随着移动通信技术的飞速发展和通信业务量的快速增长，对于通信系统的性能指标要求也不断提高，而这导致了对5G/B5G通信技术的广泛研究。大规模多输入多输出（Massive Multiple-Input Multiple-Output，Massive MIMO）作为5G/B5G的关键技术之一，能够显著提升系统容量和频谱效率，然而，由于收发端通常配备了几十根甚至上百根天线，使得Massive MIMO系统中信号检测技术面临着新的挑战。急剧增长的天线数目导致了经典信号检测算法的性能降低和复杂度的骤增，难以在性能和复杂度之间取得平衡。近来，新兴的深度学习技术拥有强大的非线性处理能力，已被应用到诸多领域中并表现出了优良的性能。因此，本文主要研究采用深度学习算法解决Massive MIMO系统中信号检测问题，首先对Massive MIMO的系统模型和信道特性进行了分析，其次研究了现有的一些经典检测算法和基于深度学习的检测算法，并对这些算法进行了仿真实验和性能分析，为后续的研究奠定基础。本文应用深度神经网络和生成对抗网络解决Massive MIMO信号检测问题，主要研究内容和创新点归纳如下：

（1）将并行干扰抵消算法（Parallel Interference Cancellation，PIC）和深度神经网络（Deep Neural Network，DNN）进行结合，提出了基于PIC-DNN的深度学习检测算法，以实现性能和复杂度的良好平衡。该算法应用PIC的思想，将MIMO系统等效为多个并行的单输入多输出（Single Input Multiple Output，SIMO）系统，接着对SIMO系统采用DNN网络进行信号检测，所设计的DNN网络将信号检测建模为深度学习的分类问题，无需信道状态信息即可实现对接收信号的盲检测。仿真结果表明，当接收天线数等于发射天线数时，所提算法在误码性能上与经典检测算法相比有明显的优势；当接收天线数多于发射天线数时，在不同调制方式下，提出的算法误码性能接近于最大似然检测算法，且具有较好的鲁棒性。

（2）将图像生成领域中具有优越性能的生成对抗网络（Generative Adversarial Network，GAN）应用于Massive MIMO系统的信号检测中，提出了基于MIMO-GAN的信号检测算法。将接收信号作为网络的输入，多次交替训练生成器和判别器以达到纳什均衡，训练完成后采用生成器对接收信号在未知信道状态信息的情况下进行检测。在传统GAN网络优化目标的基础上，对生成器引入均方差损失函数，使其优化方向正确，即朝着真实的发射信号进行优化，逐步提高检测性能。仿真结果表明，所提出的算法的误码性能在接收天线数等于发射天线数时优于经典的检测算法；当接收天线数不断增多时，其误码性能进一步提高，并且采用固定信噪比（Signal Noise Ratio，SNR）训练模式以降低实现复杂度的同时，性能损失较小。

With the rapid development of mobile communication technology and the rapid growth of communication service volume, the requirements for the performance index of communication systems have been increasing, and this has led to extensive research on 5G/B5G communication technologies. Massive Multiple-Input Multiple-Output (Massive MIMO), as one of the key technologies of 5G/B5G, can significantly improve the system capacity and spectral efficiency. However, as the transceiver is usually equipped with dozens or even hundreds of antennas, the signal detection technology in Massive MIMO system faces new challenges. The dramatic increase in the number of antennas leads to the degradation of performance and sudden increase in complexity of classical signal detection algorithms, making it difficult to balance performance and complexity. Recently, emerging deep learning techniques with powerful nonlinear processing capabilities have been applied to many fields and have shown excellent performance. Therefore, this paper focuses on solving the signal detection problem in Massive MIMO systems using deep learning algorithms. Firstly, the system model and channel characteristics of Massive MIMO are analyzed, and secondly, some existing classical detection algorithms and deep learning-based detection algorithms are studied, and simulation experiments and performance analysis of these algorithms are conducted to lay the foundation for subsequent research. In this paper, Deep Neural Network (DNN) and Generative Adversarial Network (GAN) are applied to solve the Massive MIMO signal detection problem, and the main research contents and innovations are summarized as follows：

