Performance comparisons of finite linear adaptive filters

The paper sets out to review the area of linear adaptive filters, restricted to the classification of finite impluse response filters. The exact solution to this problem based on the recursive least-squares algorithm is first derived. This algorithm is then degraded to show the evolution of self-orthogonalising adaptive algorithms, and further, to the stochastic gradient search algorithms. Computer simulations are presented to compare and contrast the performance of the algorithms in terms of their convergence bahaviour. The relative complexity and numerical stability of the algorithms is then discussed. Together, these comparisons provide a comprehensive basis on which to base an informed decision on choice of algorithm for any defined application.