Deep Learning Based Multi-user Interference Cancellation Technology

Abstract

In the paper, I proposed a neural network-based solution to multiple access interference under the Multi-antenna Input and Multi-antenna Output (MIMO) communication system. In a model of the uplink and downlink of the multiuser MIMO system. In cases of multiple access interference, each transmitter were designed with neural networks, after the transmitted signal passes through the channel, detecting received signals at receivers designed by neural network. The model could eliminate the interference between different users. The neural network-designed model adopted Rician fading channel (including Rayleigh fading channel) and simulated the Symbol Error Rate (SER) performance of multiple users under different signal-noise ratios. With respect to SER, the solution improved system performance compared with the current multiple access interference cancellation technology. Therefore, communication systems designed with neural networks face a promising future in multiple access interference cancellation.