Regularized Affine Projection Algorithm for Stereophonic Acoustic Echo Cancellation

Stereophonic acoustic echo cancellation (SAEC) requires the estimation of four unknown impulse responses by using four corresponding adaptive filters. The affine projection algorithm (APA) is an attractive solution, which has improved tracking speeds with respect to the classical alternatives. It can be employed by reinterpreting the SAEC setup and using a single complex valued adaptive filter. In this paper, we propose a regularization technique for the APA designed to mitigate the effects of double-talk situations on the adaptive process.