Enhanced adaptive equalization for MIMO underwater acoustic communications

An enhanced normalized least mean squares (NLMS) adaptive equalization scheme is proposed for single-carrier underwater acoustic (UWA) communications. The enhancement is achieved via two techniques: the sparse adaptation (SA) technique and the data reuse (DR) technique. The SA technique speeds up the convergence and improves the performance of the adaptive equalization, by taking advantage of the inherent sparsity of the equalizer. It is implemented as the selective zero-attracting NLMS (SZA-NLMS) algorithm, developed by introducing a range of attraction for the existing zero-attracting NLMS (ZA-NLMS) algorithm. Compared with the ZA-NLMS, the SZA-NLMS incurs a lower complexity and achieves a better performance. The DR technique effectively prolongs the training length and significantly reduces the training overhead. Attributed to the DR technique, a high transmission efficiency is achieved even for a block transmission. The proposed adaptive equalization is verified by the real data collected in an at-sea multiple-input multiple-output (MIMO) underwater acoustic communication trial. The experimental results show it considerably outperforms the standard NLMS adaptive equalization, especially with a low DR number.