Experimental evaluation of CE-DFE driven by NNCLMS algorithm for acoustic communication

Due to the sparsity that contained in underwater acoustic (UWA) channels, in recent years tremendous attention has been drawn in designing sparse channel estimation algorithms for improve the performance of channel equalizer, which is driven by channel estimator. However, experimental evaluation and comparison is limited regarding enhancing the performance of different norm constraint sparse channel estimation algorithms for driving the equalizer with a series of continuous UWA communication experiments on a time scale of hour. As shallow water acoustic channels pose significant difficulties such as multipath, time-space selectivity, frequency-dependent noise, and Doppler shift, this paper presents the experimental evaluation and comparison of our previously proposed non-uniform norm constraint least mean square (NNCLMS) algorithm with the classic l₀-norm and/l₁-norm LMS algorithm in the context of a shallow water acoustic communication system for driving a channel estimation based decision feed-back (CE-DFE). We reported and discussed experimental results obtained in shallow water channels in terms of the NNCLMS algorithm's sparse exploitation performance and the classic norm constraint type sparse algorithms.