Comparison of sparse adaptive filters for underwater acoustic channel equalization/Estimation

High-rate underwater acoustic (UWA) channels often demonstrate long, time-varying and sparse impulse responses. Classical and most used adaptive algorithms such as the recursive least-squares (RLS) algorithm and the normalized least-mean-square (NLMS) algorithm do not take sparseness into account when they try to match the channel. Thus, performance improvement of these algorithms is possible. Sparse adaptive algorithms developed for acoustic echo cancellation, such as the improved proportionate normalized least-mean-square (IPNLMS) algorithm and the improved proportionate affine projection algorithm (IPAPA), have shown better performance than the NLMS algorithm without any essential cost in computational complexity. In this work, we apply IPNLMS, IPAPA, RLS and NLMS in both channel estimation and decision feedback equalization (DFE) of a short-range, shallow water acoustic link. Our results confirm the superior performance of the sparse algorithms (IPAPA being the best) when the channel becomes sparse. In addition, it is shown that both IPAPA and IPNLMS have robust performance (similar to RLS) when the channel is non-sparse.