Joint Parallel Tabu Search Algorithm-Based Look-Up Table Design and Deep Learning-Based Signal Detection for OFDM-AIM

This letter first proposes a parallel Tabu Search (PTS) algorithm-based look-up table for orthogonal frequency division multiplexing with all index modulation (OFDM-AIM) called PTS-AIM. In the conventional OFDM-AIM (referred to as AIM in this letter), each subcarrier within a subblock is mapped to the same quadrature amplitude modulation (QAM) constellation points, while in the PTS-AIM, the PTS algorithm searches the optimum constellation point for each subcarrier to improve the bit error rate (BER) performance. Secondly, a deep learning (DL)-based signal detector termed as DeepAIM is proposed, which combines long short-term memory (LSTM) algorithm and deep neural network (DNN). Finally, a novel architecture, PTS-DeepAIM, is designed, where PTS-AIM and DeepAIM schemes are considered together. The simulation results show that the proposed PTS-DeepAIM outperforms AIM in BER performance and computation time, thanks to the benefit of the unified PTS-based look-up table design and DL-based signal detection architecture.