随着通信技术的不断发展，毫米波、大规模多输入多输出（Massive MIMO）与波束赋形得到了广泛的关注。频段扩展至毫米波频段解决了频谱资源稀缺问题，并且波长较短的毫米波便于天线的集成，从而实现Massive MIMO技术。一方面，传统的全数字波束赋形虽然性能较优但成本过高，而模拟波束赋形虽然成本较低但性能较差，混合波束赋形技术应运而生。混合波束赋形实现了系统成本和性能的有效折衷，但系统性能与数字波束赋形相比仍有不足。另一方面，在实际通信中受混合架构影响，信道存在秩损问题，会导致系统性能进一步损失。因此，本文在毫米波Massive MIMO系统下，采用深度学习算法，解决混合波束赋形性能不足以及实际通信信道的秩损问题。

为了进一步提升混合波束赋形算法的性能，在多用户Massive MIMO混合架构系统下，以频谱效率最大化为目标，提出了两种生成对抗网络算法来优化混合波束赋形性能，分别是基于条件生成对抗网络的数字波束赋形算法和基于双生成器的混合波束赋形算法。基于条件生成对抗网络的数字域波束赋形算法，以模拟域波束赋形矩阵作为网络的输入，生成所需的数字域波束赋形矩阵。基于双生成器生成对抗网络（DGGAN）的混合波束赋形算法，以信道矩阵作为网络的输入，两个生成器分别生成模拟域波束赋形矩阵和数字域波束赋形矩阵。仿真结果表明在所考虑仿真条件下，所提两种算法的频谱效率优于其他算法，性能更接近全数字波束赋形上界。尤其是DGGAN方法，其频谱效率与传统基于矩阵运算的算法相比提升了10.74%-20.59%，与已有的深度学习网络架构相比，性能提升了4.67%-8.31%。

为了对抗实际混合架构Massive MIMO系统固有的信道秩损所带来的性能损失，在秩损信道下多用户Massive MIMO混合架构系统中，提出了两种新的生成对抗网络算法来实现秩损信道下的混合波束赋形设计，分别是基于生成对抗网络的信道补回算法和基于多生成器的秩损信道补回以及混合波束赋形联合优化方案。基于生成对抗网络的秩损信道补回算法，以秩损信道为网络的输入，生成器对缺失信息进行补回，得到补回后的信道矩阵。基于多生成器生成对抗网络的秩损信道补回以及混合波束赋形的联合优化方案，将频谱效率直接作为网络的优化目标，得到预期的系统性能。仿真结果表明在所考虑仿真条件下，所提两种算法性能优于其他算法，信道补回算法将系统性能提升了38.81%-51.49%，联合优化算法相比于传统算法性能提升48.11%-77.79%，相比于其他深度学习算法性能提升9.40%-52.44%，体现了所提算法的鲁棒性。

With the continuous development of communication technology, millimeter wave, massive multiple-input multiple-output (Massive MIMO) and beamforming have received extensive attention. The expansion of the frequency band to the millimeter wave band solves the problem of scarcity of spectrum resources, and the shorter wavelength of millimeter wave facilitates the integration of antennas, thus realizing massive MIMO technology. On the one hand, although the traditional full-digital beamforming has better performance, the cost is too high, while the analog beamforming has lower cost but poor performance, and the hybrid beamforming technology emerges as the times require. Hybrid beamforming achieves an effective compromise between system cost and performance, but the system performance is still insufficient compared to digital beamforming. On the other hand, affected by the hybrid architecture in actual communication, there is a rank-deficient problem in the channel, which will lead to further loss of system performance. Therefore, in this paper, under the millimeter wave massive MIMO system, a deep learning algorithm is used to solve the problem of insufficient hybrid beamforming performance and the rank-deficient of the practical communication channel.

In order to further improve the performance of the hybrid beamforming algorithm, two Generative Adversarial Network (GAN) algorithms, which are digital beamforming algorithms based on conditional GAN and hybrid beamforming algorithms based on dual generators, are proposed to optimize the hybrid beamforming performance and maximize the system spectral efficiency based on multi-user massive MIMO hybrid architecture system. The digital beamforming algorithm based on conditional GAN takes the analog beamforming matrix as the input of the network to generate the required digital beamforming matrix. The hybrid beamforming algorithm based on Dual-Generator Generative Adversarial Network (DGGAN) takes the channel matrix as the input of the network, and the two generators generate the analog beamforming matrix and the digital beamforming matrix, respectively. The simulation results show that the spectral efficiency of the proposed two algorithms is better than other algorithms in the considered simulation conditions, and the performance is closer to the upper bound of full-digital beamforming. Especially the DGGAN method, its spectral efficiency is improved by 10.74%-20.59% compared with the traditional matrix-based algorithm, and compared with the existing deep learning network architecture, the performance is improved by 4.67%-8.31%.

