, In Proceedings of the fifth international conference on Recent Advances in Soft Computing, pp 104-109, 2004. [3] Bell D.A. and Wang, H., A formalism for relevance and its application in feature subset selection, Machine Learning, 41(2), pp 175-195, 2000. [4] Biesiada J. and Duch W., Features election for high-dimensionaldataĹ&#x201A;a Pearson redundancy based filter, AdvancesinSoftComputing, 45, pp 242C249, 2008. [5] Butterworth R., Piatetsky-Shapiro G. and Simovici D.A., On Feature Selection through Clustering, In Proceedings of the Fifth IEEE international Conference on Data Mining, pp 581-584, 2005. [6] Cardie, C., Using decision trees to improve case-based learning, In Proceedings of Tenth International Conference on Machine Learning, pp 25-32, 1993. [7] Butterworth R., Piatetsky-Shapiro G. and Simovici D.A., On Feature Selection through Clustering, In Proceedings of the Fifth IEEE international Conference on Data Mining, pp 581-584, 2005. [8] Cardie, C., Using decision trees to improve case-based learning, In Proceedings of Tenth International Conference on Machine Learning, pp 25-32, 1993. AUTHORS MosaNagendrahas received his B.Tech degree in Information Technology from Jaganâ&#x20AC;&#x2122;s College of Engineering & Technology, Venkatachalam affiliated to JNTU, Anantapur in 2012 and pursuing M.Tech degree in Computer Science & Engineering at Audisankara College of Engineering & Technology, Gudur affiliated to JNTU, Anantapur in (2012-2014).
ISBN NO : 378 - 26 - 13840 - 9
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 133
Proceedings of International Conference on Developments in Engineering Research
www.iaetsd.in
A Review on ECG Arrhythmia Detection based on DD-DW Transformation 1
B.Maheswari M .Tech Student
2
T.Nirmala M.Tech
Assistant professor
3
A.Rajani M.Tech
Assistant professor
Department of ECE, Annamacharya Institute of Technology and Sciences, Tirupati, India-517520. 1
mahisolmon@gmail.com nimmohan12@gmail.com 3 rajanirevanth446@gmail.com 2
Abstractâ&#x20AC;&#x201D;Machine learning of Electrocardiogram (ECG) is a core component in any of the ECG-based healthcare informatics system. Since the ECG is a nonlinear signal, the subtle changes in its amplitude and duration are not well manifested in time and frequency domains. Therefore, a machine-learning approach to screen arrhythmia from normal sinus rhythm from the ECG is proposed. The methodology consists of R-point detection using the double density discrete wavelet transformation (DD-DWT) decomposition, statistical validation of features. The average accuracy of classification is used as a benchmark for comparison. Support vector machine (SVM) kernel is used as classifier for better classification purpose. The proposed method will provide better results compared to state-ofart criteria like Signal quality indices (SQI) based feature extraction method. DD-DWT based ECG feature extraction is used in clinical purpose like intensive care unit (ICU) monitors for diagnosis of abnormalities in heart beat and is used for psychological analysis of human activities. Index Termsâ&#x20AC;&#x201D;Electrocardiogram (ECG), intensive care unit (ICU), signal quality.
I. INTRODUCTION In certain signal processing applications, like de noising, over complete transforms can offer a better tradeoff between performance and complexity, compared to critically sampled transforms. A distinguished member of the family of over complete discrete wavelet transforms (DWT) is the double density (DD) DWT [1], based on the filter bank shown in Figure 4. The input signal is split in three channels, each decimated by a factor of two. The signal on the first channel is processed by an identical filter bank.
