Performance of channel-equalized acoustic communications in the surf zone

The Surf Zone Acoustic Telemetry Experiment (SZATE) was conducted during August and September 2000 alongside the pier at Scripps Institution of Oceanography. The goal was to characterize the relationships between the performance of acoustic communications and observable oceanographic and acoustic conditions. As part of our participation in SZATE, we designed and transmitted messages by means of both non-coherent signaling methods, and channel-equalized, coherent methods. The latter involved a 3220 bps QPSK message with accompanying channel probes. More than 20,000 such waveforms were processed and analyzed, with channel conditions ranging from benign to unusable. Channel equalization was accomplished with decision feedback equalizers based on the LMS, RLS, and fast RLS implementations. Due to rapid temporal variations in the channel impulse response, the basic LMS algorithm was generally ineffective. The RLS and fast RLS algorithm performance were comparable, with the latter providing slightly superior performance in terms of residual uncorrected channel symbol errors. Equalizer tap placement was determined from a heuristic algorithm which provided excellent results. The presence of bubbles may dominate signaling performance by blocking the channel. Bubbles are either injected into the water column from breaking waves, or are advected by along-shore currents. However, the channel is generally "open" and in such cases the percent of successful messages ranged from approximately 50% to 98%, with an average of approximately 80%. Temporal diversity is shown to substantially improve this success rate. The performance in the absence of bubbles is correlated with the temporal coherence of each of the major multipath arrivals.