Fixed‐point steady‐state analysis of adaptive filters

Abstract

The steady-state performance of adaptive filters can vary significantly when they are implemented in finite precision arithmetic, which makes it vital to analyse their performance in a quantized environment. Such analyses can become difficult for adaptive algorithms with non-linear update equations. This paper develops a feedback and energy-conservation approach to the steady-state analysis of quantized adaptive algorithms that bypasses some of the difficulties encountered by traditional approaches.