Distributed Bias Compensation Algorithms And Its Application In Sparse System

In this paper,I researched adaptive algorithms and their implementation in distributed networks.In this thesis,I introduced the basic knowledge on adaptive filters,sparse system and distributed networks.I mainly focus on bias-compensation least mean squares(BCLMS)algorithms,sparse BCLMS algorithms and diffusion LMS algorithms.Based on BCLMS algorithm,and input noise variance estimation method using for-ward/backward prediction,I propose a new estimation method for white noise and two new bias-compensation algorithms.The proposed BCLMS-I and BCLMS-II algorithms have bet-ter mean square deviation(MSD)performances;I have derived the steady-state MSD of bias-compensation LMS algorithms under wide-sense stationary Gaussian white noise.Based on?₀norm and?₁norm sparse LMS algorithms,I proposed bias-compensation reweighted zero attractor LMS(BCRZA-LMS)algorithm.Based on distributed LMS algorithms,I have implemented diffusion bias-compensation LMS algorithms for sparse system(diffusion BCRZA-LMS).