Creating Society's Central Nervous System

Creating Society’s Central Nervous System I N T E L L I G E N T S Y S T E M S R E S E A R C H AT B O S T O N U N I V E R S I T Y ’ S C O L L E G E O F E N G I N E E R I N G

Advancing Society’s Central Nervous System We are poised to acquire, analyze and act on information from an extraordinary number of networked sources, significantly increasing our ability to monitor and enhance our environment and our health. At the core of this emerging “central nervous system” will be a diverse array of engineered systems that extend the reach of human intelligence and target critical societal challenges in health care, communications, energy, national security and other domains.

At Boston University’s College of Engineering, more than two dozen faculty members are taking a comprehensive approach to the design of intelligent systems that advance such societal goals. The College has established substantial prominence at the interface of information science—the sensing, communication and processing of various forms of information with the goal of developing secure networked systems for optimum decision-making and control—and systems engineering, the mathematical modeling, simulation, analysis, optimization, control and management of systems that perform in a wide range of scenarios.

Upgrading Our Quality of Life Our deep bench of expertise in both disciplines—based at the Electrical & Computer Engineering Department’s Information Sciences and Systems group, the Division of Systems Engineering, and the Boston University Center for Information and Systems Engineering—is advancing several new intelligent systems that promise to vastly improve our quality of life. These systems are designed to deliver • robotics, automation and control solutions that enable robots, unmanned aerial vehicles (UAVs) and other mobile devices to perform military surveillance, disaster recovery and other complex missions with greater autonomy and precision.

• communications and networking solutions that promote fast, secure, reliable, energy-efficient, low-cost data transmission across wireless sensor and communication networks to support activities ranging from secure smartphone communication to remote ecological monitoring.

• signal processing and decision theory solutions that exploit machine learning, signal and image processing to enable specific actions such as minimizing building energy consumption and detecting security threats in cluttered environments.

• computational biology solutions that model metabolic and genetic networks and predict protein structures to pinpoint new drug targets and treatments for tuberculosis, cancer and other diseases.

Collaborating Across Boundaries To advance such high-impact solutions, College of Engineering researchers routinely work across disciplinary and campus boundaries.

For example, faculty from the Mechanical Engineering and Electrical & Computer Engineering Departments are collaborating with researchers from the School of Management to explore the development and adoption of cyber-physical systems—from smart grids to smart lighting—that seamlessly combine physical and computational elements to reduce energy consumption, costs and pollution in an urban neighborhood setting.

Drawing on Boston University’s extensive academic resources, major government funding agencies such as the National Science Foundation and National Institutes of Health, and substantive collaborations with industry, College of Engineering faculty are at the forefront of shaping society’s emerging information infrastructure. The following pages offer a glimpse of the intelligent systems they’re designing.

1

Research at a Glance ROBOTICS, AUTOMATION AND CONTROL

COMMUNICATIONS AND NETWORKING

4 6 8 10 12

14 16 18 20 22

Flight Lessons from Bats, Birds and Insects

Always Watching

Man or Machine?

Command and Control

On-the-Fly Maneuvers

Illuminate and Communicate

Securing the Smartphone

Reliable Updates

Eyes on the Street

Remote Viewing

College of Engineering faculty affiliations that appear in this brochure include home departments—Biomedical Engineering (BME), Electrical & Computer Engineering (ECE) and Mechanical Engineering (ME), and divisions—Materials Science & Engineering (MSE) and Systems Engineering (SE).

2

SIGNAL PROCESSING AND DECISION THEORY

COMPUTATIONAL BIOLOGY

24 26 28 30 32 34

36 38 40 42 44 46

Where the Action Is

Smart Neighborhoods

Stopping Heart Attacks

Homeland Security, Automated

Engineering Smart Cities

Security by the Pixel

Decoding TB

Probing for Drug Sites

A Smarter Screen Test

Algorithms Against Cancer

Obesity 2.0

Researchers at a Glance

3

Flight Lessons from Bats, Birds and Insects Goal: Biologically-inspired flight control algorithms that enaBle UaVs to

naVigate more effectiVely in clUttered enVironments applications: Unmanned aircraft for military missions and disaster recoVery Can studying the flight dynamics of bats, birds and insects lead to a new generation of unmanned aerial vehicles that can navigate more effectively in cluttered environments for military, disaster recovery and other missions? To maneuver as well as winged animals in tight places such as forests and caves, and land as safely on variable and moving terrain, an engineered system would have to incorporate unprecedented sensing and control capabilities while satisfying complex physical design, weight and computational requirements.

Toward that end, a team of College of Engineering researchers in the Systems Engineering Division— Professors John Baillieul (ECE, ME) and Ioannis Paschalidis (ECE) and Assistant Professor Calin Belta (ME)—is developing a set of biologically-inspired flight control algorithms. In collaboration with biologist Thomas Kunz and computer scientist Margrit Betke at Boston University and multidisciplinary researchers at three other universities under an Office of Naval Research grant, the systems engineers are carefully studying and modeling the dynamics of different airborne species.

4

Intelligent Systems Research BU College of Engineering bu.edu/eng

Computer reconstruction of a 3-D Lidar image of a site with giant sequoia trees and red firs in the Sierra National Forest in California. Images of actual animal habitats will be used to reconstruct animal flight paths through the imaged habitats. (Image courtesy of Xiaoyuan Yang.)

“We’re learning how these animals move from place to place and react to obstacles, and rethinking flight control algorithms from the ground up,” says Belta. “Classical flight control algorithms emphasize stability and safety, but it may be advantageous to modify them so vehicles can react quickly to the environment.”

In one scenario, Baillieul, Belta and Paschalidis will use computer-enhanced images of bat trajectories through forests to develop algorithms approximating the bats’ flight, and ultimately test them on real vehicles.

Recognizing that bats and other winged creatures often fly in formation, the College of Engineering team’s work will leverage its previous research on multiple robot formation control and feedback control of mobile vehicles.

“We’ve worked with ground-based robots and operated them in formation,” says Baillieul. “The goal is to take what we know about controlling groups of mobile robots and apply it to aerial vehicles that must rapidly maneuver through clutter.”

5

Always Watching Goal: coordinated teams of aUtonomoUs, intelligent agents that can interpret and reason aBoUt their enVironment in UnpredictaBle conditions with minimal hUman sUperVision applications: sUrVeillance, traffic management, health monitoring

A fleet of six unmanned Navy boats patrols a stretch of Boston Harbor for suspicious individuals, vehicles entering restricted areas and incoming liquid natural gas containers. Rigged with video cameras, laser range finders, navigation and control sensors, and on-board computers—and linked together in a network— the six vessels update one another upon detecting potential security breaches.

Meanwhile, a human operator continually interacts with the network to obtain critical information, but finds herself overwhelmed by the task of supervising multiple vehicles subject to ever-changing conditions.

