Noise Constrained Incremental Least Mean Square Algorithm

Abstract

We proposed a noise constrained based distributed adaptive estimation algorithm for wireless sensor network, based on the incremental scheme. The Least Mean Square (LMS) Algorithm’s cost function is modified by using noise variance on every nodes, and noisevariance’s knowledge is used for estimation the parameter of interest. This modification result to improve convergence speed of the algorithm keeping the steady mean square error minimized. Theoretical Mean and Steady State Analysis are performed for the convergence of the algorithm and steady state mean square error. In Mean analysis the step size limit of the proposed algorithm define, and in steady state analysis the steady state mean square error define. Under different scenarios experimental results show the superiority of the proposed Noise Constrained Incremental LMS over non-constrained ILMS