A decentralised linear programming approach to energy-efficient event detection

This paper addresses the problem of decentralized event detection in a largescale wireless sensor network (WSN), where without the coordination of any fusion center, sensor nodes autonomously make decisions through information exchange with their neighbors. We formulate the event detection problem as a linear program, and propose a heuristic decentralized linear programming (DLP) approach to solve it. Compared to the existing algorithms which solve decentralized optimization problems, the DLP algorithm requires low communication cost per iteration and shows fast convergence. Convergence property of the DLP algorithm is theoretically analyzed, and then validated through simulation results. We also implement the DLP algorithm in the structural health monitoring application and demonstrate its effectiveness in decentralized event detection.