Distributed Estimation of Channel Gains in Sensor Networks

Channel estimation algorithms are proposed for the initialization phase of a wireless sensor network, during which the nodes are still unaware of their neighbors or channel qualities. Both Expectation Propagation (EP) and diffusion Least Mean Squares (LMS) algorithms are applied to this problem of channel estimation while the network employs a random duty cycling strategy to conserve power. Simulations results which show EP giving better performance are given for both algorithms for a small network.