Parallel Interference Cancellation (PIC) and Deep Neural Network (DNN) are combined to propose the PIC-DNN-based signal detection algorithm to achieve a good balance of performance and complexity. The algorithm applies the idea of PIC to equate the MIMO system to multiple parallel Single Input Multiple Output (SIMO) systems, and then uses the DNN network for signal detection of the SIMO system, and the designed DNN network models signal detection as a deep learning classification problem, which can achieve the blind detection of received signals without Channel State Information (CSI). The simulation results show that when the number of receiving antennas is equal to the number of transmitting antennas, the proposed algorithm has a significant advantage over the classical detection algorithm in terms of BER performance; when the number of receiving antennas is more than the number of transmitting antennas, the BER performance of the proposed algorithm is close to that of the maximum likelihood detection algorithm with better robustness under different modulation methods.

The Generative Adversarial Network (GAN), which has superior performance in the field of image generation, is applied to signal detection in Massive MIMO systems, and a signal detection algorithm based on MIMO-GAN is proposed. The received signal is used as the input of the network, and the generator and discriminator are trained alternately several times to achieve Nash equilibrium, and the generator is used to detect the received signal in the case of unknown Channel State Information after the training is completed. Based on the optimization objective of the conventional GAN network, the Mean Squared Error loss function is introduced to the generator to optimize it in the right direction, i.e., toward the real transmit signal, and gradually improve the detection performance. Simulation results show that the BER performance of the proposed algorithm outperforms the classical detection algorithm when the number of receive antennas is equal to the number of transmit antennas; its BER performance further improves when the number of receive antennas keeps increasing, and the performance loss is small while using a fixed Signal Noise Ratio (SNR) training mode to reduce the implementation complexity.

[1] Nguyen D C, Cheng P, Ding M, et al. Enabling AI in future wireless networks: A data life cycle perspective[J]. IEEE Communications Surveys & Tutorials, 2020, 23(1): 553-595.

[2] Mohammad A, Masouros C, Andreopoulos Y. Complexity-scalable neural-network-based MIMO detection with learnable weight scaling[J]. IEEE Transactions on Communications, 2020, 68(10): 6101-6113.

[3] Forecast G. Cisco visual networking index: global mobile data traffic forecast update, 2017–2022[J]. Update, 2019, 2017: 2022.

[4] Cisco U. Cisco annual internet report (2018–2023) white paper[J]. Cisco: San Jose, CA, USA, 2020.

[5] Shaik N, Malik P K. A comprehensive survey 5G wireless communication systems: open issues, research challenges, channel estimation, multi carrier modulation and 5G applications[J]. Multimedia Tools and Applications, 2021, 80(19): 28789-28827.

[6] Chataut R, Akl R. Massive MIMO systems for 5G and beyond networks—overview, recent trends, challenges, and future research direction[J]. Sensors, 2020, 20(10): 2753.

[7] Yang S, Hanzo L. Fifty Years of MIMO Detection: The Road to Large-Scale MIMOs [J]. IEEE Communications Surveys & Tutorials, 2017, 17(4):1941-1988.

[8] Borges D, Montezuma P, Dinis R, et al. Massive mimo techniques for 5g and beyond—opportunities and challenges[J]. Electronics, 2021, 10(14): 1667.

[9] Wei Y, Zhao M M, Hong M, et al. Learned conjugate gradient descent network for massive MIMO detection[J]. IEEE Transactions on Signal Processing, 2020, 68: 6336-6349.

[10] Zhang J, He Y, Li Y W, et al. Meta learning-based MIMO detectors: Design, simulation, and experimental test[J]. IEEE Transactions on Wireless Communications, 2020, 20(2): 1122-1137.

