Deep learning aided CSI feedback optimization with robust error recovery✩

Bo Yang; Zhanglin Zhou; Nanqi Fan; Feng Ke; Jie Tang; XiuYin Zhang

doi:10.1016/j.dcan.2025.11.002

›› 2026, Vol. 12 ›› Issue (2) :354 -363. DOI: 10.1016/j.dcan.2025.11.002

Regular Papers

research-article

Deep learning aided CSI feedback optimization with robust error recovery^✩

Author information +

South China University of Technology, Guangzhou 510641, China

^* E-mail address: fengke@scut.edu.cn (F. Ke).

Bo Yang is employed at Comba Network Systems Company Limited, serving as the Head of Research and Development, Guangzhou 510663, China (email:yangb@comba-network.com.cn).

Zhanglin Zhou received the B.S. degree in communication engineering from the Hunan University, Changsha, China in 2021. He is with the School of Electronic and Infor-mation Engineering, South China University of Technology, Guangzhou 510641, China. He is currently pursuing the M.S. degree in Electronic and Communication Engineering at the South China University of Technology. His research interests include future wireless communications and Internet of Things.

Nanqi Fan received the B.S. degree in Information Engineering from Guangdong Uni-versity of Technology, Guangzhou, China. He is currently with the School of Electronic and Information Engineering at South China University of Technology, Guangzhou 510641, China, where he is pursuing the M.S. degree in Electronic and Information Engineering. His research interests include semantic communication and Non-Terrestrial Networks.

Feng Ke received his B.S. degree, M.S. degree and Ph.D. degree in electronic en-gineering from South China University of Technology (SCUT) in 1998, 2001 and 2004 respectively. He is in the School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510641, China, and also with the Pazhou Labora-tory, Guangzhou 510330, China. He serves as the vice director of the Intelligent Com-munication Network and Computer Engineering Technology Engineering Center at SCUT and the chief scientist of Guangdong Chinese Network Electrical Engineering Research Institute. He also serves as an expert advisor of Guangzhou Science and Technology Bu-reau and Guangdong Science and Technology Department. From 2013 to 2014, he was a visiting scholar in the Department of Electronic Engineering, Missouri University of Sci-ence and Technology, Rolla. His research interests include resource allocation and signal processing for future wireless communications.

Jie Tang is with the School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China (e-mail: eejtang@scut.edu.cn).

XiuYin Zhang is with the Center for Short-Distance Wireless Communications and Network Engineering Research, Ministry of Education, School of Electronic and Infor-mation Engineering, South China University ofTechnology. Guangzhou 510000, China (e-mail: eexyz@scut.edu.cn).

Show less

History +

PDF

Abstract

Driven by the increasing demand for eﬃcient data transmission, massive Multiple-Input Multiple-Output (MIMO) systems have emerged as a key technology for future communication systems. However, eﬀective utilization of MIMO relies heavily on accurate Channel State Information (CSI) that is fed back to the base station, which poses significant challenges due to the overhead associated with CSI feedback, especially with the increasing number of antennas. To overcome these drawbacks, this paper proposes a Deep Learning (DL) scheme to improve the CSI feedback, presenting a network named CsiDNet, which compresses CSI at the user end and decompresses it at the base station side. In addition, an auxiliary module is designed to restore CSI information under error-prone scenarios, enhancing the robustness of the system. Extensive performance analysis and simulations demonstrate that CsiDNet achieves an improvement of 2.7 dB and 0.1 dB in terms of Normalized Mean Square Error (NMSE) and Square Generalized Cosine Similarity (SGCS) respectively compared to other models, while significantly reducing computational complexity. The auxiliary module further improves the NMSE and SGCS performance by 4 dB and 0.1 dB respectively, reflecting its eﬀectiveness in recovering error-prone CSI components. Overall, our research improves the accuracy and eﬃciency of CSI feedback while enhancing the system’s robustness against real-world transmission challenges.

Keywords

CSI feedback / Deep Learning / MIMO / Packet loss

Cite this article

Download citation ▾

Bo Yang, Zhanglin Zhou, Nanqi Fan, Feng Ke, Jie Tang, XiuYin Zhang. Deep learning aided CSI feedback optimization with robust error recovery^✩. , 2026, 12(2): 354-363 DOI:10.1016/j.dcan.2025.11.002

登录浏览全文

4963

注册一个新账户忘记密码

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported in part by the National Natural Science Foundation of China (NSFC) under Grant 62171188, in part by the Na-tional Foreign Expert Project of China under Grant H20241004, and in part by the Guangdong Provincial Key Laboratory of Human Digital Twin under Grant 2022B1212010004.

