Global Environmental MEMS Sensors (GEMS): A Revolutionary Observing System for the 21st Century NIAC Phase II CP_02-01 John Manobianco, Randolph J. Evans, David A. Short ENSCO, Inc. Dana Teasdale, Kristofer S.J. Pister Dust, Inc. Mel Siegel Carnegie Mellon University Donna Manobianco ManoNano Technologies, Research, & Consulting November 2003
Outline • • • •
Description Potential applications Phase I (define major feasibility issues) Phase II – Methods / Approach – Plan
• Summary
Description • Integrated system of airborne probes – – – – –
Mass produced at very low per-unit cost Disposable Suspended in the atmosphere Carried by wind currents MicroElectroMechanical System (MEMS)-based sensors • • • • • •
Meteorological parameters (temperature, pressure, moisture, velocity) Particulates Pollutants O3, CO2, etc. Acoustic, seismic, imaging Chemical, biological, nuclear contaminants
• Self-contained with power source for – – – –
Sensing Navigation Communication Computation
Description (con’t) •
Mobile, 3D wireless network with communication among – Probes, intermediate nodes, data collectors, remote receiving platforms
Broad scalability & applicability
~1010 probes Global coverage 1-km spacing Regional coverage 100-m spacing
Potential Applications Weather / climate analysis & prediction Basic environmental science Field experiments Ground truth for remote sensing Research & operational modeling
Potential Applications Planetary science missions
Manobianco et al.: GEMS: A Revolutionary Concept for Planetary and Space Exploration, Space Technology and Applications International Forum, Symposium on Space Colonization, Space Exploration Session, Albuquerque, NM, February 2004.
Potential Applications Space Environment Monitoring Planetary science missions
Manobianco et al.: GEMS: A Revolutionary Concept for Planetary and Space Exploration, Space Technology and Applications International Forum, Symposium on Space Colonization, Space Exploration Session, Albuquerque, NM, February 2004.
Potential Applications Battlesphere surveillance Intelligence gathering Threat monitoring & assessment Homeland security
Phase I (Define Feasibility Issues) Data collection/management Data impact
Cost Networking Navigation
Communication
Dispersion Deployment
Scavenging Probe design Measurement
Environmental Power
Phase II Methods / Approach Optimization of trade-offs (cost, practicality, feasibility)
Multi-Dimensional Parameter Space (Power, Deployment, Cost,‌)
Physical limitations (measurement & signal detection)
Scaling (probe & network size)
Phase II Plan • Study major feasibility issues – Extensive use of simulation • Deployment, dispersion, data impact, scavenging, power,… • System model
– Experimentation as appropriate / practical – Cost-benefit analysis • Projected per unit & infrastructure cost • Compare w/ future observing systems • Quantify benefits
• Develop technology roadmap & identify enabling technologies • Pathways for development & integration w/ NASA missions
Meteorological Issues • • • •
Deployment strategies Dispersion Scavenging Impact of probe data on analyses & forecasts – – – – –
Dynamic simulation models Virtual weather scenarios Dispersion patterns Simulated probe data & statistics OSSE (Observing System Simulation Experiments)
Deployment / Dispersion • Release (N. Hemisphere) – High-altitude balloons – 10o x 10o lat-lon
• Deployment – 4-day release – 18-km altitude – 1 probe every 6 min
• Terminal velocity – 0.01 m s-1
• Duration – 24 days – 15 Jun – 9 Jul 2001
• Total # of probes – ~200,000
Scavenging
Light Rain
Heavy Rain
Simple Collision Model
Probability of Survival
1 8 mm/hr
0.8
128 mm/hr
0.6 0.4 0.2 0 0.01
0.1
1
10
Time (minutes)
100
1000
Observing System Simulation Experiments (OSSE) Time (days) 0
1
…….. 10
11
12
13
14 …….. 29
30
Nature run (“Truth” from Model 1) Simulated observations
Benchmark (Model 2) Data insertion window (assimilate simulated observations)
Experiment 1 (Model 2) Compare with nature & control run to assess data impact
Experiments 2, 3, …(Variations on Exp. 1)
OSSE Domains Same boundary & initial conditions 30 km
30 km
10 km
10 km
2.5 km
Nature Run (Model 1) Summer / winter case Probes deployed / dispersed for 20-30 days
OSSE (Model 2)
Engineering Issues • Components – – – –
Size & shape Sensors Fundamental limits What’s next?