[11] Trotobas B, Akourim Y, Nafkha A, et al. Evaluation of the complexity, performance and implementability of geometrical MIMO detectors: The example of the exploration and exploitation list detector[J]. International Journal On Advances in Telecommunications, 2020.

[12] Zilberstein N, Dick C, Doost-Mohammady R, et al. Annealed Langevin dynamics for massive MIMO detection[J]. IEEE Transactions on Wireless Communications, 2022.

[13] Qin Z, Ye H, Li G Y, et al. Deep learning in physical layer communications[J]. IEEE Wireless Communications, 2019, 26(2): 93-99.

[14] Huo H, Xu J, Su G, et al. Intelligent MIMO Detection Using Meta Learning[J]. IEEE Wireless Communications Letters, 2022, 11(10): 2205-2209.

[15] Dargan S, Kumar M, Ayyagari M R, et al. A survey of deep learning and its applications: a new paradigm to machine learning[J]. Archives of Computational Methods in Engineering, 2020, 27: 1071-1092.

[16] Dong S, Wang P, Abbas K. A survey on deep learning and its applications[J]. Computer Science Review, 2021, 40: 100379.

[17] Nassif A B, Shahin I, Attili I, et al. Speech recognition using deep neural networks: A systematic review[J]. IEEE access, 2019, 7: 19143-19165.

[18] Otter D W, Medina J R, Kalita J K. A survey of the usages of deep learning for natural language processing[J]. IEEE transactions on neural networks and learning systems, 2020, 32(2): 604-624.

[19] Qiu X, Sun T, Xu Y, et al. Pre-trained models for natural language processing: A survey[J]. Science China Technological Sciences, 2020, 63(10): 1872-1897.

[20] Chai J, Zeng H, Li A, et al. Deep learning in computer vision: A critical review of emerging techniques and application scenarios[J]. Machine Learning with Applications, 2021, 6: 100134.

[21] Yang R, Singh S K, Tavakkoli M, et al. CNN-LSTM deep learning architecture for computer vision-based modal frequency detection[J]. Mechanical Systems and signal processing, 2020, 144: 106885.

[22] Xia J, Deng D, Fan D. A note on implementation methodologies of deep learning-based signal detection for conventional MIMO transmitters[J]. IEEE Transactions on Broadcasting, 2020, 66(3): 744-745.

[23] Shlezinger N, Fu R, Eldar Y C. DeepSIC: Deep soft interference cancellation for multiuser MIMO detection[J]. IEEE Transactions on Wireless Communications, 2020, 20(2): 1349-1362.

[24] Ma X, Gao Z. Data-driven deep learning to design pilot and channel estimator for massive MIMO[J]. IEEE Transactions on Vehicular Technology, 2020, 69(5): 5677-5682.

[25] Wei X, Hu C, Dai L. Deep learning for beamspace channel estimation in millimeter-wave massive MIMO systems[J]. IEEE Transactions on Communications, 2020, 69(1): 182-193.

[26] Mashhadi M B, Gündüz D. Pruning the pilots: Deep learning-based pilot design and channel estimation for MIMO-OFDM systems[J]. IEEE Transactions on Wireless Communications, 2021, 20(10): 6315-6328.

[27] Xu J, Zhu P, Li J, et al. Deep learning-based pilot design for multi-user distributed massive MIMO systems[J]. IEEE Wireless Communications Letters, 2019, 8(4): 1016-1019.

[28] Samuel N, Diskin T, Wiesel A. Learning to detect[J]. IEEE Transactions on Signal Processing, 2019, 67(10): 2554-2564.

[29] Sun J, Zhang Y, Xue J, et al. Learning to search for MIMO detection[J]. IEEE Transactions on Wireless Communications, 2020, 19(11): 7571-7584.

[30] Pratik K, Rao B D, Welling M. RE-MIMO: Recurrent and permutation equivariant neural MIMO detection[J]. IEEE Transactions on Signal Processing, 2020, 69: 459-473.

