A variable step-size proportionate affine projection algorithm for network echo cancellation

For a network canceler, whose input is mainly speech, the proportionate affine projection algorithm (PAPA) is expected to present faster convergence speed than the existing proportionate NLMS algorithms. However, the performance criteria of fast convergence speed conflicts with low steady-state misalignment when a constant step-size parameter is applied. In this article we introduce a variable step-size technique into the PAPA in order to achieve both fast convergence speed and low steady-state misalignment. Simulation results show that the proposed approach can greatly improve the steady-state misalignment without sacrificing the fast convergence speed of the PAPA.