A method to adjust regularization parameter of fast affine projection algorithm

Fast affine projection (FAP) algorithm is widely used recently for its NLMS like computational complexity and RLS like convergence rate. The regularization parameter in the FAP algorithm is used to guarantee a well behaved inverse of the covariance matrix, and it plays an important role in the performance of the FAP algorithm. However, the fixed value of the regularization parameter in the conventional FAP algorithm degrades the performance of the system when the signal and the acoustic environment have noticeable variations. In this paper, an efficient method is proposed to adjust the regularization parameter. Compared to the direct adjusting methods, it costs less computational burden and can be activated more frequently according to different signal and environment status