A unified approach to steady-state performance analysis of adaptive filters without using the independence assumptions

The independence assumptions are widely used conditions in the performance analysis of adaptive filters. Although not valid in general, because of the tapped-delay-line structure of the regression data in most filter implementations, its value lies in the simplifications, it introduces into the analysis. Another approach to study the performance of adaptive filters without using the independence assumptions, is to rely on averaging analysis. In this paper, we present a unified approach to study the steady-state performance of a family of affine projection and data-reusing adaptive algorithms based on the theory of averaging analysis and energy conservation relation without using the independence assumptions and assume specific models for the regression data. Finally, we provide several simulations results to evaluate the steady-state performance of a family of affine projection and data-reusing adaptive algorithms with and without using the independence assumptions.