Efficient and Fast Processing of Large Array Signal Detection in Underwater Acoustic Communications

The deployment of a large-scale array of hydrophones in underwater acoustic (UWA) communications brings numerous benefits, in terms of high spectrum and energy efficiency, and the high data rate communications. However, along with the merits, large array signal processing is known to be computationally costly and long processing delay required. Even with linear detection methods such as minimum mean-square error (MMSE) based schemes, the computational complexity is still considerable as the matrix inversion operations are involved. With Gauss-Seidel method, the matrix inversion operations are avoided, while the iterative processing achieves comparable system performance to the MMSE-based schemes. However, Gauss-Seidel method introduces successive data detection, causing significant processing delay. Meanwhile, the successive detection structure is inefficient in hardware implementation. In this paper, we propose a block Gauss-Seidel method for large array signal detection in UWA communications. In the proposed scheme, Gauss-Seidel method is performed on a set of small size block matrices, and the processing on each block can be parallelized. As a result, the total processing delay can be greatly reduced. Moreover, the parallel processing structure is quite efficient for hardware implementation. We also utilize the UWA channel model developed in recent work to investigate the performance of the proposed scheme, and the results are promising.