SCIENCE CHINA Information Sciences, Volume 64 , Issue 8 : 182304(2021) https://doi.org/10.1007/s11432-020-2993-6

Deep learning based user scheduling for massive MIMO downlink system

More info
  • ReceivedMar 28, 2020
  • AcceptedJul 20, 2020
  • PublishedJun 1, 2021



This work was supported by National Natural Science Foundation of China (Grant Nos. 61831013, 61971126, 61941104, 61921004).


[1] Rusek F, Persson D, Buon Kiong Lau D. Scaling Up MIMO: Opportunities and Challenges with Very Large Arrays. IEEE Signal Process Mag, 2013, 30: 40-60 CrossRef ADS arXiv Google Scholar

[2] Gao X Q, You X H, Cao S M. 5G移动通信发展趋势与若干关键技术. Sci Sin-Inf, 2014, 44: 551-563 CrossRef Google Scholar

[3] Wang D, Zhang Y, Wei H. An overview of transmission theory and techniques of large-scale antenna systems for 5G wireless communications. Sci China Inf Sci, 2016, 59: 081301 CrossRef Google Scholar

[4] Marzetta T L. Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas. IEEE Trans Wireless Commun, 2010, 9: 3590-3600 CrossRef Google Scholar

[5] Li X, Liu Z, Qin N. FFR Based Joint 3D Beamforming Interference Coordination for Multi-Cell FD-MIMO Downlink Transmission Systems. IEEE Trans Veh Technol, 2020, 69: 3105-3118 CrossRef Google Scholar

[6] Adhikary A, Junyoung Nam A, Jae-Young Ahn A. Joint Spatial Division and Multiplexing-The Large-Scale Array Regime. IEEE Trans Inform Theor, 2013, 59: 6441-6463 CrossRef Google Scholar

[7] Li X, Sun T, Qin N. User scheduling for downlink FD-MIMO systems under Rician fading exploiting statistical CSI. Sci China Inf Sci, 2018, 61: 082302 CrossRef Google Scholar

[8] Han Y, Hsu T H, Wen C K. Efficient Downlink Channel Reconstruction for FDD Multi-Antenna Systems. IEEE Trans Wireless Commun, 2019, 18: 3161-3176 CrossRef Google Scholar

[9] Wang T, Wen C K, Wang H. Deep learning for wireless physical layer: Opportunities and challenges. China Commun, 2017, 14: 92-111 CrossRef Google Scholar

[10] Wen C K, Shih W T, Jin S. Deep Learning for Massive MIMO CSI Feedback. IEEE Wireless Commun Lett, 2018, 7: 748-751 CrossRef Google Scholar

[11] He H, Wen C K, Jin S. Deep Learning-Based Channel Estimation for Beamspace mmWave Massive MIMO Systems. IEEE Wireless Commun Lett, 2018, 7: 852-855 CrossRef Google Scholar

[12] He H, Jin S, Wen C K. Model-Driven Deep Learning for Physical Layer Communications. IEEE Wireless Commun, 2019, 26: 77-83 CrossRef Google Scholar

[13] Goodfellow I, Bengio Y I, Courville A. Deep Learning. Cambridge: MIT Press, 2016. Google Scholar

[14] Ye H, Li G Y, Juang B H. Power of Deep Learning for Channel Estimation and Signal Detection in OFDM Systems. IEEE Wireless Commun Lett, 2018, 7: 114-117 CrossRef Google Scholar

[15] Xia W, Zheng G, Zhu Y. A Deep Learning Framework for Optimization of MISO Downlink Beamforming. IEEE Trans Commun, 2020, 68: 1866-1880 CrossRef Google Scholar

[16] Cui W, Shen K, Yu W. Spatial Deep Learning for Wireless Scheduling. IEEE J Sel Areas Commun, 2019, 37: 1248-1261 CrossRef Google Scholar

[17] Li X, Yu X, Sun T. Joint Scheduling and Deep Learning-Based Beamforming for FD-MIMO Systems Over Correlated Rician Fading. IEEE Access, 2019, 7: 118297 CrossRef Google Scholar

[18] Taricco G. Optimum Receiver Design and Performance Analysis of Arbitrarily Correlated Rician Fading MIMO Channels With Imperfect Channel State Information. IEEE Trans Inform Theor, 2010, 56: 1114-1134 CrossRef Google Scholar

[19] Boukhedimi I, Kammoun A, Alouini M S. Multi-Cell MMSE Combining Over Correlated Rician Channels in Massive MIMO Systems. IEEE Wireless Commun Lett, 2020, 9: 12-16 CrossRef Google Scholar

[20] Han Y, Zhang H, Jin S. Investigation of Transmission Schemes for Millimeter-Wave Massive MU-MIMO Systems. IEEE Syst J, 2017, 11: 72-83 CrossRef ADS Google Scholar

[21] Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift. In: Proceedings of International Conference on Machine Learning, Lille, 2015. 448--456. Google Scholar

[22] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016. 770--778. Google Scholar

  • Figure 1

    (Color online) The overall structure of the proposed scheduling neural network.

  • Figure 2

    (Color online) Average sum rate comparison.

  • Figure 3

    (Color online) Average sum rate comparison (with two trained networks).

  • Figure 5

    (Color online) Average sum rate comparison under different number of antenna.

  • Table 1  

    Table 1Layout of the scheduling network

    Layer name Output size Convolution kernel
    Conv1,BN,ReLU $20\times20\times32$ $~~3~\times~3,32$
    Res block: 1–2 $10\times10\times32$ $\left[~{\begin{array}{*{40}{c}}{3~\times~3,32}\\{3~\times~3,32}\end{array}}~\right]~\times~2~$
    Res block: 3–12 $10\times10\times64$ $\left[~{\begin{array}{*{30}{c}}{3~\times~3,64}\\{3~\times~3,64}\end{array}}~\right]~\times~10~$
    Res block: 13–14 $5\times5\times128$ $\left[~{\begin{array}{*{30}{c}}{3~\times~3,128}\\{3~\times~3,128}\end{array}}~\right]~\times~2~$
    AP $1\times128$
    512-d FC $1\times512$
    20-d sigmoid $1\times20$
  • Table 2  

    Table 2Comparison of normalized computation time

    Scheduling algorithm Normalized runtime
    Sum rate based 18.7
    Most dissimilar 12.9
    Scheduling network 1

Contact and support