Deep Learning Assisted Adaptive Index Modulation for mmWave Communications With Channel Estimation

The efficiency of link adaptation in wireless communications relies greatly on the accuracy channel knowledge and transmission mode selection. In this paper, a novel deep learning based framework is proposed for orthogonal frequency-division multiplexing (OFDM) systems with compressed-sensing-assisted index modulation, termed as OFDM-CSIM, communicating over millimeter-wave (mmWave) channels. To achieve adaptation, multi-layer sparse Bayesian (SBL) algorithm accurately instantaneously providing required state information. Meanwhile, neural networks (DNN)-assisted adaptive modulation to choose best possible maximize achievable throughput. Simulation results show that SBL enables more accurate estimation than conventional techniques. DNN-based modulator capable achieving higher throughput learning-assisted solution $k$ nearest neighbor ( -NN) algorithm, also classic average signal-to-noise ratio (SNR)-based solutions. Moreover, analysis shows both DNN-assisted better performance their respective counterparts while at significantly lower computational complexity cost.