[31] Elgabli A, Elghariani A, Aggarwal V, et al. A low-complexity detection algorithm for uplink massive MIMO systems based on alternating minimization[J]. IEEE Wireless Communications Letters, 2019, 8(3): 917-920.

[32] Liao J, Zhao J, Gao F, et al. A model-driven deep learning method for massive MIMO detection[J]. IEEE Communications Letters, 2020, 24(8): 1724-1728.

[33] Vordonis D, Paliouras V. Sphere decoder for massive MIMO systems[C]//2019 IEEE Nordic Circuits and Systems Conference (NORCAS): NORCHIP and International Symposium of System-on-Chip (SoC). IEEE, 2019: 1-6.

[34] Liao J, Zhao J, Gao F, et al. Deep Learning Aided Low Complex Sphere Decoding for MIMO Detection[J]. IEEE Transactions on Communications, 2022, 70(12): 8046-8059.

[35] Alouzi M, Chan F, D'Amours C. Semidefinite relaxation for millimeter wave massive MIMO detection[C]//2020 11th IEEE Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON). IEEE, 2020: 0734-0739.

[36] Un M W, Shao M, Ma W K, et al. Deep MIMO detection using ADMM unfolding[C]//2019 IEEE Data Science Workshop (DSW). IEEE, 2019: 333-337.

[37] Atigh S A H, Pourrostam J. Uplink Massive MIMO Detector Based on Vector Approximate Message Passing[C]//2020 10th International Conference on Computer and Knowledge Engineering (ICCKE). IEEE, 2020: 388-391.

[38] Zheng P, Zeng Y, Liu Z, et al. Deep learning based trainable approximate message passing for massive MIMO detection[C]//ICC 2020-2020 IEEE International Conference on Communications (ICC). IEEE, 2020: 1-6.

[39] Trotobas B, Nafkha A, Louët Y. A review to massive MIMO detection algorithms: Theory and implementation[M]//Advanced Radio Frequency Antennas for Modern Communication and Medical Systems. IntechOpen, 2020.

[40] Albreem M A, Alhabbash A H, Shahabuddin S, et al. Deep learning for massive MIMO uplink detectors[J]. IEEE Communications Surveys & Tutorials, 2021, 24(1): 741-766.

[41] Huang D, Jiang X Q, Chen S, et al. Deep learning based parallel detector for MIMO systems[C]//2020 5th International Conference on Communication, Image and Signal Processing (CCISP). IEEE, 2020: 28-32.

[42] Tan X, Ueng Y L, Zhang Z, et al. A low-complexity massive MIMO detection based on approximate expectation propagation[J]. IEEE Transactions on Vehicular Technology, 2019, 68(8): 7260-7272.

[43] Albreem M A, Juntti M, Shahabuddin S. Massive MIMO detection techniques: A survey[J]. IEEE Communications Surveys & Tutorials, 2019, 21(4): 3109-3132.

[44] Yan L, Wang Y, Zheng N. 5G Massive MIMO Signal Detection Algorithm Based on Deep Learning[J]. Computational Intelligence and Neuroscience, 2022, 2022.

[45] Kumar S, Singh A, Mahapatra R. DLNet: Deep learning-aided massive MIMO decoder[J]. AEU-International Journal of Electronics and Communications, 2022, 155: 154350.

[46] Qin Z, Ye H, Li G Y, et al. Deep learning in physical layer communications[J]. IEEE Wireless Communications, 2019, 26(2): 93-99.

[47] Samuel N , Diskin T , Wiesel A . Deep MIMO Detection[C]// 2017IEEE 18th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), 2017,1-5.

[48] He H, Wen C K, Jin S, et al. A model-driven deep learning network for MIMO detection[C]//2018 IEEE Global Conference on Signal and Information Processing (GlobalSIP). IEEE, 2018: 584-588.