In the Robotic Urban-Like Environment (RULE) in the Belta Lab, Aiming to radically reduce the workload for human operators of semiKhepera III car-like robots move autonomously on streets while autonomous underwater, ground and aerial vehicles in military and civilian contexts, staying in their lanes, obeying traffic rules and avoiding collisions.

Professor Calin Belta (ME, SE) uses RULE to test autonomous agent Assistant Professor Calin Belta (ME, SE) and Professor Christos Cassandras (ECE, decision-making performance in an urban context.

SE) are developing autonomous, intelligent single agents—entities that compute, communicate and control—that can interpret and reason about their environment in changing conditions, as well as networks of multiple agents that can safely and efficiently coordinate their activities with other agents and human operators. Potential applications include persistent surveillance, automated parking and medical monitoring.

6

A persistent surveillance scenario in which intelligent agents coordinate to survey a complex scene with areas weighted by importance and recognize abnormal activity. (Photo courtesy of the

Intelligent Systems Research BU College of Engineering

Office of Naval Research.) bu.edu/eng

Collaborating with machine learning and control theory experts from MIT, University of Pennsylvania and University of California, Berkeley under an Office of Naval Research Multidisciplinary University Research Initiative grant, Cassandras and Belta are devising optimization methods to model the behavior of single and multiple agents and probabilistic techniques to model uncertain conditions, and deploying small robots that carry cameras, communicate with each other and perform tasks in simulated settings.

For instance, to test autonomous agent decision-making performance in an urban context, Belta sends robotic cars on various missions in a model city graced with plastic towers, makeshift roads and computercontrolled traffic lights.

“We’re trying to come up with formal proofs for our controls and communications strategies and to ensure they’re bug-free,” he says. “We want to make sure that our control systems always work, regardless of operating conditions.”

7

Man or Machine?

Goal: new strategies to optimize partition of hUman and roBot tasks in hUman/roBot teams applications: mixed hUman/roBot team missions in UnpredictaBle enVironments If you are assembling a team of humans and robots to accomplish a specific objective, what are the right ways to partition tasks?

Who is in charge under what circumstances, and what psychological factors are involved?

Addressing these questions in a five-year, Air Force Office of Scientific Research-funded project, a team of mathematicians, cognitive and social psychologists, and engineers—including Professors John Baillieul (ECE, ME, SE) and David Castanon (ECE, SE)— is conducting several experiments to improve joint human/robot decision-making. Potential applications include air force missions involving cooperation between human controllers and unmanned aerial vehicles (UAVs) in unpredictable, hostile environments.

One set of experiments explores decision-making in reconnaissance simulations in which human controllers collaborate with sensor-based robots to measure pollutant concentration levels, such as radiation fallout after a nuclear accident. The controllers direct mobile platforms (UAVs or underwater vehicles) equipped with sensor-based control algorithms to detect various substances and their concentrations.

Mobile Khepera III robots emulate the motions of salsa dancers in experiments focused on communication through action.

8

Intelligent Systems Research BU College of Engineering

“If you just let people know the clock is running and they have to explore as much territory as possible and report back as much detail as possible, how will they trade off the level of detail against the amount of time that they have to acquire information?” says Baillieul. “Working with the robots, they have to decide when is enough enough, and when is it time to gather information somewhere else.”

In a second set of experiments on communication through action, computer-assisted video cameras deconstruct the movements of salsa dancers.

“We’d like to understand exactly how you can automate/predict the motion evolution in dance and competitive sports,” Baillieul explains, “and how to write simple computer programs to enable mobile robots to react appropriately to the motions and gestures they were seeing on the part of human team members.”

bu.edu/eng

9

Command and Control Goal: analysis and control strategies for systems operating in UnpredictaBle enVironments applications: moBile roBot and Biological networks, power systems and other complex, dynamic systems

10

Command to mobile robot: “Moving only through regions determined to be safe, travel first to a region containing medical supplies, and then to a region containing an injured person.�

Easier said than done. In mobile robots, biological networks, power systems and other complex, dynamic systems, electronic noise can divert the system from executing a specified task. For instance, in a mobile robot positioned at a T-intersection of two hallways, sensor noise can lead the robot to misjudge its location and initiate a turn at the wrong place, and actuator noise can cause it to turn right instead of left.

But there is a way to get a handle on this noise and help keep mobile robots and other complex systems on task.

In a National Science Foundation-funded project, Assistant Professors Sean Andersson and Calin Belta (both ME, SE) are developing control algorithms for systems subject to sensor and actuator noise—algorithms that maximize the probability that the system will respond correctly to a given command. The project is one of the first to take a probabilistic approach to this problem and to address noise in complex, dynamic systems.

Intelligent Systems Research BU College of Engineering bu.edu/eng

As a robot moves through its environment, it must make choices as to what immediate actions to perform so as to achieve a specified task. For example, in this image, as the robot moves into the intersection, should it go right, straight or left? This research strives to develop algorithms that will allow the system to autonomously determine the choice that maximizes the probability of achieving the task in the face of sensor and actuator noise.

“With the algorithms we are developing, the system would be handed the task and then autonomously determine and carry out the best control policy,” says Andersson. “Our research is best suited for complex systems in uncertain environments, from automated lawnmowers to search-and-rescue robots.”

Incorporating a mathematical model of a given system, real-time sensor data and instructions in a specialized control language that specifies system tasks and attaches probabilities to them, the researchers’ algorithms produce a control choice at every time-step. Although focused on fundamental, theoretical development, this research also involves the implementation of the algorithms on real robots.

11

On-the-Fly Maneuvers Goal: algorithms to enaBle dramatic improVements in real-time, UaV onBoard decision-making capaBility applications: sUrVeillance, air traffic control, disaster recoVery While unmanned aerial vehicles (UAVs) require no onboard pilots to accomplish their missions, ground-controlling them is highly labor intensive. Though automated to follow a trajectory and stabilize amid wind gusts, UAVs cannot perform more sophisticated maneuvers, such as adjusting a flight plan based on unforeseen events and variable weather conditions, without human input. As a result, for each UAV performing a mission, multiple Air Force pilots are needed to provide ground control.

But if algorithms that Professors David Castañón and Christos Cassandras (both ECE, SE) are developing gain traction, those pilots could end up controlling squadrons of eight to ten UAVs rather than teaming up to operate a single vehicle. Funded by the Air Force Office of Scientific Research, Castañón and Cassandras are applying systems engineering techniques to achieve dramatic improvements in UAV onboard decision-making capability.

“Currently, UAVs will go where you want them to go, but they don’t know why,” says Castañón. “We’re trying to develop approaches where teams of UAVs, given sufficient processed information, can determine what tasks each UAV should be doing next.”