In order to combat the performance loss caused by the inherent channel rank-deficient of the practical hybrid architecture massive MIMO system, based on the multi-user massive MIMO hybrid architecture system under rank-deficient channels, two new GAN algorithms are proposed, which are a channel compensation algorithm based on GAN and a rank-deficient channel compensation and hybrid beamforming joint optimization scheme based on multi-generators, to realize the hybrid beamforming design under rank-deficient channels. Based on the rank-deficient channel compensation algorithm of GAN, the rank-deficient channel is used as the input of the network, and the generator compensates the missing information to obtain the compensated channel matrix. The joint optimization scheme of rank-deficient channel compensation and hybrid beamforming based on multi-generator GAN, the spectral efficiency is directly regarded as the optimization goal of the network, and the expected system performance is obtained. The simulation results show that the performance of the proposed two algorithms is better than other algorithms in the considered simulation conditions. The channel compensation algorithm improves the system performance by 38.81%-51.49%. The joint optimization algorithm improves the performance by 48.11%-77.79% compared with the traditional algorithms, and the performance is improved by 9.40%-52.44% compared with other deep learning algorithms, which reflects the robustness of the proposed algorithms.

[1] 王东明, 张余, 魏浩, et al. 面向5G的大规模天线无线传输理论与技术[J]. 中国科学:信息科学, 2016, 46(1):3-21.

[2] 赵亚军, 郁光辉, 徐汉青. 6G移动通信网络:愿景,挑战与关键技术[J]. 中国科学F辑, 2019，49(8):963-987.

[3] ERICSSON, Ericsson Mobility Report: Special edition world economic forum, white paper[M]. Stockholm, Sweden: Ericsson, 2019.

[4] David K, Berndt H. 6G vision and requirements: Is there any need for beyond 5G?[J]. IEEE vehicular technology magazine, 2018, 13(3): 72-80.

[5] Strinati E C, Barbarossa S, Gonzalez-Jimenez J L, et al. 6G: The next frontier: From holographic messaging to artificial intelligence using subterahertz and visible light communication[J]. IEEE Vehicular Technology Magazine, 2019, 14(3): 42-50.

[6] Zong B, Fan C, Wang X, et al. 6G technologies: Key drivers, core requirements, system architectures, and enabling technologies[J]. IEEE Vehicular Technology Magazine, 2019, 14(3): 18-27.

[7] Rappaport T S, Xing Y, Kanhere O, et al. Wireless communications and applications above 100 GHz: Opportunities and challenges for 6G and beyond[J]. IEEE access, 2019, 7: 78729-78757.

[8] Bogale T E, Le L B. Massive MIMO and mmWave for 5G wireless HetNet: Potential benefits and challenges[J]. IEEE Vehicular Technology Magazine, 2016, 11(1): 64-75.

[9] Roh W, Seol J Y, Park J, et al. Millimeter-wave beamforming as an enabling technology for 5G cellular communications: Theoretical feasibility and prototype results[J]. IEEE communications magazine, 2014, 52(2): 106-113.

[10] Wei L, Hu R Q, Qian Y, et al. Key elements to enable millimeter wave communications for 5G wireless systems[J]. IEEE Wireless Communications, 2014, 21(6): 136-143.

[11] Kutty S, Sen D. Beamforming for millimeter wave communications: An inclusive survey[J]. IEEE communications surveys & tutorials, 2015, 18(2): 949-973.

[12] Chen Y, Xia Y, Xing Y, et al. Low complexity hybrid precoding for mmWave massive MIMO systems[C]//2017 26th Wireless and Optical Communication Conference (WOCC). IEEE, 2017: 1-5.

[13] Wang X, Chen R, Xu Y, et al. Low-complexity power allocation in NOMA systems with imperfect SIC for maximizing weighted sum-rate[J]. IEEE Access, 2019, 7: 94238-94253.

[14] Srivastava S, Mishra A, Rajoriya A, et al. Quasi-static and time-selective channel estimation for block-sparse millimeter wave hybrid MIMO systems: Sparse Bayesian learning (SBL) based approaches[J]. IEEE Transactions on Signal Processing, 2018, 67(5): 1251-1266.

[15] Ding Z, Fan P, Poor H V. Random beamforming in millimeter-wave NOMA networks[J]. IEEE access, 2017, 5: 7667-7681.

[16] Sheemar C K, Thomas C K, Slock D. Practical hybrid beamforming for millimeter wave massive MIMO full duplex with limited dynamic range[J]. IEEE Open Journal of the Communications Society, 2022, 3: 127-143.

[17] Moon S, Lee H, Choi J, et al. Low-Complexity Beamforming Optimization for IRS-Aided MU-MIMO Wireless Systems[J]. IEEE Transactions on Vehicular Technology, 2022.

