Phase-Only Reconfigurable Sparse Array Beamforming Using Deep Learning

Abstract

The paper considers phase-only reconfigurable sparse arrays (RSAs) for receive beamforming to maximize signal-to-interference plus noise ratio (MaxSINR). We develop a design approach based on supervised deep neural network (DNN) to learn and mimic a phase-only sparse MaxSINR beamformer. The proposed approach strives to match the SINR performance of data driven sparse Capon beamformer. The problem is posed as a multi-label classification problem, where the received antenna correlations is the input to the fully connected neural network (FCNN) which outputs the optimum sensor locations for effective interference mitigation. We evaluate the performance of DNN based optimization of RSAs in terms of the ability of the classified sparse array to mitigate interference and maximize signal power using phase-only beamforming. The phase-only DNN-based sparse sensor placement reduces hardware requirements, shifts optimization algorithm complexity to satisfying training data sufficiency, and is amenable to real-time implementation.