12

Intelligent Systems Research BU College of Engineering bu.edu/eng

Castañón and Cassandras’ algorithms seek to provide enough automation to enable UAV teams not only to determine and execute an optimal coordinated mission, but also to depart from that plan when unexpected conditions arise.

“Our approach combines both strategic and tactical decision-making,” Castañón explains. “The algorithms use optimization techniques to enumerate the contingencies robots would encounter over time, steer them into positions where they’re likely to succeed at their missions, and empower them to make their own decisions based on real-time information—rather than preplanning the entire mission ahead of time.”

By significantly reducing manpower requirements, these algorithms could free up more UAVs to perform surveillance, air traffic control, disaster relief and other missions.

Students conduct experiments on a robotic testbed developed by ECE/SE Professors Christos Cassandras (second from right) and David Castañón, in which small, sensor-toting robots represent coordinated UAV teams.

13

Illuminate and Communicate Goal: to create immersiVe lighting that exploits noVel, controllaBle led lUminaries that can improVe health, safety and energy efficiency as well as indoor wireless networking performance applications: immersiVe, intelligent indoor spaces that react to hUman needs in a cost-effectiVe and ecologically-friendly way

Imagine flipping a digital wall switch to activate a white LED ceiling lamp that illuminates your living room and connects your laptop, PDA and other networked electronic devices in the room to the Internet. Requiring far less energy than incandescent or compact fluorescent bulbs, a new generation of highly adaptable and controllable solid state “smart lights” could illuminate a defined space and facilitate optical wireless communication among electronic devices within that space.

“LEDs can be turned on and off very quickly, or modulated, to achieve data transmission. Mobile devices such as smartphones or laptops can receive the pulsed light in a point-to-point or point-to-multipoint broadcast,” says Professor Thomas Little (ECE, SE). “This scheme can vastly improve the wireless capacity of indoor spaces when used by itself or in conjunction with existing WiFi.”

14

Intelligent Systems Research BU College of Engineering bu.edu/eng

Little and Research Associate Tarik Borogovac (ECE) and Assistant Professor Hatice Altug (ECE, MSE) are developing this technology as part of the National Science Foundation Smart Lighting Engineering Research Center (ERC). Boston University is one of three core institutions in an international, interdisciplinary effort to advance intelligent lighting systems and a robust smart lighting industry.

The BU team’s current prototype consists of a three-by-six-inch board comprised of two sections—one containing an array of nine high-brightness, white LEDs that transmit data to other transceivers, the other with an array of three photodiodes that receive data.

Eventually, each board will serve as an Internet and data communication access point for any electronic device within range, and thus enable multiple devices to communicate with one another.

“Our research is directed at the intersection of networking, free-space optical communications and the ‘anywhere’ computing that they enable,” says Little. “Examples include achieving high definition video streaming in high bandwidth-density scenarios such as aircraft seating in-flight entertainment, and improving Researchers at the National Science Foundation Smart Lighting road safety via robust vehicle-to- Engineering Research Center, of which Boston University is a core partner, are enhancing the BU Smart Lighting prototype in vehicle communications.” collaboration with academic and industrial partners.

15

Securing the Smartphone Goal: new methods and algorithms to identify threats, detect attacks, protect User priVacy and aUthenticate Users in programmaBle smartphones applications: smartphone secUrity

In the past few years, more and more of us have become reliant on smartphones for on-the-go financial, business and social transactions. But the security of the information we store on these devices could decline considerably as hardwired or proprietary features increasingly give way to opensource software programs that users can customize to fit their needs. Identifying, understanding and mitigating new security risks to these “open softphones” will be critical to ensuring their continued viability and success in the mobile communications marketplace.

To that end, Professor Mark Karpovsky (ECE) and Associate Professors David Starobinski and Ari Trachtenberg (both ECE, SE) plan to address hardware, software and networking challenges in making softphones more secure. Their effort is part of a multidisciplinary National Science Foundation-funded project based at Boston University’s Center for Reliable Information Systems and Cyber Security.

“Our goal is to preempt major security problems before the technology becomes mature,” says Trachtenberg, one of the project’s chief architects. “We seek to understand these problems and identify new opportunities for solving them.”

16

Screenshot of an Android-based phone application requesting permissions from the phone user. Many users simply click OK without having any idea about the significance of the requested permissions.

Intelligent Systems Research BU College of Engineering bu.edu/eng

Their strategy is to leverage the devices’ unique features, which include sensors, accelerometers, GPS and digital cameras. These technologies could be exploited to identify threats, detect attacks, protect user privacy and authenticate users. For example, a softphone’s sensors could be programmed to confirm its user’s biometric signature before granting access to the device.

Collaborating with experts in computer networking, security and algorithms, cryptography, and telecommunications, Karpovsky, Starobinski and Trachtenberg aim to develop more effective ways to authenticate users and callers, and design more secure networking protocols and hardware.

“In response to a growing interest in smartphone security, several researchers are attempting to fix individual hardware or software components,” says Trachtenberg.

“We’re part of an exceptionally broad, multidisciplinary team that’s addressing different aspects of the problem in a holistic and cohesive manner.”

17

Reliable Updates Goal: enaBle efficient, reliaBle third-party software dissemination in wireless networks applications: software Updates for smart phone, sensor and other wireless networks Commercial providers are increasingly allowing third parties to develop and implement their own applications on wireless devices, ranging from sensors to cellular phones. As a result, reliable, large-scale data dissemination is becoming a key enabling technology, providing fundamental services such as software upgrades and security updates. But the typically high density of wireless devices on a network makes it difficult to deliver such services efficiently.

One reason for this is the significant volume of feedback that these devices provide to cell sites and other broadcast sources to confirm receipt of data through wireless communication channels of varying reliability.

“On a network you may need to reprogram thousands of devices, and if every one of them reports back to the source/transmitter, ‘I received/I didn’t receive that information,’ you get a feedback implosion,” says Associate Professor David Starobinski (ECE, SE). “This control traffic needs to be minimized to significantly reduce network traffic congestion and thereby increase the speed of applications.”

In experiments conducted with this wireless network of sensor motes, the researchers’ algorithms minimized software update control traffic, energy consumption and delivery times.

18

Intelligent Systems Research BU College of Engineering bu.edu/eng

With funding from the National Science Foundation and Deutsche Telekom, Starobinski, Associate Professor Ari Trachtenberg (ECE, SE) and Sachin Agarwal (ECE’05) aim to reduce control traffic considerably through algorithms that combine novel theoretical and practical methods.

Pairing extreme value theory—which predicts the longest it will take a broadcast source to transmit information to all nodes—with rateless coding—which enables receivers to detect information useful to them, independent of the quality of the channel—these algorithms have shown a drastic reduction in control traffic, energy consumption and software delivery times in several sensor network experiments.

“Using our algorithms, whenever you have a new update, you broadcast a signal and one node requests the update on behalf of the whole network,” Starobinski explains. “The other nodes piggyback on the one node, thereby eliminating most of the control traffic.”

