A Novel Affine Projection Algorithm Based on Variable Projection Order

Abstract

As affine projection adaptive algorithm performance relates to the projection order, it is necessary to analyze the projection order. When the projection order increases, a high convergence speed will be acquired. Whereas, it also brings the huge computational cost and the large mean square error (MSE). It seems advisable to reduce the projection order when it is necessary. Therefore, a novel method which dynamically adjusts the projection order is presented. It changes the projection order by different convergence state. The goal is to find the optimized projection order which can balance the steady-state and convergence performance. Experiments demonstrate that the proposed algorithm equips good convergence ability and low MSE.