Efficient Antenna Selection for Adaptive Enhanced Spatial Modulation A Deep Neural Network Approach

In this letter, we employ a machine learning algorithm based on transmit antenna selection (TAS) for adaptive enhanced spatial modulation (AESM). Firstly, channel state information (CSI) is used to predict the TAS problem in AESM. In addition, a low-complexity multi-class supervised learning classifier of deep neural network (DNN) is introduced. Meanwhile, adaptive gradient (AdaGrad) is applied to optimize the network structure and reduce network training time. Finally, the simulation results show that the proposed scheme efficiency is higher than traditional TAS in the AESM system and provides a similar bit error rate (BER) performance while the computational complexity of the system is lowest.