[18] Lin T, Cong J, Zhu Y, et al. Hybrid beamforming for millimeter wave systems using the MMSE criterion[J]. IEEE Transactions on Communications, 2019, 67(5): 3693-3708.

[19] Khalid F. Hybrid beamforming for millimeter wave massive multiuser MIMO systems using regularized channel diagonalization[J]. IEEE Wireless Communications Letters, 2018, 8(3): 705-708.

[20] Luo Z, Zhao L, Liu H, et al. Robust hybrid beamforming in millimeter wave systems with closed-form least-square solutions[J]. IEEE Wireless Communications Letters, 2020, 10(1): 156-160.

[21] Ying K, Gao Z, Lyu S, et al. GMD-based hybrid beamforming for large reconfigurable intelligent surface assisted millimeter-wave massive MIMO[J]. IEEE Access, 2020, 8: 19530-19539.

[22] Nasir A A, Tuan H D, Duong T Q, et al. Hybrid beamforming for multi-user millimeter-wave networks[J]. IEEE Transactions on Vehicular Technology, 2020, 69(3): 2943-2956.

[23] Lyu S, Wang Z, Gao Z, et al. Lattice-based mmWave hybrid beamforming[J]. IEEE Transactions on Communications, 2021, 69(7): 4907-4920.

[24] 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(4): 1071-1092.

[25] Mao Q, Hu F, Hao Q. Deep learning for intelligent wireless networks: A comprehensive survey[J]. IEEE Communications Surveys & Tutorials, 2018, 20(4): 2595-2621.

[26] Alom M Z, Taha T M, Yakopcic C, et al. A state-of-the-art survey on deep learning theory and architectures[J]. Electronics, 2019, 8(3): 292.

[27] Tanuwidjaja H C, Choi R, Baek S, et al. Privacy-preserving deep learning on machine learning as a service—a comprehensive survey[J]. IEEE Access, 2020, 8: 167425-167447.

[28] Ouyang W, Chu X, Wang X. Multi-source deep learning for human pose estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2014: 2329-2336.

[29] Zhou Y, Hu Q, Liu J, et al. Combining heterogeneous deep neural networks with conditional random fields for Chinese dialogue act recognition[J]. Neurocomputing, 2015, 168: 408-417.

[30] Zhao L, Hu Q, Wang W. Heterogeneous feature selection with multi-modal deep neural networks and sparse group lasso[J]. IEEE Transactions on Multimedia, 2015, 17(11): 1936-1948.

[31] Peken T, Adiga S, Tandon R, et al. Deep learning for SVD and hybrid beamforming[J]. IEEE Transactions on Wireless Communications, 2020, 19(10): 6621-6642.

[32] Elbir A M. CNN-based precoder and combiner design in mmWave MIMO systems[J]. IEEE Communications Letters, 2019, 23(7): 1240-1243.

[33] Song N, Ye C, Hu X, et al. Deep learning based low-rank channel recovery for hybrid beamforming in millimeter-wave massive MIMO[C]//2020 IEEE Wireless Communications and Networking Conference (WCNC). IEEE, 2020: 1-6.

[34] Lin T, Zhu Y. Beamforming design for large-scale antenna arrays using deep learning[J]. IEEE Wireless Communications Letters, 2019, 9(1): 103-107.

[35] Lizarraga E M, Maggio G N, Dowhuszko A A. Deep reinforcement learning for hybrid beamforming in multi-user millimeter wave wireless systems[C]//2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring). IEEE, 2021: 1-5.

[36] Almagboul M A, Shu F, Abdelgader A M S. Deep-Learning-Based Phase-Only Robust Massive MU-MIMO Hybrid Beamforming[J]. IEEE Communications Letters, 2021, 25(7): 2280-2284.

[37] Osama I, Rihan M, Elhefnawy M, et al. Deep Learning Based Hybrid Precoding Technique for Millimeter-Wave Massive MIMO Systems[C]//2021 International Conference on Electronic Engineering (ICEEM). IEEE, 2021: 1-7.

[38] Elbir A M, Mishra K V. Robust hybrid beamforming with quantized deep neural networks[C]//2019 IEEE 29th International Workshop on Machine Learning for Signal Processing (MLSP). IEEE, 2019: 1-6.

[39] Tao J, Xing J, Chen J, et al. Deep neural hybrid beamforming for multi-user mmWave massive MIMO system[C]//2019 IEEE Global Conference on Signal and Information Processing (GlobalSIP). IEEE, 2019: 1-5.

[40] Sung L, Cho D H. Multi-User Hybrid Beamforming System Based on Deep Neural Network in Millimeter-Wave Communication[J]. IEEE Access, 2020, 8: 91616-91623.