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	F.A. Pereira de Figueiredo, An overview of massive mimo for 5g and 6g, IEEE Latin Am. Trans. 20 (6) (2022) 931-940.

[2]	B.M. Lee, Massive mimo for massive industrial Internet of things networks: opera-tion, performance, and challenges, IEEE Trans. Cogn. Commun. Netw. 10 (6) (2024) 2119-2135.

[3]	R.M. Dreifuerst, R.W. Heath, Massive mimo in 5g: how beamforming, codebooks, and feedback enable larger arrays, IEEE Commun. Mag. 61 (12) (2023) 18-23.

[4]	A. Krishnamoorthy, R. Schober, Downlink massive mu-mimo with successively-regularized zero forcing precoding, IEEE Wirel. Commun. Lett. 12 (1) (2023) 114-118.

[5]	J. Pauls, M. Vaezi, Secure precoding in mimo-noma: a deep learning approach, IEEE Wirel. Commun. Lett. 11 (1) (2022) 77-80.

[6]	X. Jiang, F. Kaltenberger, Channel reciprocity calibration in tdd hybrid beamforming massive mimo systems, IEEE J. Sel. Top. Signal Process. 12 (3) (2018) 422-431.

[7]	J. Guo, C.-K. Wen, M. Chen, S. Jin, Environment knowledge-aided massive mimo feedback codebook enhancement using artificial intelligence, IEEE Trans. Commun. 70 (7) (2022) 4527-4542.

[8]	P.K. Gadamsetty, K.V.S. Hari,A fast dictionary learning algorithm for csi feedback in massive mimo fdd systems, in:2023 National Conference on Communications (NCC), 2023, pp. 1-6.

[9]	Z. Qin, J. Fan, Y. Liu, Y. Gao, G.Y. Li, Sparse representation for wireless communi-cations: a compressive sensing approach, IEEE Signal Process. Mag. 35 (3) (2018) 40-58.

[10]	P.-H. Kuo, H.T. Kung, P.-A. Ting, Compressive sensing based channel feedback pro-tocols for spatially-correlated massive antenna arrays, in: 2012 IEEE Wireless Com-munications and Networking Conference (WCNC), 2012, pp. 492-497.

[11]	I. Daubechies, M. Defrise, C. De Mol, An iterative thresholding algorithm for linear inverse problems with a sparsity constraint, Commun. Pure Appl. Math. 57 (11) (2004) 1413-1457.

[12]	D.L. Donoho, A. Maleki, A. Montanari, Message passing algorithms for compressed sensing: II. analysis and validation, in: 2010 IEEE Information Theory Workshop on Information Theory (ITW 2010, Cairo, 2010, pp. 1-5.

[13]	C. Li, W. Yin, Y. Zhang, Tval3: Tv minimization by augmented Lagrangian and al-ternating direction algorithms, Tech. Rep, Department of Computer and Applied Mathematics, Rice University, 2009.

[14]	C.A. Metzler, A. Maleki, R.G. Baraniuk, From denoising to compressed sensing, IEEE Trans. Inf. Theory 62 (9) (2016) 5117-5144.

[15]	C.-K. Wen, W.-T. Shih, S. Jin, Deep learning for massive mimo csi feedback, IEEE Wirel. Commun. Lett. 7 (5) (2018) 748-751.

[16]	Z. Wang, C. Qing, L. He, H. Jiang, Y. Yang, Y. Sun, Sensing information-assisted csi feedback, IEEE Commun. Lett. 29 (2) (2025) 313-317.

[17]	C. Qing, Z. Wang, Q. Ye, W. Liu, J. Wang, X. Cai, Gats-net: Gat-based superimposed csi feedback, IEEE Trans. Green Commun. Netw. 9 (2) (2025) 536-548.

[18]	Q. Chaojin, Y. Qing, L. Wenhui, W. Zilong, W. Jiafan, C. Jinliang, Los sensing-based superimposed csi feedback for uav-assisted mmwave systems, Chin. J. Aeronaut. 36 (12) (2023) 349-360.

