Robust communications in uncertain shallow water acoustic channels

Underwater acoustic communications systems need to cope with multipath, time varying propagation conditions. In this paper, Maximum-Likelihood (ML) estimators for recovering the transmitted signal under three different models for shallow water acoustic communications are presented. The models use different levels of knowledge on the propagation conditions, resulting in estimation methods with different levels of robustness or sensitivity to channel mismatch. In addition, a Constant Modulus (CM) algorithm have been used in order to handle CM signals. The three estimators have been compared for Gaussian and BPSK signals. The results show that in the presence of channel mismatch, using robust estimators can significantly improve the estimator performance.