Fast estimation of array shape and direction of arrival using sparse Bayesian learning for manoeuvring towed line array

Abstract

The sparse Bayesian learning (SBL) algorithm has demonstrated its advantage in the direction of arrival (DOA) estimation. However, it requires a lot of computational cost to iteratively estimate the SBL hyperparameters from measurement data. This paper focuses on fast estimating the array shape and DOAs using SBL for a short towed line array (TLA) during manoeuvring. A parabolic model is utilised to describe the bent TLA whose bow as a hyperparameter is estimated in the SBL iterative process. Then, the basis vector pruning strategy is considered in the iteration to reduce computational cost by neglecting the impossible directions of signal presence. The converged speed of the joint estimation algorithm is further improved by approximately calculating the posterior probability density with the message passing approach. The effectiveness of the optimising joint estimation algorithm is verified using the experimental results from South China Sea.