Acoustic source identification using an off-grid and sparsity-based method for sound field reconstruction

Abstract

Acoustic source identification based on the microphone array is an attractive technique, and the effective identification of off-grid sources is still a challenging issue. To improve the accuracy and efficiency of source identification, a novel method based on the off-grid model and group sparsity is proposed. Specifically, both source locations and strengths are considered in an integrated model to reconstruct the sound field with the microphone array at arbitrary positions, and the group minimax concave penalty is employed to promote the group sparsity of identification results. Then, the performance of the proposed method is examined with numerical simulations. Results indicate that the proposed method can identify off-grid sources with higher accuracy than existing methods within a broad range of noise levels and source frequencies, and less computing time is needed than using on-grid approaches. Benefit from the capacity to approximate propagation functions, the computational efficiency can be further enhanced by utilizing coarser search grids. Moreover, experimental results demonstrate the satisfactory performance and application potential of the proposed method.