Error and Convergence Properties of the Fast Affine Projection and Least Mean Square Echo Cancellation Algorithms
The fast affine projection (FAP) algorithm is a relatively new type of adaptive filter which is reported to have complexity comparable with the least mean square (LMS) algorithm but with performance matching that of recursive least squares (RLS) techniques. This paper compares the error performance of the FAP and LMS algorithms when applied to the task of acoustic echo cancellation using both real and simulated data. The results confirm that the convergence of the FAP algorithm is faster than that of LMS but that care must be taken with respect to the stability of the FAP algorithm which can diverge with certain types of input signal or with unsuitable update parameters.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.