Error rate improvement in underwater MIMO communications using sparse partial response equalization

Point-to-point links using multiple transmitters and receivers or MIMO (multiple input, multiple output) configurations and associated signal processing at the receiver, can provide significant improvements in both data rate and reliability and is a promising technology for enhancing communications in the band-limited and highly-dynamic underwater acoustic (UWA) channel. Although the underwater channel impulse response extent can span tens to hundreds of symbols, it is generally sparse in nature and well suited for sparse partial response equalization (sPRE). This equalization scheme, which is not restricted to MIMO configurations, does not attempt to suppress intersymbol interference completely, rather it retains residual ISI in a controlled manner. This is accomplished by setting a sparse residual impulse response target generally similar in magnitude and time to the dominant, yet also sparse, arrivals within the actual channel impulse response. The resultant partial response equalizer is followed by a complexity-reduction detection scheme known as "belief propagation (BP)" which is an alternative to the optimal Viterbi or MAP (maximum aposteriori probability) detector. The complexity of the optimal schemes grows exponentially with the total number of taps, regardless of structure; whereas the complexity of BP grows exponentially only with the non-zero taps. Thus the entire receiver structure, sPRE followed by BP, is suitable for the long, sparse channels since it allows more efficient exploitation of the channel structure. The proposed symbol recovery scheme was applied to data collected during a comprehensive multi-institution MIMO Experiment conducted within the Makai Experiment in 2005 off the northwest coast of Kauai, Hawaii. We have demonstrated a reduction in error rates over receiver algorithms using conventional decision feedback equalization techniques due to increased multipath diversity.