Broadband Aperture Extension of A Passive Sonar Array Using Multi-Dimension Autoregression Model

This paper presents a broadband aperture extension approach of a passive sonar array based on multi-dimension autoregression (AR) model, integrated with failure element compensation. The multi-dimension AR model is proposed to build relations between broadband complex signals of array elements in frequency domain and predict the virtual element signals for wideband aperture extension, instead of the traditional time-domain AR model suitable for narrow-band aperture extension. In addition, the broken element compensation algorithm is designed to recover the failure array elements to ensure the feasibility and accuracy of AR modeling and virtual element prediction. Experimental studies verify the effectiveness of the presented approach for broadband array aperture extension.