Using normal mode channel structure for narrow band underwater communications in shallow water

Multipath and high temporal and spatial variability of the propagation environment causes severe signal degradation in shallow water acoustic digital communications. Among the many solutions that have been proposed the most known is adaptive equalisation where cyclic training signals are used to adapt the equaliser to the variability of the acoustic channel. When the channel is rapidly changing, equaliser coefficients are frequently adapting and the effective transmitting rate rapidly decreases. Another approach consists in using a priori information obtained from acoustic propagation models. These models can give a deterministic estimate of the true channel impulse response that can be used to detect the transmitted signals. In practice, the use of deterministic acoustic models is mainly dependent of the accuracy of the input environmental parameters. As a first step, this paper presents an exhaustive study of the signal detection sensitivity to model parameters mismatch. The scenario used is composed of a 100 m depth water column with range dependent characteristics. The water column is located over a 10 m thick sediment layer with variable properties. Source-receiver communication is made over a variable distance between 500 and 600 m with the source near the bottom and the receiver near the surface. The communication signals are narrow band (1.5 kHz) pulse amplitude modulated with a carrier frequency of 15 kHz, and the detector is based on the maximum-likelihood sequence detector (MLSD).