Abstract: To overcome the problem of huge hardware cost and computational load associated with the planar array in 3D sonar imaging application, a parallel subarray beamforming algorithm based on the cross array is proposed. The cross array can achieve 3D sonar image by two 1D linear arrays, which greatly simplifies the hardware complexity compared to planar array. The proposed algorithm is applied to optimize the receiving beamforming procedure of the cross array, and subdivide the receiving array into several subarrays. The first-stage frequency-domain beamforming is performed simultaneously in each subarray, and the second-stage beam-domain beamforming is implemented between the subarrays. The two-stage parallel and pipeline framework efficiently reduces the computational requirement of the beamforming. Moreover, the architecture of the algorithm on Field Programmable Gate Array (FPGA) is given. The simulation and analysis results demonstrate that the proposed algorithm can achieve similar beam performance but save half of the computational requirement compared to the direct method. As the number of beams increases, the superiority of the proposed algorithm becomes more obvious.

Key words: information processing, parallel subarray beamforming, subarray decomposition, cross array, computational requirement