New normalized least mean M-estimate algorithm for stereophonic acoustic echo cancellation

This paper proposes a new robust adaptive filtering algorithm for stereophonic acoustic echo cancellation in impulsive noise environment. The well-known conventional two channel normalized least mean square (TCNLMS) algorithm performs very poorly when either the input or the desired signal is corrupted by impulsive noise. By employing robust M-estimate technique, a new two channel normalized least mean M-estimate (TCNLMM) algorithm is proposed, which only imposes mild additional computational overhead but achieves improved robustness to impulsive noises over its NLMS counterpart. Experiments are also conducted to verify the efficiency of the new algorithm.