Underwater acoustic channel estimation based on variational Bayesian inference under reliable acoustic path in deep sea

In the deep-sea reliable acoustic path (RAP) area, the underwater acoustic channel exhibits obvious long delay, clustered and sparse characteristics. Aiming at the above characteristics of the RAP channel, a channel estimation method based on vector approximate message passing (VAMP) and variational Bayesian inference (VBI) is developed under the framework of Turbo iterative equalization. Firstly, the RAP channel is modeled as a series of sub-channels of different clusters, and each sub-channel of each cluster is estimated iteratively using VBI. Then, to address the problem of high computational complexity of VBI, VAMP is embedded in the VBI framework to estimate the posterior distribution of each cluster channel with low complexity. Finally, to address the problem of strong time-varying nature of the reflected phonon channel, a joint estimation method of the channel and symbol based on the time correlation of the direct sound is proposed under the VBI framework. The proposed method is verified using deep-sea experimental data collected in the South China Sea. The results show that the proposed method has better channel estimation performance and lower computational complexity under the deep-sea RAP channel.