Abstract:

In order to improve the detection performance of multiple-input multiple-output (MIMO) sonar in Gaussian noise and reverb background, an adaptive detection method incorporating priori knowledge is proposed. Considering an interference scenario in which Gaussian noise and reverb coexist, firstly, Bayesian theory is introduced to model the unknown reverb covariance matrix as a random matrix with inverse complex Wishart distribution. Secondly, two sets of training data are jointly exploited to devise a two-step estimation method of the interference covariance matrix. Finally, interference covariance estimate is used in place of its true value and the adaptive matched filter is obtained under the Bayesian framework. The simulation results show that the proposed detection method can achieve more accurate estimation of the interference covariance matrix and has a robust detection performance when the training data is insufficient.

Key words: multiple-input multiple-output (MIMO) sonar, adaptive detection, Gaussian background, inverse complex Wishart distribution, Bayesian framework