ISBN NO : 378 - 26 - 13840 - 9
The DD-DWT [1] is expansive with a factor of two, compared to the critically sampled DWT arrhythmias based on their characteristics features extracted from ECG signals. Intended for Premature ventricular contraction (PVC) beat exposure RR interval ratio and power of beat is calculated. Dominant notched R wave, dominant S wave, QRS duration and direction of T wave are used for the detection of Left bundle branch block (LBBB) and Right bundle branch block ( RBBB). The rest of this paper is organized as follows: presents the ECG signal processing, ECG De noising, ECG parameter calculation, DD-DW Transformation, Feature Extraction and Selection. Describes the detection of arrhythmia. Finally, the summarizes the result & conclusion of this work. II. EXISTING APPROACH A. Pre-processing of ECGs Each channel of ECG was filtered to remove baseline wander and low frequency noise using a high pass filter with a cut-off at 1 Hz. QRS detection was performed on each channel individually using two open source QRS detectors (eplimited and wqrs). since eplimited is less sensitive to noise, we prefer eplimited detector. B. Signal quality indices Six signal quality indices (SQIs) were chosen based on earlier work and run on each of the m = 12 leads separately, producing 72 features per recording: 1. lSQI: The percentage of beats detected on each lead which were detected on all leads. 2. bSQI: The percentage of beats detected by wqrs that were also detected by eplimited.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 134
Proceedings of International Conference on Developments in Engineering Research
3. fSQI: The ratio of power P(5-20Hz)/P(0-fnHz), where fn=62.5 Hz is the Nyquist frequency. 4. sSQI: The third moment (skewness) of the distribution. 5. kSQI: The fourth moment (kurtosis) of the distribution. 6. pSQI: The percentage of the signal xm which appeared to be a flat line (dxm/dt < ÇŤ where ÇŤ =1mV). C. Support Vector Machine Classification In this section, a brief description of the two and multi-class SVM classification concept is reviewed. Support Vector Machines (SVMs) [17] are very popular and powerful in pattern learning because of supporting high dimensional data and at the same time, providing good generalization properties. Moreover, SVMs have many usages in pattern recognition and data mining applications, phoneme recognition, 3D object detection, image classification, bio-informatics etc. At the beginning, SVM was formulated for two-class (binary) classification problems. The extension of this method to multi-class problems is neither straight forward nor unique. DAG SVM is one of the methods that have been proposed to extend SVM classifier to support multi-class classification [17]. D. Binary Support vector machine formulation Let be a set of n training samples, where is an mdimensional sample in the input space. In a support vector machine, the optimal hyper plane is obtained by maximizing the generalization ability of the SVM. However, if the training data are not linearly separable, the obtained classifier may not have high Generalization ability, even though the hyper planes are determined optimally. To enhance linear separability, the original input space is mapped into a high-dimensional dot-product space called the feature space. Now using the nonlinear vector function that maps the m-dimensional input vector x into the l-dimensional feature space. E. Multi Class Support vector machine As described before, SVMs are intrinsically binary classifiers but the classification of ECG signals often involves more than two classes.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
In order to face this issue, a number of multiclass classification strategies can be adopted. III. PROPOSED CRITERIA A.ECG Signal Processing The proposed method includes processing and parameter calculation of ECG and then detection of cardiac arrhythmia using an algorithm developed in MATLAB 7.12 simulation tool. The algorithm is tested over MIT-BIH Arrhythmia database. B.ECG De noising ECG signals are usually corrupted by several noises like 50 Hz power line interferences, baseline wander and electro mayogram (EMG). Therefore, the signal needs to be preprocessed before applying any detection algorithm. Wavelet de noising and S- Golay Filter is used for removal of baseline wander and high frequency noise. ECG unfiltered data is passed through baseline wandering removal function, followed by wavelet based high frequency noise removal. The data is then smoothed. Further using SGolay filter. All the modules are implemented and simulated in MATLAB. C. ECG Parameter Calculation The purpose of the feature extraction process is to select and retain relevant information from original signal. The Feature Extraction stage extracts diagnostic information from the ECG signal. In order to detect the peaks, specific details of the signal are selected. The detection of R peak is the first step of feature extraction. R peak is detected by using PanTompkins algorithm. The intervals QRS, PR and QT are calculated by searching for corresponding onset and offset points in the wave. The separate logic was implemented for identifying P-onset, Q and S points, once R-peak was located using Pan-Tompkins algorithm. The window is selected around R-wave and the minimum of the points within this window are declared as Q and S points. In the differentiated signal ECG, a window of 155 ms is defined starting 225 ms before R-peak position. In this window, we search for maximum and minimum signal value. The P-wave peak is assumed to occur at the zerocrossings between maximum and minimum values within the selected window. Once P-wave peak is found, we proceed to locate waveform boundary-P-
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 135
Proceedings of International Conference on Developments in Engineering Research
wave onset. Similarly T wave ms with abnormal morphology. Premature â&#x20AC;&#x201D; i.e. occurs earlier than would be expected for the next sinus impulse. Discordant ST segment and T wave changes. There are five different types Of PVC, first Bigeminy every other beat is PVC, second Trigeminy every third beat is a PVC, third Quadrigeminy every fourth beat is a PVC, forth Couplet two consecutive PVCs and last Triplet three consecutive PVCs. The main characteristic of PVCs is its premature occurrence. This characteristic is measured by relating the RR interval lengths of heart cycles adjacent to the PVC. In case of a PVC, these lengths should be notoriously different .The method for classifying the abnormal complexes from the normal ones is based on the concepts of RR interval ratio of detected R peaks and energy analysis of ECG signal.
www.iaetsd.in
In particular, the considered beat types refer to following classes: normal sinus rhythm (N), right bundle branch block (RB), left bundle branch block (LB), and paced beat (P). In Figure 2, sample of four N, RB, LB, and P beats are noticeable. ii) Feature Description For each signal nineteen temporal features such as RR interval, PQ interval, PR interval, PT interval and three morphological features are recognized. These features are manually extracted for each beat and put into a separate vector. Each vector is tagged with one the four possible labels N, P, LB, RB.
