11-04-2012, 01:01 PM
Advances in Sensor Networking:An Algorithmic Approach
Wireless-Sensor-Networks.pdf (Size: 5.85 MB / Downloads: 138)
Mica2 Architecture: Processor Board
Atmel Atmega128L processor (8 MHz) – Micro-controller
Program flash memory: 128 KB
EEPROM: 4 KB
Serial flash memory: 512 KB
(non-volatile storage)
Chipcon CC1000 radio transceiver
– Bidirectional communication: 38.4 Kbps
– FSK modulated radio
– Frequency band
Mica2 Programming
TinyOS: Components based OS designed for sensor
networks with limited resources
Supports modularity and event-based programming
Programming language: NesC
Provides framework for concurrency: never poll and
never block
TinyOS executes only one program consisting of
selected system components and custom components
needed for a single application
Forest detection WSN Applications: Civilian
- fire / Earthquake detection, Flood control
- Temperature monitoring and vent control in buildings
- Environment, Habitat, Agriculture monitoring
- Traffic / Vehicle monitoring and control
- Structural health monitoring
- Smart homes and Health care
- Bio and Chemical sensors detection
WSN Design Paradigms
Long-lived, un-tethered and unattended systems
– Low-duty cycle operation with bounded latency
Distributed, Self-Organization
– Scalability and robustness
– Self-healing, adaptive to unpredictable environments (and failures)
– Short range communication energy efficiency
In-network processing
–Exploit localized computation (near data) to reduce communication
– Exploit temporal and spatial redundancy
Aggregation, distributed source coding, and compression
Data-centric and attribute-based addressing
– Decouple identify for routing and aggregation
S. K. Das
Application-specific network protocol design
Modeling Compromises: Epidemic Defense
Premise: Node compromises
–Capture node deployment, key distribution, topology
–Outbreak possible unless controlled
Objectives:
–Construct a model and analyze the spread of node
compromises in WSNs based on Epidemic Theory
–Characterize outbreak transition point of compromise process
–Study the impact of infectivity duration of a compromised
node on the process
–Capture the time dynamics of the spread
–Identify critical parameters to prevent outbreaks
Trust in WSNs: Motivation
In WSNs, data are noisy uncertain) and unreliable
Redundancy from highly dense deployed sensors may
provide “side information” for data fusion
- e.g., How to exploit redundancy for abnormality detectection?
Precise fusion is difficult with multiple questionable data
–How to represent uncertainty in the aggregation result?
e.g., Is there any measure to interpret the ignorance in fusion?
– How to quantify uncertainty when fusion results are propagated?
e.g., How to evaluate a hierarchical, bottom-up fusion result?