A block selection adaptive delay filter algorithm for echo cancellation

Abstract

The block selection adaptive delay filter algorithm for echo cancellation presented exploits the fact that the impulse response of an echo will have active regions and dead-band regions. By identifying and tracking the active regions, short-length adaptive transversal filters can be centered about these regions and the computational effort of adjusting taps which converge to zero can be reduced. The block selection adaptive delay algorithm was applied to two unknown systems whose impulse response was linear and time invariant. In both cases, the adaptive delay filter was able to accurately identify the significant delays and the short length filter bank was able to adapt to active regions of the echo. The input for both cases was white noise simulated by a uniform random number generator with zero mean and unit variance. In the first example the filter bank contains two adaptive filters with 12 tap weights. In the second example, the filter bank contains three adaptive filters again with 12 taps for each filter. >