19

Eyes on the Street Goal: long-dUration, low-cost, software-optimized sensor network

to monitor for harmfUl agents and eVents applications: nUclear/Biological/chemical material detection, intelligence gathering, sUrVeillance

To detect the potential release or transport of nuclear, biological or chemical material, American cities typically deploy a small number of expensive, large sensors in strategic locations. But Professors Ioannis Paschalidis and Christos Cassandras (both ECE, SE) are turning that paradigm on its head.

With funding from the National Nuclear Security Administration in the Department of Energy and in consultation with the Los Alamos National Laboratory, they are designing a long-duration monitoring system that relies on a network of multiple, cheap, often mobile sensors rather than a few large, expensive, stationary ones. Such a system could be used not only for material detection but also for intelligence gathering in remote locations.

To maximize system performance, the researchers have devised strategies to keep sensors up and running, collecting data on potential threats and communicating it across the network.

20

Intelligent Systems Research BU College of Engineering bu.edu/eng

An iRobot programmable robot equipped with a netbook, a camera and a sensor node used in Professor Ioannis Paschalidis’ (ECE, SE) lab.

“We’ve developed algorithms that do everything from routing information to preserve energy across the network, to optimizing sensor positions to maintain good coverage of areas we’d like to monitor,” says Paschalidis.

Among other things, the algorithms maximize area coverage and minimize energy consumption by using the fewest possible sensors to cover a designated space and directing them to strategic intersections of streets and building corridors. They also track the location of each sensor by reading the strength of the signal emanating from a small radio antenna attached to the sensor, and pinpoint the source of a harmful agent release based on multiple observations of increased concentration levels near the source.

“We’ve built a testbed at the BU Photonics Center where, in a small-scale environment, we’re experimenting with these algorithms,” says Paschalidis. “Our work over the past five years provides a fundamental science base for government and industry to further develop wireless sensor networks for long term surveillance applications.”

21

Remote Viewing Goal: enaBle remote ecological oBserVation with Video in settings that are difficUlt to access, are

distUrBed By hUman presence, and span large temporal and spatial scales applications: ecological monitoring, coastal erosion stUdies, monitoring of marine protected areas Monitoring species’ behavior usually involves direct human observation or tracking of individual animals with telemetry devices.

However, many species exist in large populations, over large geographical expanses, or are otherwise impractical to observe in any large scale or for long durations.

To overcome these barriers, Assistant Professor Prakash Ishwar (ECE, SE) and Professors Janusz Konrad (ECE), Thomas Little (ECE, SE) and Thomas Kunz (Biology) proposed to deploy remotely-controlled wireless sensor networks of video cameras that collect and stream data to their workstations. It’s an approach that requires the deft integration of sensor networking—which assumes long-term deployment of large numbers of units with low data rates—with video streaming, which provides rich visual detail but needs significant energy resources to sustain data recording, communication and storage.

To that end, the researchers developed technologies for low-cost, wireless video sensor networks that operate with low energy consumption (via solar power) in remote, wireless settings and are suitable for observing wide fields of view.

Cameras at each network node cooperate to maximize data return to investigators.

A remote solar-powered camera was installed at Great Point Rip in Nantucket, Massachusetts to observe the seal population.

22

Intelligent Systems Research BU College of Engineering bu.edu/eng

Funded by the National Science Foundation, the researchers have deployed several pilot installations in Massachusetts to observe species such as seals and bats. Typical questions sought in these installations involve species censusing (e.g., “how many seals are present today?” or “what are the effects of disease on the seasonal population of bats at this site?”).

Other features of interest exist at long time scales (e.g., measurement of coastal erosion, or behavioral changes due to climate change).

To pinpoint scientifically interesting events, the team has developed action recognition algorithms to identify particular activities within the video, and a video condensation algorithm that enables investigators to bypass irrelevant video frames.

“There’s so much footage in ecological monitoring, where information is very intermittent and sporadic,” Ishwar explains. “Video condensation provides an efficient way of capturing specific information rather than going through hours of video.”

23

Where the Action Is Goal: algorithms that aUtomatically detect and recognize hUman

actions from camera-recorded digital Video signals applications: homeland secUrity, health care, ecological monitoring When watching a video, children have no problem recognizing that a person is walking, jumping or waving. Give the video footage to a computer, however, and the same task becomes daunting. For instance, different actions, such as walking and running, may appear quite similar when observed from different viewpoints, whereas the same action performed by different people may look quite different. Despite significant progress over the past decade, action recognition—the automatic detection and identification of animate actions from camera-recorded digital video signals— remains a challenging problem.

But in a National Science Foundation-funded project, Associate Professor Prakash Ishwar (ECE, SE), Professor Janusz Konrad (ECE) and graduate student Kai Guo (ECE, PhD ’11) have risen to the challenge. The researchers have developed a new action recognition algorithm that consistently exceeds the performance of state-of-the-art methods and, due to low storage and computational requirements, is suitable for real-time use.

24

Intelligent Systems Research BU College of Engineering bu.edu/eng

Unique in its combination of features developed for objecttracking and classification framework based on compressive sampling methods, the algorithm may ultimately be exploited for homeland security, healthcare A typical surveillance video from the SDHA contest and a magnified runner showing low resolution of the area of interest. monitoring, ecological monitoring, automatic sign-language recognition and other applications.

For this work the researchers won the Best Paper Award at the IEEE International Conference on Advanced Video and Signal-Based Surveillance.

And at the International Conference on Pattern Recognition, their algorithm defeated eight competing teams to win the “Aerial View Activity Classification Challenge” in the Semantic Description of Human Actions (SDHA) contest.

“There are plenty of other real-world engineering and algorithmic challenges to overcome,” says Ishwar, “but the fact that our method has performed so well consistently across several datasets, including the low-resolution SDHA dataset, is exciting.”

“Although we are very happy to have won the contest, in very difficult realworld scenarios our algorithm still misses about five times out of 100,” Konrad cautions. “This is not acceptable in practice, but we are hopeful of making further progress soon.”

25

Smart Neighborhoods Goal: new method to minimize BUilding energy consUmption and costs, and transact mUtUally Beneficial electric energy exchanges with electric Utilities Using smart micro-grid systems and societal networks applications: “smart neighBorhoods� that minimize fossil fUel consUmption/costs and maximize adoption of renewaBle energy and new sUstainaBle loads Past efforts to design more sustainable buildings have largely focused on reducing their energy consumption in isolation.

Now a new project drawing on mechanical engineering and architecture faculty at Boston University and MIT, respectively, promises to deliver substantial carbon footprint and energy cost reductions not only to individual buildings, but also to the surrounding built environment and electricity consumers in general.

