Double-talk robust fast converging algorithms for network echo cancellation

There is a need for echo cancelers for echo paths with long impulse respons es (greater than or equal to 64 ms). This in turn creates a need for more r apidly converging algorithms in order to meet the specifications for networ k echo cancelers. Faster convergence, however, in general implies a higher sensitivity to near-end disturbances, especially "double-talk." Recently, a fast converging algorithm has been proposed called proportionate normalize d least mean squares (PNLMS) algorithm, This algorithm exploits the sparsen ess of the echo path and has the advantage that no detection of active coef ficients is needed. In this paper we propose a method for making the PNLMS algorithm more robust against double-talk. The slower divergence rate of th ese algorithms in combination with a standard Geigel double-talk detector i mproves the performance of a network echo canceler considerably during doub le-talk. The principle is based on a scaled nonlinearity which is applied t o the residual error signal. This results in the robust PNLMS algorithm whi ch diverges much slower than PNLMS and standard NLMS. Tradeoff between conv ergence and divergence rate is easily adjusted with one parameter and the a dded complexity is about seven instructions per sample which is less than 0 .3% of the total Load of a PNLMS algorithm with 512 filter coefficients. A generalization of the robust PNLMS algorithm to a robust proportionate affi ne projection algorithm (APA) is also presented. It converges very fast, an d unlike PNLMS, is not as dependent on the assumption of a sparse echo path response, The complexity of the robust proportionate APA of order two is r oughly the same as that of PNLMS.