Supervised Learning Classifier Based Transmit Antenna Selection for SM-MIMO System

In this paper, we propose the supervised learning classifier (SLC) based transmit antenna selection (TAS) for spatial modulation (SM) multiple-input multiple-output (MIMO) system. Firstly, the classical Euclidean distance based antenna selection (EDAS) method is investigated. Secondly, we convert the typical TAS problem into a classification problem and propose a low complexity multi-output classifier design. Then, two SLC schemes, random forest decision (RFD) and deep neural network (DNN) with adaptive momentum (ADAM) optimization, are employed to solve the TAS problem. Numerical results illustrate that the proposed DNN based TAS (DNN-TAS) scheme obtains better average bit error rate (ABER) performance than the RFD based TAS (RFD-TAS) scheme and outperforms the typical TAS methods with much lower complexity. Moreover, the proposed DNN-TAS scheme achieves the suboptimal ABER performance at high signal-to-noise ratio (SNR) region compared with the optimal high complexity EDAS algorithm.