Efficient Selection on Spatial Modulation Antennas: Learning or Boosting
Yue Zhang, Jintao Wang, Xuesi Wang, Yonglin Xue, Jian Song
Abstract
In this letter, a novel deep learning-based transmit antenna selection (TAS) scheme for the multiple-input multiple-output (MIMO) with spatial modulation (SM) system is proposed. We formulate the generalized TAS pipeline in both neural networks (NN) and gradient boosting decision trees (GBDT), in which the importance of different features reflecting the different elements from channel state information (CSI) is analyzed regarding to the empirical data as well. Furthermore, the bit error rate (BER) performance and the complexity comparison of two structures is investigated. Simulation results confirm that GBDT can be efficiently implemented for real-time application with near-optimal performance.