AN AFFINE PROJECTION ALGORITHM WITH VARIABLE PROJECTION ORDER BASED ON MEAN-SQUARE DEVIATION ANALYSIS

摘要

We present an affine projection algorithm (APA) that automatically determines projection order in order to improve convergence performance. Determination of the projection order is achieved by mean-square deviation (MSD) analysis. The projection order of the proposed algorithm varies in accordance with the steady-state MSD. Experimental results indicate that the proposed algorithm has a faster convergence rate and a smaller misalignment than the existing affine projection algorithms with variable projection order.