Steady state behavior of normalized frequency domain adaptive filter in noncausal circumstances

Abstract

Normalized frequency domain LMS algorithm is attractive due to its low computational burden and fast convergence speed. However, it suffers from the deteriorated steady-state behavior in noncausal circumstances, which is always inevitable in many application scenarios. In this paper, the influence of several factors on the steady state behavior of the normalized frequency domain LMS algorithm is investigated in a typical noncausal condition -a compact active noise control system. It is very interesting to find that the noise reduction performance is even worse with the increase of the control filter length.