D. Feature Extraction and Selection In this section explain the characteristics of the extracted feature from the ECG signals and the procedure designed for the extraction. Figure 1, presents the block diagram of the proposed arrhythmia classification.
Fig 3: sample features, ST interval, TP interval and RR interval
Fig.1.Block diagram of proposed arrhythmia classification
The three morphological features by computing the maximum and the minimum values of a beat in ECG signal. Signals of each beat are scaled, using the following formula, such that the range of every signal is between zero and one.
i) Dataset Description In this experiments were conducted on the ECG data as the basic signal for classification. In recent researches, the annotated ECG records, available at the MIT-BIH arrhythmia database, have been widely used for the evaluation of the performance of different classifiers. The database has 48 records with each record being an ECG signal for the duration of 30 minutes.
Fig 2: Sample signal of Normal, Paced, LBBB and RBBB
ISBN NO : 378 - 26 - 13840 - 9
The minimum and maximum voltages between the first and the second R feature is computer first and the normalization action is performed [0 1]. As mentioned before, we considered percent that are higher than 0.2, 0.5 and 0.8 as three features. Six of the 22 features called basic features are: R1, S, T, P, Q, R2 and the rest are called derived features. The derived features are calculated using the basic features via a semiautomatic procedure. We suggest first and second R point to expert using an algorithm based on maximum-minimum. Then the expert distinguishes appropriate points(R, S, T, P, Q, and R).
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 136
Proceedings of International Conference on Developments in Engineering Research
E.Double Density Wavelet Transform 1)1-D Double Density DWT (DD) The DD was consisted of two stages of filter banks as shown in Fig. 4: (i)Analysis In the analysis filter banks, three filters were implemented and the original signals were downsampled by 2 in order to decompose the signals into three sub-bands. The low frequency sub-band c(n) was produced by low pass filter h (-n) and the two 0
high frequency sub-bands d (n) and d (n) were 1
2
produced by high pass filters h (-n) and h (-n). 1
2
(ii) Synthesis The synthesis filter banks were the inverse of analysis filter banks where the three sub-bands were up-sampled by 2, filtered by the high pass filter h (n) 0
and the two low pass filters h (-n) and h (-n). The 1
2
filtered signals were combined to form the output signal x(n).
Fig.4.Filter bank diagram of Double Density DWT 2) Double Density Complex DWT (DDC) The input data, were processed by two parallel iterated filter banks h (n) and g (n) where i = 0, 1, 2. i
i
The real part of a complex wavelet transform [2] was produced by the sub-band signals of the upper DWT and the imaginary part was produced by the lower DWT.
Fig.5.Filter bank diagram of Double Density Complex DWT
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
F.DETECTION OF ARRHYTHMIA A premature ventricular contraction (PVC) also known as a premature ventricular complex, ventricular premature contraction (complex or complexes) (VPC), is Broad QRS complex (â&#x2030;Ľ 120) range [0, 1]. Therefore, S SQI and k SQI were normalized by subtracting the mean and dividing by the standard deviation, with both of which computed on the training set. Sensitivity (Se) measures the proportion of poor quality signals that have been correctly identified as such. Specificity (Sp) measures the proportion of good quality signals that have been correctly identified as acceptable, and accuracy (Ac) corresponds to the proportion of signals that have correctly been classified. These statistical measures are calculated from the number of true positive (TP), true negative (TN), false positive (FP) and false negative (FN) with Se = TP/(TP+FN), Sp = TN/(FP+TN) and Ac = (TP+TN)/(TP+TN+FP+FN). The energy of ECG signal is calculated for each beat and RR interval ratio is also calculated. The threshold for energy is taken as 65% of maximum energy and for ratio 70% of maximum ratio value. If RR interval ratio and energy is less than threshold PVC beat