Funded by the National Science Foundation, BU College of Engineering Professors Michael Caramanis and John Baillieul (both ME, SE) and their MIT collaborators plan to develop a new method to retrofit existing buildings and design new ones that minimize internal energy consumption and costs, and transact mutually beneficial electric energy exchanges with electric utilities.

They envision equipping individual buildings with a smart micro-grid that can monitor and control smart appliances, plug-in hybrid electric vehicles and other grid-friendly devices, as well as generate electricity from rooftop photovoltaic panels and wind turbines. Each building would also be configured to exchange electric energy with external energy markets, enabling it not only to draw on external power sources but also to sell competitive energy services to the grid.

College of Engineering faculty members are leading an effort to develop a new framework for advanced sustainable buildings. (Illustration by Denise Joseph and

26

Phyllis McKee.)

Intelligent Systems Research BU College of Engineering bu.edu/eng

“Our framework will enable advanced sustainable buildings to interact with nextgeneration electricity markets, including synergistic interactions between the built environment, transportation and urban infrastructures that expand the use of wind, solar and other intermittent, renewable energy sources,” says Caramanis. “We consider this adaptive interaction capability the major contribution of our research toward a sustainable energy future.”

To that end, the research team aims to create a two-layer technology platform that will enable a building to continuously optimize its energy consumption under dynamically changing internal, building-side-of-the-meter conditions—including building capabilities, safety requirements and occupant power use preferences—and external, utility-side-of-the-meter factors, such as weather and energy market trends, requirements and costs.

27

Stopping Heart Attacks Goal: enaBle high resolUtion, high signal-to-noise imaging of coronary artery inflammation throUgh pet-ct scans applications: heart disease diagnostics

The leading cause of mortality in industrialized nations, coronary artery disease is characterized by chronic inflammation and plaque formation in the arterial walls, which, if not properly diagnosed and treated, can lead to a heart attack. Accurate detection of the degree of inflammation could help prevent heart attacks, but current diagnostic techniques, some quite invasive, fall short. Positron emission tomography (PET) can noninvasively image arterial inflammation but is compromised by severe blurring induced by cardiac and respiratory motion during lengthy PET acquisitions.

Now College of Engineering Professor W. Clem Karl (ECE, SE), graduate student Sonal Ambwani (ECE, PhD’11) and physicians from Massachusetts General Hospital have joined forces to develop an advanced technique that combines PET and CT scanning to eliminate that blurring and, for the first time, provide noninvasive, direct, sharp in vivo images of developing coronary plaques.

28

Intelligent Systems Research BU College of Engineering bu.edu/eng

In a study evaluating the feasibility of high resolution PET imaging to enhance coronary artery inflammation detection, College of Engineering researchers showed that their method corrects for both cardiac and respiratory motion in a dual step algorithm.

“This could be a huge game-changer for personalized medicine (for individuals with high risk factors, it could be used to examine their plaques and determine if they’re getting better or worse) and for drug design (it would enable clinicians to test if a drug is effective in reducing inflammation),” says Karl.

Funded by the National Science Foundation, the new PET-CT coronary artery inflammation detection method compensates for both cardiac and respiratory motion and boosts image resolution.

“We’re combining super-resolution techniques from image processing with model-based techniques for motion estimation to correct the motion and increase the resolution, allowing the direct visualization of these arteries with inflammation,” says Karl.

In computer simulations the proposed technique outperformed conventional PET reconstruction and two other plaque imaging methods in reconstructing a well-defined, localized lesion. The researchers next aim is to conduct a clinical study using in vivo data from coronary artery disease patients.

29

Homeland Security, Automated Goal: machine learning, optimization and image processing techniqUes to

improVe explosiVes detection capaBilities applications: whole-Body imaging, lUggage inspection, Video sUrVeillance Airport luggage inspection machines scan one bag every six seconds, but the conventional medical imaging technology they use could easily overlook potential threats. That’s why College of Engineering Professor David Castañón is working to equip these and other explosives detection machines with a wider range of sensors and pattern recognition tools, and more sophisticated signal processing algorithms to analyze the data in real time.

Since 2008, Professors Castañón and W. Clem Karl and Associate Professor Venkatesh Saligrama (all ECE, SE) and five systems engineering PhD students have Using advanced modeling and algorithms, the ALERT team at BU aims to enable airport screening machines to provide threeaddressed these questions for whole-body imaging, video surveillance and other dimensional cross-sections of a bag so that individual items can be clearly separated. Images show top and back view of applications for the Department of Homeland Security initiative Project ALERT: Awareness boom box and other objects. (Image courtesy of Professor W. Clem and Localization of Explosive Related Threats. Castañón serves as associate director and Boston Karl (ECE, SE).)

University principal investigator of the project, which draws on multidisciplinary experts from 15 academic institutions to improve the nation’s explosives detection capability.

The BU team focuses on mathematical problems in machine learning, optimization and image processing. “We model the capabilities of different sensors, develop algorithms to combine the information they gather to form decisions concerning the presence of a potential risk, and intelligently sequence sensor data to ensure that the system as a whole performs well,” says Castañón.

30

Intelligent Systems Research BU College of Engineering bu.edu/eng

One promising solution emerging from Castañón’s team is an “adaptive training” system that uses video cameras to monitor pedestrian and traffic behavior, and “learns” on-thejob to detect abandoned packages and vehicles by tracking changes in image pixels of sidewalks and streets. The team is also designing an intelligent sensor network to monitor crowds with infrared cameras, chemical sniffers and other devices.

“In explosives detection applications, most researchers focus on improving the performance of individual components, such as sharper imaging quality,” says Castañón. “We’re exploring ways of combining components and examining tradeoffs to see how different data streams can complement each other to get a more accurate system.”

31

Engineering Smart Cities Goal: systems that reconfigUre themselVes in response to Unexpected eVents and

fast-changing conditions applications: “smart cities” that collect information aBoUt the enVironment from distriBUted sensors, make optimal decisions and inVoke actUators to execUte those decisions An estimated 30 percent of vehicles on the downtown streets of major metropolitan areas are cruising for a parking spot, and it takes the average driver 7.8 minutes to locate one. Finding a desirable parking spot in any big city can be daunting, but Professor Christos Cassandras (ECE, SE) and graduate student Yanfeng Geng (SE) have produced a preliminary version of a system designed to do just that.

They envision a smart GPS-based system that takes your request for a destination such as “Colonial Theatre” along with some personal preferences, and directs you to the best parking spot for you—all while managing hundreds of similar requests. Not only do you arrive at your destination faster; the city benefits from reduced traffic congestion and accidents, and increased revenue through more efficient parking capacity utilization.

32

Intelligent Systems Research BU College of Engineering

“We’re trying to design sensor networks as a closed-loop control system that not only collects information but acts on it as well,” says Cassandras. “This is one of the features that will ultimately define the smart city.”

