Joint Estimation and Gossip Averaging for Sensor Network Applications

This note presents an efficient distributed approach for computing a spatial average of parameters estimated by sensors in a wireless network. The most intuitive approach would rely on a two-step procedure. First, the nodes would estimate the local quantities, and second a distributed process would average these estimates over the network. Instead, the proposed algorithm combines both processes to foster the convergence while fulfilling the usual wireless sensor network requirements: simplicity, low memory/CPU usage, and asynchronicity.