Bias-Compensated MCSE Algorithm for Widely Linear Complex-Valued Adaptive Filtering with Noisy Inputs

In this paper, based on minimum complex Shannon entropy (MCSE), a novel widely linear complex-valued estimated-input MCSE (WLC-EIMCSE) algorithm is proposed, which can not only make unbiased estimation in the environment where the input signal has noise, but also show superiority over WLC-EILMS and WLC-EIMCCC in the non-Gaussian noise whose output noise is bimodal Gaussian distribution with non-zero mean. The convergence of the proposed algorithm is analyzed, and the simulation of system identification verifies its superiority.