DM-DETECT – A Deep MIMO Detector for Beyond 5G Networks
Muhammad Yunis Daha, Joseph Rafferty, Muhammad Ikram Ashraf, Muhammad Usman Hadi · 2023
The revolution of Artificial Intelligence (AI) transforms the Multiple Input Multiple Output (MIMO) technology into Massive MIMO (Ma-MIMO) technology. However, despite the promising benefits of Ma-MIMO technology, it is very difficult to design a reliable and energy-efficient detector at the receiver end. To overcome this research challenge, this paper presents a new deep Ma-MIMO Detection (DM-DETECT) scheme for Ma-MIMO detection. The DM-DETECT uses deep learning (DL) to construct an AI-based network model. The DM-DETECT model is extensively trained and optimized to achieve high performance in realistic MIMO and Ma-MIMO use cases. Simulation results depict that the DM-DETECT performs better at a certain limit than the conventional detectors in terms of Symbol Error Rate (SER). Moreover, this study presents an in-depth analysis of the activation functions for deep neural networks at different SNR ranges, which offers valuable insights into optimizing the performance of the proposed DM-DETECT for Ma-MIMO technology.