Cost and performance comparisons of several implementations of adaptive recursive filters

A dual-delay-line adaptive recursive filter is compared to a standard non-recursive adaptive filter with attention to performance and cost considerations. The noise cancellation problem is used to evaluate the Stearns-White recursive gradient algorithm, the Fientuch gradient approximation algorithm, and a new algorithm based on the Gitlin recursive-like echo cancellation structure. While the Stearns-White algorithm is found superior to the other algorithms in terms of performance, the increased order available for the same cost in the alternate algorithms is favorable to the simple non-recursive filter.