• Network – Cost of basic operations – Mesh network implementation – Limitations & scaling challenges
• Optimization
Probe Components Power: • Solar cell (~1 J/day/mm2) • Batteries ~1 J/mm3 • Capacitors ~0.01 J/mm3 • Fuel Cell ~30 J/mm3
Sensing & Processing: • Temperature, pressure, RH sensors • Analog Front-end • Digital Back-end
Communication: • RF antenna (shown) • Optical receiver
230 µJ: 25 µm2 solar cell
Sample, compute, listen, talk (RF) once per hour for 10 days
• Goal: Probe dropped at 20 km remains airborne for hours to days • Strategies: – Dust sized particles – Materials – Buoyancy control: positively buoyant probes – Probe shape: dandelion/maple seed
Fall Time Increase
Probe Size & Shape
Particle Size Decrease
Sensors • •
MEMS temperature, pressure & RH sensors well-established Need to optimize range for atmospheric measurements Sensirion humidity & temperature:
Range: 0-100% RH, -40-124 ºC ±0.2% RH ±0.4 ºC $18
5 mm
Intersema pressure:
Range: 300-1100 mbar, -10-60 ºC ±1.5 mbar µW per measurement $18
9 mm
Shrinking Probes
• • • • • • •
Circuit Board Layout TI MSP430f149 16-bit processor 60kB flash, 2 kB RAM Temp, battery, RF signal sensors 7 12-bit analog inputs 16 digital IO pins 902-928 kHz operation
• • • • • •
8 bit uP 3k RAM OS accelerators World record low power 8 bit ADC (100kS/s, 2uA) HW Encryption support 900 MHz transmitter
Limiting Factors: µ-Fabricated Components • Moore’s Law • Thermal Noise: kT/2 (10-21 J) • Sensors: – Fabrication limitations (aspect ratio) – Sensitivity (lower limit: molecules in Brownian motion?) – Inherent structural motion/vibration
The Next Generation: Nano Dust? • • • •
Nanotube sensors Nanotube computation Nanotube hydrogen storage Nanomechanical filters for communication!
Cost of Basic Operations Current
Operation
[A]
Time [s]
Charge [A*s]
Sleep
3µ
Sample
1m
20µ
0.020µ
Talk to neighbor
25m
5m
125µ
10m
8m
80µ
Sound an alarm
25m
1s?
25,000µ?
Listen for alarm
2m
2m
4µ
15 byte payload
Listen to neighbor 15 byte payload
QAAbattery = 2000mAh = 7,200,000,00 µA*s
Mesh Network Routing & Localization Probe network determines optimal route to gateway, and locates probes based on signal strength and GPS sensors. Specialized GPS motes send position information to gateway.
Three motes’ routing paths Limit: Message traffic increases near gateway
Communication Limits • RF noise limit: Preceived > kTB Nf SNRmin Signal Power Received
Thermal Noise -174+53 dBm
Receiver Noise +9 dBm
Signal to Noise required by downstream processing +10 dBm
Sensitivity ≈ -102 dBm (<0.1 pW) But, transmit power must be greater due to path loss • Network communication must be rapid enough to avoid errors or loss of path due to probe motion
Link Budget ↑ Probe Spacing = ↑ Transmission Power
Transmit Power vs. Probe Spacing
Transmit Power Required (W)
0.14 0.12 0.1 0.08
Transmit Power Required for 0.1 pW at Receiver
0.06 0.04
10 GHz Antenna Gain = 3
0.02 0 0
2000
4000
6000
8000
10000
12000
Probe Spacing (m)
14000
16000
18000
20000
Network Scaling • • •
Message traffic limited near gateway Next step: event-based reporting (1-way communication) Beyond: local event-based subnet formation & reporting – any mote becomes a gateway Motes near event “wake up” and report
Lots of message traffic near gateway
Optimization: Trade-offs ↓SIZE
+ + -
↓POWER
+ Smaller power supply required - Decrease transmission distance & sampling frequency - Shorter mote life
↑# PROBES
Min Environmental Impact Slow descent Decreased power storage Decrease SNR
+ Improved network localization + Improved forecast - Increased message traffic
Demonstration Pressure
Humidity/Temperature
X,Y-Acceleration
Light
Cost / Benefit Analysis • Cost issues – – – –
Per unit cost Deployment / O&M cost Global versus regional (targeted) observations Estimates for future observing systems (in situ v. remote)
• Benefit issues – $3 trillion dollars of U.S. economy has weather / climate sensitivity – How much can we reduce sensitivity with improved observations / forecasts? – Example (hurricane track forecasts) • 72-h track forecast error ≈ 200 mi • Evacuation cost = $0.5M per linear mile • Potential savings with 10% error reduction = $10M for storms affecting populated areas
Summary • Advanced concept description – Mobile network of wireless, airborne probes for environmental monitoring
• Phase I results – Define major feasibility issues – Validate viability of the concept
• Phase II plans – Study feasibility issues – Cost-benefit – Generate technology roadmap including pathways for development / integration with NASA missions
Acknowledgments • Universities Space Research Association NASA Institute for Advanced Concepts – Phase I funding – Phase II funding
• Charles Stark Draper Laboratory – James Bickford – Sean George