Convolutional-Neural-Network-Based Detection Algorithm for Uplink Multiuser Massive MIMO Systems

In the coming 6th generation (6G) and beyond in wireless communication, an increasing number of ultrascale intelligent factors, including mobile robot users and smart cars, will result in interference exploitation. The management of this exploitation will be a great challenge for detection algorithms in uplink massive multiple-input and multiple-output (MIMO) systems, especially for high-order quadrature amplitude modulation (QAM) signals. Artificial intelligence technology employing machine learning is one of the key approaches among the 6G technical solutions. In this paper, a convolutional-neural-network-based likelihood ascent search (CNNLAS) detection algorithm is proposed on the basis of a graphical detection model for uplink multiuser massive MIMO systems. Compared with other algorithms, the proposed CNNLAS detection algorithm has a stronger robustness against the channel estimation errors, and requires lower average received signal-to-noise ratios to obtain better bit error rate performance and to achieve the theoretical spectral efficiency with a lower polynomial average per symbol computational complexity, both for the graphical low-order and high-order QAM signals in uplink multiuser massive MIMO systems.