Underwater acoustic channel estimation via fast Bayesian matching pursuit

Abstract

Underwater acoustic (UWA) channel estimation based on fast Bayesian matching pursuit (FBMP) is investigated in this paper. In UWA multipath channel, the positions of UWA channel taps usually obey the Bernoulli distribution, while the coefficients of UWA channel taps obey the complex-valued Gaussian distribution. Such knowledge of distribution is considered as a prior for UWA channel estimation based on the Bayesian framework. The positions of UWA channel taps are first determined. Then the coefficients of UWA channel taps are estimated. Regarding the complexities for Bayesian estimation, the fast implementation is also studied, where the stop condition for the proposed iterative algorithm is also optimized. Simulation results show that FBMP outperforms orthogonal matching pursuit (OMP). In order to achieve the same bit error rate (BER) of 0.016, FBMP can save around 4dB of signal to noise ratio (SNR) compared to OMP.