Deep Learning based MAP Selection Techniques for Media Based Modulation Scheme
IEEE Access, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1109/access.2026.3715650
- Dergi Adı: IEEE Access
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Directory of Open Access Journals
- Anahtar Kelimeler: capacity-optimized map selection, deep learning, Euclidean distance optimized map selection, index modulation, map correlation based map selection, Media-based modulation, MIMO systems, Rayleigh fading channel, spectral efficiency
- Yıldız Teknik Üniversitesi Adresli: Evet
Özet
In this paper, a new high data rate, high energy-efficient, and low-complexity telecommunication system is proposed by combining the recently popular media-based modulation (MBM) technique, mirror activation pattern (MAP) selection techniques, and a deep learning framework. In this proposed system, in addition to the conventional modulated symbols, the MBM technique carries additional information in the MAP indices that occur depending on whether the RF mirrors are on/off without incurring additional energy. Furthermore, performance is enhanced by employing threeMAPselection strategies: Euclidean distance map selection (EMS), map correlation-based selection (MCMS), and capacity-optimized map selection (CMS). To significantly reduce the complexity of choosing MAPs a deep neural network (DNN) is introduced. This DNNefficiently identifies the most favorableMAPsubsets based on the channel conditions, thereby reducing computational burden and accelerating real-time system operation. These methods are collectively referred to as deep learning-based Euclidean distance map selection assistedMBM(DL-EMS-MBM), deep learning-based map correlation assisted MBM (DL-MCMS-MBM), and deep learning-based capacity-optimized map selection assisted MBM (DL-CMS-MBM). Among these, EMS offers the best performance with the highest complexity, while CMS has the lowest complexity and performance, and MCMS lies in between. Simulation results over Rayleigh fading channels with M-ary quadrature amplitude modulation (QAM) demonstrate that the proposed DL-aided MBM techniques not only outperform traditional MBM but also offer performance comparable to or better than their conventional EMS, MCMS, and CMS counterparts, all while substantially reducing complexity. In addition, the applicability of the proposed learning-based selection has been illustrated under Nakagami-m fading for the CMS/DL-CMS case.