is detected. IV. EXPERIMENTAL RESULTS To report the two examples of design. The SDP problem has been solved using the library SeDuMi. The optimization problem has been solved with the large scale version of the Matlab function fmincon, with default parameters, excepting TolCon (the tolerance for satisfying the constraints), which has been set. The tolerance ÇŤ from has been set , which means that the PR constraints are satisfied with very good precision. The results obtained after 10 refinement iterations, although in the second example fewer iterations give approximately the same result. V. CONCLUSION The contribution of this is twofold. Firstly, the Hilbert transform [4] FIR approximation problem can be expressed as an SDP problem and solved reliably. Secondly, a complete algorithm to compute the second filter bank of a dual-tree double-density DWT. The algorithm comprises an initialization step based on SDP, followed by iterative refinement via nonlinear optimization.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 137
Proceedings of International Conference on Developments in Engineering Research
Previous method Using SQI Specificity
0.7500
Proposed method using DDDWT 0.92
Sensitivity
0.8000
0.87
Accuracy
0.7931
0.82
Precision
0.9524
0.97
REFERENCES [1] I.W. Selesnick, “The Double-Density Dual-Tree DWT,” IEEE Trans. Signal Processing, vol. 52, no. 5, pp. 1304–1314,May 2004. [2] N.G. Kingsbury, “Complex wavelets for shift invariant analysis and filtering of signals,” J. Applied and Computational Harmonic Analysis, vol. 10, no.3, pp. 234–253,May 2001. [3] I.W. Selesnick, R.G. Baraniuk, and N.G. Kingsbury,“The Dual-Tree ComplexWavelet Transform,”IEEE Signal Proc.Magazine, vol. 22, no. 6, pp. 123–151, Nov. 2005. [4] I.W. Selesnick, “Hilbert Transform Pairs of Wavelet Bases,” IEEE Signal Proc. Letters, vol. 8, no. 6, pp. 170–173, June 2001. [5] R. Yu and H. Ozkaramanli, “Hilbert Transform Pairs of Orthogonal Wavelet Bases: Necessary and Sufficient Conditions,” IEEE Trans. Signal Processing, vol. 53, no. 12, pp. 4723–4725, Dec. 2005. [6] B. Dumitrescu, “SDP Approximation of a FractionalDelay and the Design of Dual-Tree Complex Wavelet Transform,” IEEE Trans. Signal Proc.,2008, to appear. [7] B. Dumitrescu, Positive trigonometric polynomials and signal processing applications, Springer, 2007. [8] T.N. Davidson, Z.Q. Luo, and J.F. Sturm, “Linear Matrix Inequality Formulation of Spectral Mask Constraints with Applications to FIR Filter Design,” IEEE Trans. Signal Proc., vol. 50, no. 11, pp. 2702– 2715, Nov. 2002. [9] B. Alkire and L. Vandenberghe, “Convex optimization problems involving finite autocorrelation sequences,” Math. Progr. ser. A, vol. 93, no. 3, pp. 331–359, Dec. 2002.
ISBN NO : 378 - 26 - 13840 - 9
www.iaetsd.in
[10] B. Dumitrescu, I. T˘abus¸, and P. Stoica, “On the Parameterization of Positive Real Sequences and MA Parameter Estimation,” IEEE Trans. Signal Proc., vol. 49, no. 11, pp. 2630–2639, Nov. 2001. [11] Y. Genin, Y. Hachez, Yu. Nesterov, and P. Van Dooren, “Optimization Problems over Positive Pseudopolynomial Matrices,” SIAM J. Matrix Anal. Appl., vol. 25, no. 1, pp. 57–79, Jan.2003. [12] J.F. Sturm, “Using SeDuMi 1.02, a Matlab Toolbox for Optimization over Symmetric Cones,” Optimiza-tion Methods and Software, vol. 11, pp. 625–653,1999. [13] J. Gao, H. Sultan, J. Hu and W. W. Tung, “De noising nonlinear time series by adaptive filtering and wavelet shrinkage: a comparison,” IEEE Sig. Proc. Let., vol. 17, no. 3, 2010, pp. 237-240. [14] Fraser HSF, Blaya J. Implementing medical information systems in developing countries, what works and what doesn’t. AMIA Annu Symp Proc 2010;2010:232–236. [15] Waegemann CP. mHealth: the next generation of telemedicine? Telemed J E Health 2010;16(1):23–25. [16] Li Q, Mark RG, Clifford GD. Robust heart rate estimation from multiple asynchronous noisy sources using signal quality 1. indices and a Kalman filter. Physiological measurement 2008;29(1):15–32. [17] Bishop CM. Pattern Recognition and Machine Learning. Springer Verlag, 2006. [18] Monasterio V, Clifford GD. Robust apnoea detection in the neonatal intensive care unit. Annals of Biomedical Engineering 2011;In Submission.
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT 138