Robust localization and identification method of moving sound sources based on worst-case performance optimization

Abstract

Based on the passive synthetic aperture principle, a new robust high-resolution focused array signal processing method of moving sound source localization and identification is proposed in this paper. By means of the integrated optimization, this method generates the coordinates of a virtual array and the data matrix through the vector maximum likelihood focused algorithm, then utilizes the sparse virtual array focused algorithm based on the worst-case performance optimization to obtain the robust high-resolution localization and recognition effects. The theory and the simulation analysis show that this method is applicable to the complex experimental situations such as non-uniform motion and tipsy array, and the focused spatial spectrum indicates the greater dynamic range, the sharper focused peak, and the stronger ability to suppress the fluctuations of the background noise. The higher robustness and better results of this proposed method are verified in the lake experiment. Under the same experimental condition, the dynamic range of high-resolution MVDR focused algorithm is only 3.5dB, however, it can reach 50 dB by the proposed method.