On Distributed Optimization Using Peer-to-Peer Communications in Wireless Sensor Networks

We describe and evaluate a suite of distributed and computationally efficient algorithms for solving a class of convex optimization problems in wireless sensor networks. The problem class has wide applications in estimation, detection, localization, coordination and resource-sharing. We focus on peer-to-peer algorithms where nodes only exchange data with their immediate neighbors, and consider three distinct alternatives: a dual-based broadcast algorithm, a novel stochastic unicast algorithm, and a linear broadcast algorithm tailored for least-squares problems. We implement the algorithms in the network simulator NS2 and present extensive simulation results for random topologies.