Distributed Stochastic Approximation Algorithm for Time-Varying Regression Function Over Network

In this paper we consider the distributed root-tracking problem for a sum of time-varying regression functions over a network, where each agent only has local observation and aims at finding the root of the global regression function. Since the regression functions are time-varying, we have to track the changing roots. A distributed stochastic approximation algorithm is introducted for solving the problem and that the consensus and convergence of the estimates are proved.