Error Coded Affine Projection-Like Algorithm with Evolving Order and Variable Resolution for Acoustic Echo Cancellation

Adaptive filters used in acoustic echo canceler (AEC) systems are composed of a large number of taps, resulting in high computational complexity systems and slow convergence. To overcome this problem, there is a need for an algorithm that provides a better convergence rate. The error coded affine projection-like (ECAP-L) algorithm with evolving order is a novel algorithm that presents good convergence characteristics and a lower computational burden compared with those of the algorithms based on affine projections. Nevertheless, such algorithm has only been applied in active noise control systems. In this work, we evaluate for the first time the performance of the (ECAP-L) algorithm with evolving order for an AEC system, furthermore, we propose a strategy to adjust dynamically the resolution. The simulation results show the efficiency of the algorithm in terms of good convergence rate, lower final residual error and a great reduction in computational cost.