Block Sparse Bayesian Learning Using Weighted Laplace Prior for Super-Resolution Estimation of Multi-Path Parameters
Qiyan Song, Xiaochuan Ma · Global Oceans 2020: Singapore – U.S. Gulf Coast · 2020
In order to overcome the conventional methods of time delay estimation suffering from low resolution in a multi-path environment, in this paper, we propose block sparse Bayesian learning using weighted Laplace prior (WL-BSBL). We impose the weighted Laplace prior on the TOA. And a greedy iterative strategy is proposed to solve the WL-BSBL model. Furthermore, to improve the computation efficiency, we incorporate block idea into the WL-BSBL model by dividing the potential TOA time domain into connected blocks based on the first few iterations result. WL-BSBL model can take advantage of the active sonar receiving data to estimate the number, time delay, and amplitude of multi-path accurately. The advantages of WL-BSBL include low computation complexity, super-resolution. The simulation results show that WL-BSBL runs fast and retains good performance in low signal-to-noise ratio (SNR) environment.