Bias-compensated Sparse RLS Algorithms Over Distributed Networks

In this paper, we propose a bias-compensated method based on the L1-RLS algorithm and the diffusion L1-RLS algorithm for sparse system identification. Our proposed algorithms improve the estimation accuracy of traditional L1-RLS when the input data is corrupted by input noises. Furthermore, we give simulation results to verify that proposed algorithms have better estimation accuracy than other sparse RLS algorithms without bias compensation, it also proves that results are unbiased under input noises.