distributed sensor networks - boston university · distributed sensor networks angela m. mielke1,...
TRANSCRIPT
UNCLASSIFIED LA-UR-06-1509
1
Distributed Sensor Networks
Angela M. Mielke1, Sean M. Brennan1, Mark C. Smith1, Janette R. Frigo1, Diana Jackson1, David C. Torney2, James T. Rutledge3, Douglas M. Alde3,
Matthew S. Nassar4, Stephan J. Eidenbenz4, James Horey5, Arthur B. Maccabe5, Violet Syrotiuk6, Charles Colbourn6
1ISR-3 - Space Data Systems, 2T-10 - Theoretical Biology and Biophysics,3EES-11 - Geophysics, 4CCS-5 - Discrete Simulations Sciences,
5The University of New Mexico,6Arizona State University
UNCLASSIFIED LA-UR-06-1509
2
What are DSN’s and why do we care?
• DSN’s are self-organizing networks of sensors with computation and communication capabilities.
• DSN’s are ideally designed for use in application areas such as:
Remote sensingPersistent surveillanceCovert operations
Event
UNCLASSIFIED LA-UR-06-1509
3
The Classical DSN approach employs a Star Topology.
Advantages:• Simple network architecture
• Raw data available at the Central Processing Station
Disadvantages:• Catastrophic single-point failures
• Potential long-range communication
• Large quantities of transmitted data
• Difficult network scale-up
Central Processing StationEvent
UNCLASSIFIED LA-UR-06-1509
4
Collective Computation is an alternative network architecture providing in-situ data processing.
Advantages:• Network size scalable
• Fault-tolerance
• Conclusions to user node(s)
End-user(s) receivingtimely simple conclusion
Sensors inter-communicatefor collective computation
Disadvantages:• Complicated node design
• Complicated network design
UNCLASSIFIED LA-UR-06-1509
5
The Distributed Sensors Simulator is a simulation infrastructure for network prototyping.
• Developed due to shortcomings in other commonly available sensor network simulators.
• Provides the infrastructure for environmental simulations and wireless network simulations.
• Developed in conjunction with the University of New Mexico.
• DOE approval for Open Source Distribution of the Distributed Sensors Simulator (DSS), Version 0.8 - was obtained on April 14, 2004.
UNCLASSIFIED LA-UR-06-1509
6
Let’s step through a simulation example…
• Sensor node placement is highlighted here.
UNCLASSIFIED LA-UR-06-1509
12
Simulator Developments - DSS to NSIM • Improved framework for environmental
phenomenology – DSS included basic information about source
propagation and sensor detections.
– NSIM will include physical dynamics and obstacle models allowing for direct, reflected and refracted propagation paths as applicable.
• Improved wireless simulation characteristics– DSS relied on simple transmission models.
– NSIM will utilize probabilistic models that include obstructions and collisions.
• NSIM is an Emulation engine enabling simple compiler changes for code ports from simulator to actual hardware.
UNCLASSIFIED LA-UR-06-1509
13
Commercial Wireless Sensor Networks
• Mesh Network• Self-organizing• Self-healing• Encryption• Spread Spectrum• Multi-hopping• Typically Star Topology• Provide Sensor Network
Backbone• Manufactures include:
DUST, Crossbow, Ember, Millenial
Crossbow Mica2/Mica2dot
Millennial Net I-bean
Ember Modules
UNCLASSIFIED LA-UR-06-1509
14
The Sponsor Challenge - Application Development
• Demonstrate simulation results• Evaluate commercial hardware• Investigate two alternative wireless network
approachesMote approach - small, low power devices, advanced wireless capabilities, minimal data processing capabilities, small operating system (TinyOS)PDA approach - moderate-sized and moderate-powered devices, 802.11b wireless capability, significant data processing capabilities, general purpose operating system (Linux)
UNCLASSIFIED LA-UR-06-1509
15
Target Application - Roadway Monitoring for Radioactive Materials
• Radioactive source detection
• Focus of this application is on the mitigation of the potential for terrorist usage of RDDs.
• Mote network detects vehicle presence.
• Motes queue the PDA network to begin coherent addition techniques for material detection.
• Radiation background updates and data evaluation occur on the PDAs.
• Source detection is exfiltrated to a command-and-control console.
UNCLASSIFIED LA-UR-06-1509
16
Vehicle Detection Mote Network
• Deployed network of Mica2 Crossbow motes
• Deployed along a roadway
• On-board sensors on the Crossbow MTS310 multi-sensor board included:
2-axis accelerometer
2-axis magnetometer
Light, temperature
Microphone, sounder
• Self-organizing
• Multi-hop capability
• Magnetometers used for vehicle detection
UNCLASSIFIED LA-UR-06-1509
17
PDA Network
• PDA network utilized the Linux-based Sharp Zaurus SL-5600.
• In field testing -Range limit is approximately 450’UNM delivered software for multi-hop routing on the PDAs.
• PDA network connects to the radiation detectors as well as to the Crossbow gateways thus bridging the two networks.
• Radiation detectors are off-the-shelf serial Geiger-Mueller tubes with a 1.75 inch active window and a length of 0.5 inches.
UNCLASSIFIED LA-UR-06-1509
18
Challenges Encountered in Initial Field Experiments
• Network Timing
• Accelerometer Insensitivity
• RS232 Non-compliance of the PDA Devices
• Mote Failures
• Inconsistent Sensor Readings
• Transmission Range Inconsistencies
• Battery Lifetime
• PDA Power Consumption
UNCLASSIFIED LA-UR-06-1509
19
Research Areas under Current Investigation
• Sensor characterization and exploitation through testbed development
• Transmission protocol characterization and refinement
• Algorithm development for event detection– Vehicle classification using geophone data
– Hyperspectral Imaging
• Power consumption and management studies
• Network scale-up
• Network and device security (ie. spoofing and tamper resistance)
• Improved radiation detection techniques
UNCLASSIFIED LA-UR-06-1509
20
Transmission Protocol Characterization Studies
• Development of a non-invasive monitoring system utilizing the Crossbow Stargates
• Parametric Routing Protocols developed in CCS-5 (Destination-Attractor and Directed-Transmission) are being evaluated against simulation results.
• Additional protocols will be tested– WSN, DSDV, AODV
• Protocols are implemented in TinyOSon the Crossbow Mica2 motes
• Current efforts focus on validation of the experimental results
– Collisions, Network Stability, Timing, Reception/Transmission
UNCLASSIFIED LA-UR-06-1509
21
Power Consumption/Energy Harvesting
• Battery lifetime studies for different battery types and different operational conditions
• RF power versus transmission distance studies
• Solar recharge capabilities
• Outfitting the Crossbow Stargates for an off-grid configuration
UNCLASSIFIED LA-UR-06-1509
22
Research Areas for Future Investigation• Long-range data exfiltration
• Expansion of university and industrial collaborations
• Expansion of application tools and sponsors
Air surveillance
Border security
Event monitoring
Internal security perimeters
Methods of energy harvesting for long-term deployments
System-wide power management
Incorporation of knowledge discovery techniques for data management
UNCLASSIFIED LA-UR-06-1509
23
Project Strengths
Ability to initially design and evaluate systems in simulation
Development of persistent surveillance applications utilizing COTS technology
Implementation of networks utilizing complicated sensors
Collective ComputationMoving from Data Collection to Event Detection
UNCLASSIFIED LA-UR-06-1509
24
More Information?
www.lanl.gov/dsnAngela Mielke(505) [email protected]