Toward that end, the system continuously monitors parking spot sensor data, vehicle locations and destination requests, and traffic conditions; optimizes a solution based on this constantly changing information; and measures overall performance of the system.

This smart parking system is part of a larger, National Science Foundation-funded project at Boston University focused on developing enterprise-level, information-driven systems that automatically reconfigure themselves in response to unexpected events and fast-changing conditions. Professors Ioannis Paschalidis (ECE, SE) and Azer Bestavros (Computer Science, SE) and two researchers from the University of Massachusetts and the University of Connecticut are also contributing to the project.

The BU-led team is one of five nationwide now exploring reconfigurable systems and their applications with NSF support. Three of these teams are pursuing ways to create “smart cities” that exploit ubiquitous wireless networking; collect data from distributed sensors; make optimal decisions about traffic, transportation, garbage collection, communication, power consumption and other urban concerns; and invoke actuators to execute those decisions.

bu.edu/eng

Robotic Urban-Like Environment (RULE) with traffic lights, parking spots and cars for smart parking experimentation at Boston University’s Control of Discrete Event Systems (CODES) laboratory.

33

Security by the Pixel Goal: fast, reliaBle method to process Video data and detect sUspicioUs oBjects or eVents in highly clUttered UrBan enVironments applications: homeland secUrity, military and domestic sUrVeillance

Each week more than 30 million surveillance cameras produce nearly 4 billion hours of video footage, but the reams of data they produce exceed the processing capacity of human analysts. Even where software is used to sift through the data for suspicious activity, the algorithms are not always up to the task, especially in busy urban areas.

Now two College of Engineering researchers—Professor Janusz Konrad (ECE) and Associate Professor Venkatesh Saligrama (ECE, SE)—and Pierre-Marc Jodoin, an assistant professor of computer science at the University of Sherbrooke in Canada, have devised a technique to process video data and pinpoint unusual objects or events in cluttered urban environments that’s much faster and more reliable than conventional approaches.

Rather than classify and track objects in a video stream, the new technique breaks footage down to a sequence of snapshots, compares pixels in subsequent snapshots for subtle changes, and uses statistical methods to identify and locate pixel-level changes that depart from normal activity within the monitored scene. Data collected on these anomalies can then be tracked via conventional software systems.

“Typical approaches entail tagging, identifying and tracking every single object, but in an urban setting with too many moving objects, you can’t track them all,” says Saligrama. “Our idea is to collect pixel-level statistics and monitor variations over time. Using cameras with embedded algorithms, we’ve shown that pixel-level anomaly detection can work.”

34

Intelligent Systems Research BU College of Engineering bu.edu/eng

“Not only is the algorithm effective at detecting anomalous dynamics in the scene, but it is also efficient in that it requires relatively low-power computing hardware and its memory footprint is minimal compared to state-ofthe-art algorithms,” notes Konrad, “enabling implementation in the camera instead of a central server.”

The project is funded by the National Science Foundation, Department of Homeland Security, National GeospatialIntelligence Agency and Office of Naval Research.

College of Engineering researchers have developed a novel statistical technique to identify and locate pixel-level changes that depart from normal activity within a monitored scene—changes that could indicate potential security threats. Here the technique identifies a streetcar as an anomaly within the clutter of traffic.

35

Decoding TB Goal: reVerse-engineer regUlatory and metaBolic networks in tUBercUlosis

BacteriUm to determine which proteins and genes trigger the disease applications: simpler, faster, more targeted diagnostics and drUgs for tB Since 2008, Associate Professor James Galagan’s (BME) group and four collaborating research teams have been building molecular maps of the regulatory and metabolic networks of the bacterium that causes tuberculosis, in a systematic effort to determine which proteins and genes in these networks trigger the disease. This knowledge could lead to simpler, faster, more targeted diagnostics and drugs for TB, which accounts for up to three million deaths per year.

“We want to understand this bug like we would understand a computer or a car or any other system that has interacting parts,” says Galagan. “Starting with the genome for this bacterium, we’re focused on the set of molecules, proteins and genes that allow TB to inflict damage on the human host.”

36

bu.edu/eng

Control and Prevention.)

Intelligent Systems Research BU College of Engineering

Myobacterium tuberculosis bacteria. Associate Professor James Galagan (BME) is taking a systematic approach to pinpointing the genes and proteins in the TB bacterium that trigger the disease. (Image courtesy of Centers for Disease

To map out the bacterium’s regulatory network, Galagan and his collaborators apply genome sequencing technology to obtain a “circuit diagram” of the network’s 180 known transcription factors, interconnected proteins that activate or deactivate the approximately 4,000 genes that control the cell. Using an advanced synthetic biology technique, they next activate the transcription factors—individually and in groups, or “subcircuits”—to see which genes get turned on or off as a result. Finally, they integrate the data into a predictive computer model, which they use to simulate the bacterium’s regulatory circuitry and pinpoint promising gene and protein subcircuits to knock out. Galagan’s team takes a similar approach to investigating the cell’s metabolic network.

“What’s unique about our approach to TB is that we’re trying to map out the terrain in a systematic, unbiased and comprehensive way, rather than focus on a select number of the 180 transcription factors,” says Galagan. “We believe it will work and produce a foundation that will allow us to think about infectious disease in a fundamentally different way that’s more quantitative, systems-oriented and comprehensive.”

37

Probing for Drug Sites Goal: compUtational techniqUe that emUlates laB-Based methods to pinpoint the most faVoraBle protein Binding sites for potential drUgs applications: drUg design for cancer and other diseases Despite the unprecedented achievements of the genomic revolution, the sequencing of the human genome has brought about very few new drugs.

Determining druggability—whether a particular validated protein target in the body is capable of binding drug-sized, small molecules designed to fight disease—remains a largely unsolved problem.

One promising solution is the use of mapping methods, which place molecular probes—small molecules or functional groups—on the surface of proteins in order to identify the most favorable binding sites for potential drugs. But such methods can be very expensive and time-consuming. Now Professor Sandor Vajda (BME, SE) and Research Assistant Professor Dima Kosakov (BME), in collaboration with Associate Professor Adrian Whitty and Professors Karen Allen and John Porco (all Chemistry), have developed and experimentally tested computational methods designed to emulate lab-based mapping methods.

38

Computational mapping of the expression initiation factor protein eIF4E, an important cancer target. The potential druggable sites are detected by the binding of small “probe” molecules. The figure also shows a novel eIF4E inhibitor fitted into the druggable sites. Indicated in blue on the protein surface are residues that, based on NMR experiments, interact with ligands.

Intelligent Systems Research BU College of Engineering bu.edu/eng

Modeling the interaction of proteins with small molecules on a supercomputer, their software-based solution replicates two industrial-scale experiments—one in which protein crystals are soaked in organic solvents to reveal regions of the protein that bind small molecules, the other which uses nuclear magnetic resonance to determine protein- small molecule interactions. The software simulates small molecular probes distributed on the protein surface in search of regions where the protein-small molecule interactions are the strongest—i.e., where the maximum number of small molecules clusters.

