Simple variable tap‐length algorithm for high‐noise environment

Abstract

In the variable tap-length algorithm, fast-convergence rate and small steady-state fluctuation of the tap-length are two important criteria to evaluate performance of the algorithm. Furthermore, the simplicity and robustness to the noise are the key points for the algorithm to be applied in practise. Along this line, a very simple variable tap-length algorithm is processed where the adaptation rule for fractional tap-length is modified by incorporating strategy of limiting the amplitude of difference between squared output error and squared segmented error. Only at a slight cost of computational complexity, the proposed algorithm can achieve both small steady-state fluctuation and fast-convergence rate of the tap-length under a high stochastic Gaussian noise condition (signal-to-noise ratio = 0 dB) or a deterministic impulsive noise condition.