Sparse equalization for real-time digital underwater acoustic communications


Due to the very long reverberation time of many ocean channels, the size of the adaptive filters required for conventional equalization becomes large, rendering the computational complexity of the adaptive receiver unacceptable for many cases of practical interest. To overcome this problem we exploit the natural sparseness of the reverberation pattern. By focusing only on those intervals which contain a significant portion of the signal energy, the sparse equalization method provides data detection using a minimum complexity adaptive receiver subject to an upper bound on the signal estimation error. Experimental results demonstrate an order of magnitude reduction in computational complexity with a negligible loss in performance.