Optimum design of the hard-switching dual sign algorithm for adaptive echo cancellation

Abstract

In this paper, a practical design method is presented to optimize the hard-switching dual si n algorithm (HS-DSA), formed by the sign algorithm ( S i ) with a time switch for the step size, for the fastest convergence in adaptive echo cancellation. For the optimized HS-DSA, the two step sizes and the switching instant are optimally chosen to yield the most rapid convergence under an initial and desired mean-square error. As compared to previously known variants of the SA, the optimized HS-DSA gains improvements in convergence performance and retains the advantage of hardware simplicity. Computer simulations are employed to show the correctness and effectiveness of the algorithm.1) The transmitted signal a(k) and the received far-end signal u(k) are zero-mean widesense stationary mutually independent stochastic processes with variances ua2 and uu2, respectively. Moreover, a(k) is a white process and u(k) is Gaussian.2) The probability density function of E(k)=e(k)-Z(k) is Gaussian with variance uE2(k). This is a reasonable assumption if the impulse response of the echo path is sufficiently long. 0788 329.1.1