“We have demonstrated that we can get similar results computationally at a fraction of the cost,” says Vajda. Adds Kozakov: “Each individual lab study costs about $100,000, whereas our computational approach has essentially no cost and takes only a few hours.”

Now focused on determining the druggability of potential cancer targets, Vajda and Kozakov aim to develop their computational solution into a drug design technology over the next two years. Based at the Biomolecular Engineering Research Center, their research is primarily funded by the National Institutes of Health.

39

A Smarter Screen Test Goal: a new, inexpensiVe, genotyping techniqUe that detects aBnormal repeats of short dna seqUences within the hUman genome applications: screening indiVidUals’ sUsceptiBility to selected diseases or enaBling personalized drUg treatment

40

Short DNA sequences called short tandem repeats (STRs) appear frequently in thousands of regions of the human genome. For example, in the short sequence of nucleotide bases “AGCAGCAGCAGCAGC,” the triplet “AGC” occurs five times. Individuals occasionally differ from the norm in their number of triplet copies within the short sequence—and thus the length of the sequence itself. Scientists have long linked

Intelligent Systems Research BU College of Engineering bu.edu/eng

abnormal length variations in STRs to Huntington’s disease, schizophrenia and other major diseases, but until now have never investigated repeats at the whole genome level.

Applying computational analyses to existing sequence Now a novel approach to genotyping that exploits STR length variation has data, the research team compiled and assembled the emerged from a team of scientists that includes Biomedical Engineering Department first complete list of short tandem repeats (STRs) in regions of human and chimp genomes where the researchers Simon Kasif, professor of bioinformatics and biomedical engineering and repeated element is a triplet. For each of these regions, they compared the lengths of triplet repeats in the gene director of the Computational Genomics Lab; Professor Emeritus Charles Cantor (BME) of transcripts of the three human and one chimp genomes, and eventually observed a high rate of length variation the Center for Advanced Biotechnology; and Michael Molla, a Center for Biodynamics in longer STRs. (Image courtesy of PNAS.) fellow and post-doc in Professor James Collins’ (BME, MSE, SE) lab.

Using computational analyses to compare STRs in three well-known versions of the human genome as well as that of a chimpanzee, the researchers discovered a high rate of length variation in STRs that extend beyond 20 nucleotides. Their finding suggests a new set of sequences in personal DNA samples to examine for disease susceptibility. In the process, they also introduced a new, inexpensive, whole-genome assay technology to detect abnormal repeats that could be used to screen individuals’ susceptibility to selected diseases.

“We’re hoping to scale up the assay technology to enable us to examine STR length variations in human individuals with a particular disease, such as cancer; compare them to the normal population; and check whether the individual has a predisposition to the disease,” says Kasif.

This research is funded by the National Human Genome Research Institute, National Science Foundation and National Institutes of Health Informatics to Bedside Consortium.

41

Algorithms Against Cancer Goal: algorithms to predict the strUctUre of complexes formed By protein-protein interactions

inVolVed in metaBolic control, signal transdUction, gene regUlation and other critical processes applications: new drUgs to comBat cancer and inflammatory diseases An interdisciplinary team of College of Engineering faculty members—Professor Sandor Vajda (BME, SE), Research Assistant Professor Dima Kozakov (BME), Professor Ioannis Paschalidis (ECE, SE) and Associate Professor Pirooz Vakili (ME, SE)—are developing a family of powerful optimization algorithms for predicting the structures of complexes that form when two cell proteins bond together—structures that, in some cases, generate erroneous cell signaling pathways that can trigger cancer and inflammatory diseases.

A joint effort of Boston University’s Center for Information and Systems Engineering and Biomolecular Engineering Research Center funded by the National Institutes of Health, the project combines Paschalidis’ and Vakili’s expertise in optimization and systems theory with Vajda and Kozakov’s knowledge of biophysics and bioinformatics.

“Given the three-dimensional structure of two proteins, you’d like to predict with great accuracy the structure of the complex formed once these two proteins bind,” says Paschalidis, who compares the process to characterizing all the possible structures that pairs of Lego blocks can form out of an initial set of 1,000 blocks. “Based on laws of thermodynamics, we’ve developed optimization algorithms that have succeeded in doing just that.”

42

Intelligent Systems Research BU College of Engineering bu.edu/eng

Many biologically important protein-protein interactions (PPIs) produce fragile complexes that do not remain intact long enough to be subject to direct experimental analysis, but optimization algorithms such as those developed by the College of Engineering team can determine the structure of these complexes with great accuracy based on the structures of the component proteins.

Judged the world’s best in the most recent global blind prediction experiment conducted by the Critical Assessment of Predicted Interactions organization, the team’s high-precision computational methods are also playing a key role in a related BU-based project that seeks to develop small molecules—potential drugs—that can inhibit selected PPIs that produce structures that may provoke illness.

An interdisciplinary College of Engineering team has developed computational methods to predict the structures that form when two cellular proteins interact.

43

Obesity 2.0 Goal: determine how oBesity affects the response to infection applications: targeted diagnostics and drUgs for infections in oBese patients

44

Intelligent Systems Research BU College of Engineering bu.edu/eng

Macrophages from lean and obese mice are exposed to P. gingivalis infection. The corresponding gene expression data is fed into the researchers’ computational tools to predict genes that are essential to this response.

Experiments by Boston University Goldman School of Dental Medicine (BUGSDM) Professor Salomon Amar indicate that obese organisms have a deficient immune response to infection, leading to significantly longer-lasting illnesses. To understand the mechanisms by which high body mass index impacts the immune response—and to try to reverse the process, Amar, director of the BUGSDM Center for Anti-inflammatory Therapeutics, has teamed up with an expert in computational modeling of gene and other networks, College of Engineering Assistant Professor Calin Belta (ME, SE).

“Obesity and response to infection are well-defined, but the connection between them is not understood,” says Belta. “We’re trying to find a gene in obese organisms and change the function of that gene (knock it out or over-express it) so the organism responds better to infection.”

Funded by the National Institutes of Health, the researchers envision their work leading to a medication that would make obese individuals less prone to infectious diseases and better equipped to mount an effective immune response. Achieving such a breakthrough would reduce the complications of obesity—and the costs, which total about $150 billion annually in the U.S. To determine what impedes the immune response in obese organisms, Amar and Belta are developing a genome-scale mathematical model of metabolism, signaling and gene networks in the immune system cells of lean and fat strains of mice infected with P. gingivalis a microorganism implicated in periodontal disease (gum disease). The researchers obtain gene expression data, make computational predictions about which genes are restricting the immune response, and then knock out or over-express those genes in the lab to see if they lead to sharp differences in immune response in the two strains.

