Adaptive multiuser detection for underwater acoustical channels

An underwater acoustic local area network (ALAN) provides multipoint-to-point telemetry between many high-rate, ocean-bottom sensors and a central, surface-deployed receiver in the 10-30 kHz vertical acoustical channel. Ocean-bottom modems initiate the transmission process by requesting data channel time slots via a common narrow-band request channel. Request packets overlap in time and frequency in this channel, and the throughput and average transmission delay rely heavily on the successful resolution of the request packet collisions. This paper presents the design, analysis, and experimental demonstration of a request channel receiver capable of resolving collisions between several asynchronous and cochannel packets. The receiver algorithm differs from standard capture schemes (by demodulating the data from both strong and weak transmitters), conventional spread-spectrum receivers (by overcoming the near-far problem), and existing multiple-access demodulation techniques (by adapting to the number of interfering signals, and the unknown phase, Doppler, amplitude, and timing of each signal in the collision). The receiver demodulates the collided packets by decision-directed techniques through a novel method of estimating the interference for each user which minimizes error propagation due to inaccurate tentative decisions. An inwater experiment illustrates that this technique is extremely desirable for collision resolution in underwater acoustic local area networks, and also for underwater autonomous vehicles with both sidescan sonar as well as acoustic telemetry links