Sequential analysis for underwater communications

Abstract

The state-of-the-art in high-rate, single-carrier wideband signaling for acoustic communications is represented by the decision feedback equalizer (DFE). Though often effective, its performance is far from the optimal obtained from soft decision maximum a posteriori probability (MAP) and maximum-likelihood sequence detectors (MLSD). While these algorithms have optimal performance, their complexity increases exponentially with the duration of the channel impulse response. In practice, such methods are only used for multicarrier modulation, after mode filtering or other form of channel shortening, time-spatial pre-processing equalization. The optimal joint channel estimation and data decoding algorithm is derived and analyzed. The combination of pre-processing, channel response shortening equalization, and joint channel and data recovery have shown excellent performance in shallow water acoustic communications experiments. A sequential estimation alternative to MLSD-based decoding is “almost” as effective in a probability sense, given a modest increase in signal-to-noise ratio (SNR). That approach combines very high performance with a small computational burden relative to the MLSD approach. The practical result of the paper is the investigation of the replacement of the DFE with a sequential implementation (FANO) of a likelihood sequence estimator. Comparative performance of the two using at-sea experiments in very shallow water is provided.