“We’ve identified some genes with the potential to be the key genes that cause the differences,” said Belta. “We’ve done this in-silico with partial validation in vitro, but we’re working to validate those genes more comprehensively in further experiments.”

45

Researchers at a Glance Assistant Professor Hatice Altug (ECE, MSE) 14 Illuminate and Communicate

Assistant Professor Sean Andersson (ME, SE) 10 Command and Control

Professor John Baillieul (ECE, ME, SE) 4 Flight Lessons from Bats, Birds and Insects 8 Man or Machine?

26 Smart Neighborhoods

Assistant Professor Calin Belta (ME, SE) 4 6 10 44

Flight Lessons from Bats, Birds and Insects Always Watching Command and Control Obesity 2.0

Professor Michael Caramanis (ME, SE) 26 Smart Neighborhoods

Professor Christos Cassandras (ECE, SE) 6 Always Watching 12 On-the-Fly Maneuvers 32 Engineering Smart Cities 46

Professor David Casta帽贸n (ECE, SE) 8 12 20 30

Man or Machine?

On-the-Fly Maneuvers Eyes on the Street Homeland Security, Automated

Associate Professor James Galagan (BME) 36 Decoding TB

Assistant Professor Prakash Ishwar (ECE, SE) 22 Remote Viewing 24 Where the Action Is

Professor W. Clem Karl (ECE, SE) 28 Stopping Heart Attacks 30 Homeland Security, Automated

Professor Mark Karpovsky (ECE) 16 Securing the Smartphone

Professor Simon Kasif (BME) 40 A Smarter Screen Test

Professor Janusz Konrad (ECE) 22 Remote Viewing 24 Where the Action Is 34 Security by the Pixel

Research Assistant Professor Dima Kosakov (BME) 38 Probing for Drug Sites 42 Algorithms Against Cancer

Professor Thomas Little (ECE, SE) 14 Illuminate and Communicate 22 Remote Viewing

Associate Professor Ari Trachtenberg (ECE, SE) 16 Securing the Smartphone 18 Reliable Updates

Professor Sandor Vajda (BME, SE) 38 Probing for Drug Sites 42 Algorithms Against Cancer

Associate Professor Pirooz Vakili (ME, SE) 42 Algorithms Against Cancer

Professor Ioannis Paschalidis (ECE, SE) 4 20 32 42

Flight Lessons from Bats, Birds and Insects Eyes on the Street Engineering Smart Cities Algorithms Against Cancer

Associate Professor Venkatesh Saligrama (ECE, SE) 30 Homeland Security, Automated 34 Security by the Pixel

College of Engineering faculty affiliations include home departments— Biomedical Engineering (BME), Electrical & Computer Engineering (ECE) and Mechanical Engineering (ME), and divisions—Materials Science & Engineering (MSE) and Systems Engineering (SE).

Associate Professor David Starobinski (ECE, SE) 16 Securing the Smartphone 18 Reliable Updates 47

Degree Programs and Research Organizations Intelligent systems are a major focus of several College of Engineering and Boston University-wide research organizations and College of Engineering degree programs.

Division of Systems Engineering (SE) bu.edu/se The Division of Systems Engineering (SE) is a unique interdisciplinary graduate program with select faculty from all engineering departments at BU. It offers PhD, MS and MEng degrees to graduate students with interests in information, decision and control sciences, and in all application areas encompassing the modeling, analysis, simulation, control, optimization and management of complex systems. Equipped with the unique skills to adapt their knowledge and expertise to different application domains, SE graduates are in high demand.

The Division offers research opportunities through the Center for Information and Systems Engineering (CISE). The cross-disciplinary SE curriculum, together with the CISE, leverage the expertise of Engineering, Computer Science, Mathematics, and Management faculty from the College of Engineering, the College of Arts & Sciences, and the School of Management. Research activities focus on automation, control and robotics; communication and networking; computational and systems biology; information sciences; and production, service systems and supply chain management.

Center for Information and Systems Engineering (CISE) bu.edu/systems CISE is a virtual, umbrella organization that convenes researchers from across BU to investigate the design, analysis and management of complex systems. CISE faculty members and the students they advise reside in home departments predominantly in the College of Engineering, but also in the College of Arts & Sciences and School of Management.

Since its founding in 2002 by a dozen College of Engineering professors, CISE has grown to represent 31 faculty members and more than 100 graduate students pursuing projects funded by the National Science Foundation, Department of Defense, Department of Energy and other major federal agencies.

CISE members have made seminal contributions in control systems, optimization and decision theory; applied probability and simulation; networking; information sciences; computational biology; and production systems.

Major accomplishments include innovative techniques for assessing network and server performance and pricing Internet services, novel image processing techniques with applications in radar and biomedical imaging, new algorithms for machine learning and pattern recognition with applications in explosives detection, new computer simulation methodologies that have been adopted by leading software companies, and advanced computational methods in structural biology.

48

ECE Department Information Sciences and Systems (ISS) Research Group bu.edu/iss The ISS group’s mission is research, education and technology transfer in all areas related to the sensing, communication and processing of information, encompassing an extensive range of natural and man-made phenomena, as well as the design and synthesis of secure networked systems for optimum decision-making and control. Comprising a third of the ECE department, currently ISS is home to 14 dynamic faculty of international renown, several post-doctoral researchers, over 50 doctoral candidates, scores of master’s students as well as a number of undergraduates exploring state-of-the-art research outside their regular curriculum.

ISS research centers on the sensing, communication and processing of various forms of information with the objective of designing and synthesizing secure networked systems for optimum decision-making and control. ISS members have a broad range of research interests but share a common approach to problem-solving, the pursuit of foundational research, and the development and utilization of sophisticated analytic and algorithmic tools from mathematics, statistics, computer science and physics.

Additional Research Organizations • Boston University Photonics Center bu.edu/photonics • Center for Biodynamics cbd.bu.edu • Center for Computational Science ccs.bu.edu • Center for Reliable Information Systems and Cyber Security (RISCS) bu.edu/riscs • Clean Energy and Environmental Sustainability Initiative (CEESI) bu.edu/energy • Computational Genomics Laboratory bu.edu/bme/research/labs/cg • Control of Discrete Event Systems (CODES) Laboratory vita.bu.edu/cgc/CODES • Hybrid and Networked Systems Laboratory hyness.bu.edu/Home.html • NSF Smart Lighting Engineering Research Center at BU bu.edu/smartlighting • Structural Bioinformatics Lab structure.bu.edu

Produced by Boston University’s College of Engineering Design and production by Crocker & Company

Nonprofit U.S. Postage

PAID 44 Cummington Street Boston, MA 02215

Boston MA Permit No. 1839

An equal opportunity, affirmative action institution.

facebook.com/BUCollegeofENG twitter.com/BUCollegeofENG

30%

bu.edu/eng