[41] Vishwakarma A, Meda D K. Optimal Detection of data symbol in UDMT Massive MIMO 5G System model for high spectral efficiency and energy efficiency[C]//2021 4th International Conference on Recent Developments in Control, Automation & Power Engineering (RDCAPE). IEEE, 2021: 501-505.

[42] Mishra S K, Panda A V, Mahapatro A, et al. Impact of Spatial Correlation on Sum-Rate Capacity in MU-MIMO Broadcast Channels[C]//2021 IEEE 18th India Council International Conference (INDICON). IEEE, 2021: 1-5.

[43] An M, Sun Y, Zhai H, et al. A Method for Increasing the Channel Capacity of MIMO Antenna for Base Station Array[C]//2021 International Conference on Microwave and Millimeter Wave Technology (ICMMT). IEEE, 2021: 1-3.

[44] Lu A Y, Chen Y F. Adaptive Hybrid Beamforming Schemes in Millimeter Wave MIMO Systems[C]//2021 IEEE International Symposium on Radio-Frequency Integration Technology (RFIT). IEEE, 2021: 1-3.

[45] Ezzine R, Wiese M, Deppe C, et al. Outage Common Randomness Capacity Characterization of Multiple-Antenna Slow Fading Channels[C]//2021 IEEE Information Theory Workshop (ITW). IEEE, 2021: 1-6.

[46] Khalid F. Hybrid beamforming for millimeter wave massive multiuser MIMO systems using regularized channel diagonalization[J]. IEEE Wireless Communications Letters, 2018, 8(3): 705-708.

[47] Lim Y G, Cho Y J, Sim M S, et al. Map-based millimeter-wave channel models: An overview, data for B5G evaluation and machine learning[J]. IEEE Wireless Communications, 2020, 27(4): 54-62.

[48] Goldsmith A . Wireless Communications[M]. 人民邮电出版社, 2007.

[49] Yamashita R, Nishio M, Do R K G, et al. Convolutional neural networks: an overview and application in radiology[J]. Insights into imaging, 2018, 9(4): 611-629.

[50] Ojha V K, Abraham A, Snášel V. Metaheuristic design of feedforward neural networks: A review of two decades of research[J]. Engineering Applications of Artificial Intelligence, 2017, 60: 97-116.

[51] Dyer C, Kuncoro A, Ballesteros M, et al. Recurrent neural network grammars[J]. arXiv preprint arXiv:1602.07776, 2016.

[52] Ni W, Dong X, Lu W S. Near-optimal hybrid processing for massive MIMO systems via matrix decomposition[J]. IEEE transactions on signal processing, 2017, 65(15): 3922-3933.

[53] Tsai T H, Chiu M C, Chao C. Sub-system SVD hybrid beamforming design for millimeter wave multi-carrier systems[J]. IEEE Transactions on Wireless Communications, 2018, 18(1): 518-531.

[54] Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets[J]. Advances in neural information processing systems, 2014, 27.

[55] Karras T, Aila T, Laine S, et al. Progressive growing of gans for improved quality, stability, and variation[J]. arXiv preprint arXiv:1710.10196, 2017.

[56] Zhu J Y, Park T, Isola P, et al. Unpaired image-to-image translation using cycle-consistent adversarial networks[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2223-2232.

[57] Antipov G, Baccouche M, Dugelay J L. Face aging with conditional generative adversarial networks[C]//2017 IEEE international conference on image processing (ICIP). IEEE, 2017: 2089-2093.

[58] Ledig C, Theis L, Huszár F, et al. Photo-realistic single image super-resolution using a generative adversarial network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 4681-4690.

[59] Yang L C, Chou S Y, Yang Y H. MidiNet: A convolutional generative adversarial network for symbolic-domain music generation[J]. arXiv preprint arXiv:1703.10847, 2017.

[60] Donahue C, McAuley J, Puckette M. Adversarial audio synthesis[J]. arXiv preprint arXiv:1802.04208, 2018.

[61] Wang T C, Liu M Y, Zhu J Y, et al. Video-to-video synthesis[J]. arXiv preprint arXiv:1808.06601, 2018.

[62] Mirza M, Osindero S. Conditional generative adversarial nets[J]. arXiv preprint arXiv:1411.1784, 2014.

[63] Ni W, Dong X. Hybrid block diagonalization for massive multiuser MIMO systems[J]. IEEE transactions on communications, 2015, 64(1): 201-211.

[64] Alkhateeb A, El Ayach O, Leus G, et al. Channel estimation and hybrid precoding for millimeter wave cellular systems[J]. IEEE journal of selected topics in signal processing, 2014, 8(5): 831-846.

[65] He K, Sun J. Convolutional neural networks at constrained time cost[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 5353-5360.