Deep Learning for Multiple-Input Multiple-Output communication system

Abstract

Multiple-Input Multiple-Output (MIMO) technology is an emerging and promising technology for wireless communications, especially in the implementation of the fifth generation (5G). Due to implementation of Internet-of-Things (IoT), 5G needs to account for higher user density. With the strive for high spectral efficiency, high energy efficiency and low latency, the study of MIMO is crucial to enhance 5G performance. As such, the use of Deep Learning (DL), modelled by the human brain, can be a useful tool in the exploration of various aspects of MIMO systems. Taking into considerations the multiple antenna characteristics of MIMO systems, the study of Rayleigh fading and Additive White Gaussian Noise (AWGN) needs to be incorporated into wireless transmission techniques. Along with the generation of Bit Error Rate (BER), Signal-to-Noise Ratio (SNR), the study of MIMO channel capacity and detection of data symbols are vital in the implementation of a DL model. With the nature of DL, we also consider DeepMIMO and various dataset options to bring about the best fit model. Analysis results using Detection Network (DETNet) show the success of DL in the study of channel capacity and BER.