A simplified statistical analysis of the Affine Projection Algorithm

This paper develops a statistical analysis of the Affine Projection Algorithm (APA) for adaptive filters to theoretically calculate filter convergence behavior. Assuming that the number of tap weights of the adaptive filter is sufficiently large, we derive simple difference equations that only require matrix computations. In experiments with some examples, it is observed that simulated and theoretically calculated filter convergence curves are generally in good agreement that justifies the approximations used and shows the validity of the analysis for practical use.