RIS-Assisted Dual-Polarized Spatial Modulation With CVCNN-Based Detector

In this letter, a RIS-assisted dual-polarized generalized spatial modulation (RIS-DPGSM) is proposed, which jointly utilizes the spatial dimension and the polarization states of transmit antennas to convey extra information than the traditional wireless schemes. Meanwhile, based on deep learning (DL), a complex-valued convolutional neural network (CVCNN) detector is designed to address the trade-off issue between the complexity and BER for the proposed scheme. The performance of the employed CVCNN detector outperforms that of the block zero-force (B-ZF) detector and modified ordered block minimum mean-squared error (MOB-MMSE) detector with the BER approaching to that of the maximum likelihood (ML) detector. By comparing the benchmark scheme called the RIS-assisted unit-polarized GSM (RIS-UPGSM), simulation results show that the RIS-DPGSM scheme has better BER and spectral efficiency (SE) performance than the existing one. The results demonstrate that the proposed RIS-DPGSM significantly improves SE with considerable BER performance and low complexity using CVCNN, showcasing its strong potential and scalability in addressing the challenges of next-generation communication networks.