Robust Bias-Compensated Subband Adaptive Filtering

Fan Song, Fuyi Huang, Sheng Zhang, Haiqiang Chen, Pengwei Wen

Abstract: This brief proposes a robust bias-compensated subband adaptive filtering (R-BC-SAF) algorithm for noisy input and large outliers. We first performs the maximum a posteriori estimate subject to a constraint on the squared norm of the weight vector difference in order to obtain a variable step-size method in update and combat large outliers. An unbiasedness criterion is then employed to insert a bias compensation term in the update to reduce the estimation bias caused by noisy input. Besides, estimators are designed for the input and output noise variances. Simulations in system identification and echo cancellation scenarios for different input signals demonstrate that our method outperforms the comparing algorithms.