Distributed parameter estimation in sensor networks based on stochastic approximation

A distributed stochastic approximation algorithm with expanding truncations (DSAAWET) is proposed to estimate the unknown parameter linearly appearing in the sensor networks. Each agent updates its local estimate by averaging its neighbors' estimates with weights and by processing its local current observation. The estimates are shown to converge to the true parameter with probability one. A numerical example is given to demonstrate the obtained theoretic result.