Estimation of a shallow water channel impulse response via an MCMC method with application to an oceanic acoustic tomography experiment

The impulse response of shallow water channels that have to be measured during Oceanic Acoustic Tomography experiments consists of many paths with decreasing amplitudes. The estimation of such channels must cope with several problems; in particular, the transmitter transducer strongly distords the signal. In addition, close paths are generally not resolved at the output of the matched filter. Further processing is therefore required. In this paper, we present a method for estimating the impulse response of a shallow water channel. We first propose to estimate the distorded waveform by means of an iterative procedure based on a Matching Pursuit algorithm. Second, to achieve precise estimation, we model the channel impulse response by means of a Bernoulli-Gaussian process. Then, an estimator is computed from simulated samples drawn according to the resulting a posteriori log-likelihood by means of an MCMC method. These methods have been applied to data measured during a tomographic experiment made by the SHOM: the PASSTIME experiment.