Deep Learning Based Antenna Selection Aided Space-Time Shift Keying Systems

In this paper, we propose an antenna selection aided space-time shift keying(STSK) multiple-input multiple-output(MIMO) system based on deep learning method. More specifically, we first extract the channel characteristics from the channel state information, then labeling every sample by maximizing the key performance indicator. Finally, we consider antenna selection as multi-class classification learning and classify the samples through neural network to select the optimal antenna subset for supporting the actual communications. Simulation results demonstrate that neural network may achieve higher accuracy than other machine learning methods, and the corresponding performance stays close to the optimal ergodic search method.