A novel received signal strength indicator method for modeling Massive MIMO beamforming via multi-task deep learning

Abstract

To achieve the best performance in terms of accuracy and complexity of massive multiple-input multiple-output (Massive MIMO) in wireless communication systems, hybrid beamforming (HBF) is a promising technique that provides high data rate multiplexing gains and enhances the spectral efficiency (SE) of the system. In this paper, a novel received signal strength indicator (RSSI) method is proposed to design an HBF for Massive MIMO BF via multitasking deep learning (DL) that minimizes the reliance on the channel state information (CSI) feedback. The trade-off between the enhancement SE of the system and the deep neural networks (DNNs) performance is optimized, and the results reveal that the proposed novel DL techniques achieve predicted spectral efficiencies with accuracy of 99.23% and 95.64% for Deep-HBF and Deep-AFP, respectively. The processing times for Deep-HBF and Deep-AFP are 709.2914 sec and 1425.864 sec, respectively. Notably, Deep-AFP exhibits a higher range of computational complexity compared to Deep-HBF. It is worth mentioning that the proposed techniques utilize the same DNN architecture.