[49] Khani M , Alizadeh M , Hoydis J , et al. Adaptive Neural Signal Detection for Massive MIMO[J]. IEEE Transactions on Wireless Communications, 2020, 19(8):5635-5648.

[50] Sun L, Wang Y, Swindlehurst A L, et al. Generative-adversarial-network enabled signal detection for communication systems with unknown channel models[J]. IEEE Journal on Selected Areas in Communications, 2020, 39(1): 47-60.

[51] Guan X, Hu J, Qi J, et al. Observer-based H∞ sliding mode control for networked systems subject to communication channel fading and randomly varying nonlinearities[J]. Neurocomputing, 2021, 437: 312-324.

[52] 赵子坤,房晨,陈小敏,张涛涛,朱秋明.面向5G/6G大规模MIMO信道实时模拟研究[J].微波学报,2022,38(01):30-35+40.

[53] 杨婧文,朱秋明,王健,姚梦恬,陈小敏,仲伟志.无人机空对地毫米波通信路径损耗预测[J].应用科学学报,2021,39(03):398-408.

[54] 吴丽娜,何丹萍,艾渤,王剑,官科,钟章队.基于多层感知器神经网络的路径损耗预测研究[J].电波科学学报,2021,36(03):396-404.

[55] Wang C X, Huang J, Wang H, et al. 6G wireless channel measurements and models: Trends and challenges[J]. IEEE Vehicular Technology Magazine, 2020, 15(4): 22-32.

[56] Wu Z, Pan S, Chen F, et al. A comprehensive survey on graph neural networks[J]. IEEE transactions on neural networks and learning systems, 2020, 32(1): 4-24.

[57] Yang Y, Zhang S, Gao F, et al. Graph neural network-based channel tracking for massive MIMO networks[J]. IEEE Communications Letters, 2020, 24(8): 1747-1751.

[58] Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11): 139-144.

[59] He K, Wang Z, Li D, et al. Ultra-reliable MU-MIMO detector based on deep learning for 5G/B5G-enabled IoT[J]. Physical Communication, 2020, 43: 101181.

[60] Pang B, Nijkamp E, Wu Y N. Deep learning with tensorflow: A review[J]. Journal of Educational and Behavioral Statistics, 2020, 45(2): 227-248.

[61] Gurobi Optimization L L C. Gurobi optimizer reference manual[J]. 2021.

[62] Baek M S, Kwak S, Jung J Y, et al. Implementation methodologies of deep learning-based signal detection for conventional MIMO transmitters[J]. IEEE Transactions on Broadcasting, 2019, 65(3): 636-642.

[63] Lemayian J P, Hamamreh J M. Hybrid MIMO: a new transmission method for simultaneously achieving spatial multiplexing and diversity gains in MIMO systems[J]. RS Open Journal on Innovative Communication Technologies, 2021, 2(4).

[64] Khalid S, Mehmood R, Abbas W B, et al. Joint transmit antenna selection and precoding for millimeter wave massive MIMO systems[J]. Physical Communication, 2020, 42: 101137.

[65] Ha V N, Kaddoum G, Poitau G. Joint radio resource management and link adaptation for multicasting 802.11 ax-based wlan systems[J]. IEEE Transactions on Wireless Communications, 2021, 20(9): 6122-6138.

[66] Tomić I, Davidović M, Drajić D, et al. On the impact of network load on CQI reporting and Link Adaptation in LTE systems[C]//IcEtran 2021. 2021.

[67] Imoize A L, Ibhaze A E, Atayero A A, et al. Standard propagation channel models for MIMO communication systems[J]. Wireless Communications and Mobile Computing, 2021, 2021: 1-36.

[68] Jaeckel S, Turay N, Raschkowski L, et al. Industrial indoor measurements from 2-6 GHz for the 3GPP-NR and QuaDRiGa channel model[C]//2019 IEEE 90th Vehicular Technology Conference (VTC2019-Fall). IEEE, 2019: 1-7.