On the DOA Estimation Performance of Optimum Arrays Based on Deep Learning

Abstract

In this paper, we investigate the optimality of deep learning-based optimal sparse arrays in comparison to well known conventional sparse linear arrays. Deep learning-based sparse arrays are realized through a deep learning-based approach which was proposed recently for antenna selection purposes as a measure towards reducing high hardware and computational cost in radar systems. Through numerical examples, we demonstrated that the proposed approach yields sparse arrays whose performance and configurations are comparably closer to conventional sparse arrays.