Algorithms for Distributed Sensor Networks

A distributed sensor network is many (100-10000) autonomous sensor nodes spread out over a large area. Each node is equipped with a processor, mission-specific sensors, and short-range communications. Local interactions between sensor nodes allow them to reach global conclusions from their data. This work develops algorithms that allow:

· The group to establish robust spatial patterns of messages
· The group to develop a communications network by dividing tasks among themselves
· Each mote to determine its position in physical space based on their location in the network topology
· Each mote to determine if it is on the boundaries of the network by measuring global constants through local interactions
· The group to project the path of a target moving through the network

To verify our algorithms, we have constructed two simulation environments. One is based in software and allows for very rapid proof-of-concept development. The other is a hardware version that still allows rapid development, yet provides all the problems of real hardware for a high fidelity simulation.

Paper located here.

McLurkin J
University of California, Berkeley