[19]	J. Guo, C.-K. Wen, S. Jin, G.Y. Li, Convolutional neural network-based multiple-rate compressive sensing for massive mimo csi feedback: design, simulation, and analysis, IEEE Trans. Wirel. Commun. 19 (4) (2020) 2827-2840.

[20]	P. Liang, J. Fan, W. Shen, Z. Qin, G.Y. Li, Deep learning and compressive sensing-based csi feedback in fdd massive mimo systems, IEEE Trans. Veh. Technol. 69 (8) (2020) 9217-9222.

[21]	Z. Lu, J. Wang, J. Song,Multi-resolution csi feedback with deep learning in massive mimo system, in: ICC 2020-2020 IEEE International Conference on Communications (ICC), 2020, pp. 1-6.

[22]	b. Cheng j. Zhao y. Hu, Multi-scale and multi-channel networks for csi feedback in massive mimo system, J. Comput. Commun. 9 (10) (2021) 132-141.

[23]	Q. Cai, C. Dong, K. Niu, Attention model for massive mimo csi compression feedback and recovery, in: 2019 IEEE Wireless Communications and Networking Conference (WCNC), 2019, pp. 1-5.

[24]	Y. Xu, M. Yuan, M.-O. Pun,Transformer empowered csi feedback for massive mimo systems, in:2021 30th Wireless and Optical Communications Conference (WOCC), 2021, pp. 157-161.

[25]	Y. Cui, A. Guo, C. Song, Transnet: full attention network for csi feedback in fdd massive mimo system, IEEE Wirel. Commun. Lett. 11 (5) (2022) 903-907.

[26]	B. Zhou, S. Ma, G. Yang,Transformer-based csi feedback with hybrid learnable non-uniform quantization for massive mimo systems, in:2023 32nd Wireless and Optical Communications Conference (WOCC), 2023, pp. 1-5.

[27]	W. Shan, C. Huang, C. Zhu, Q. Yang, Z. Yang, Attention guided deep learning for csi compression and feedback, in: 2023 IEEE 23rd International Conference on Commu-nication Technology (ICCT), 2023, pp. 1608-1613.

[28]	X. Zhang, J. Wang, Z. Lu, H. Zhang, Continuous online learning-based csi feedback in massive mimo systems, IEEE Commun. Lett. 28 (3) (2024) 557-561.

[29]	B. Zhang, H. Li, X. Liang, X. Gu, L. Zhang, Model transmission-based online updat-ing approach for massive mimo csi feedback, IEEE Commun. Lett. 27 (6) (2023) 1609-1613.

[30]	Y. Cui, J. Guo, C.-K. Wen, S. Jin, S. Han, Unsupervised online learning in deep learning-based massive mimo csi feedback, IEEE Commun. Lett. 26 (9) (2022) 2086-2090.

[31]	J. Guo, Y. Zuo, C.-K. Wen, S. Jin, User-centric online gossip training for autoencoder-based csi feedback, IEEE J. Sel. Top. Signal Process. 16 (3) (2022) 559-572.

[32]	S. Yang, W. Lyu, B. Ning, Z. Zhang, C. Yuen, Flexible precoding for multi-user mov-able antenna communications, IEEE Wirel. Commun. Lett. 13 (5) (2024) 1404-1408.

[33]	A. Krishnamoorthy, R. Schober, Downlink massive mu-mimo with successively-regularized zero forcing precoding, IEEE Wirel. Commun. Lett. 12 (1) (2023) 114-118.

[34]	H. Gui, Y. Xie, Q. Song, Y. Qian, R. Zhang, Improved sic-svd hybrid precoding algo-rithm for indoor sub-thz systems, IEEE Access 11 (2023) 93691-93700.

[35]	D.P. Kingma,J. Ba, Adam: a method for stochastic optimization, arXiv preprint, arXiv:1412.6980.

[36]	S. Ioﬀe, C. Szegedy, Batch normalization: accelerating deep network training by re-ducing internal covariate shift, CoRR, arXiv:1502.03167.

[37]	R. Xiong, Y. Yang, D. He, K. Zheng, S. Zheng, C. Xing, H. Zhang, Y. Lan, L. Wang, T. Liu, On layer normalization in the transformer architecture, CoRR, arXiv:2002. 04745.

[38]	W. Liu, W. Tian, H. Xiao, S. Jin, X. Liu, J. Shen, Evcsinet: eigenvector-based csi feedback under 3gpp link-level channels, IEEE Wirel. Commun. Lett. 10 (12) (2021) 2688-2692.