A Sparse Arrays Sidelobe Suppression Method: Clean Algorithm for Joint Target Number Estimation

Sparse arrays provide high angular resolution and avoid the grating lobes, but the average and peak sidelobe level of the spase arrays are relatively high, that may lead to the occurrence of false targets. Although the traditional CLEAN method can effectively suppress the sidelobe in high SNR situation, it may fail under the severe channel environments. The traditional CLEAN algorithm needs to recalculate the threshold value every iteration, since threshold calculations depend on the results of each iteration, it may lead to false positives. In this paper, we propose a CLEAN algorithm for joint target number estimation. It rounds off the traditional clean algorithm's step of calculating the threshold for each iteration, and uses the target number as the CLEAN algorithm iteration number. The target number estimation based on deep neural network has better estimation performance under the severe channel environments. Therefore, we first use the deep neural network model to predict the number of targets, and then use the obtained result as the iterative termination condition of the CLEAN algorithm. Simulation result shows the effectiveness of the proposed method.