Abstract
With the increasing number of antennas at the base station (BS), the feedback overhead of traditional codebook in frequency division duplexing (FDD) mode becomes overwhelming, since the number of codewords in codebook increases quickly. Alternatively, we can directly feedback the channel state information (CSI) to the BS for precoding. To reduce the overhead of CSI feedback, this paper proposes three CSI compression models based on autoencoder network. The first two of them, adopting deep learning (DL) structure, are named FCNet and CNet, respectively. FCNet employs full-connected network architecture, while CNet is designed based on convolutional neural network with lightweight convolution kernels and multi-channel architecture. By applying principal component analysis (PCA) on CSI feedback, the third one, i.e., PCANet, is also studied and analyzed in details. Experiments show that CNet has best accuracy performance at the cost of high computational complexity, while FCNet shows medium accuracy and complexity among the three models. Besides, the accuracy of PCANet is nearly the same as CNet in some specific channel conditions. Compared with the state-of-the-art of CsiNet, the proposed models have their own advantages and limitations in different scenarios.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Since each value in \( \mathbf {s} \) belongs to (0, 1) (the sigmoid function), we minus 0.5 before round function, quantize the rounded value to bits and then transmit them to the decoder. Once the decoder received these bits, it dequantizes them, plus 0.5 and then divide them by \({2^B}\) to get the quantization value.
References
3GPP Technical Specification (TS) 38.211: NR; Physical channels and modulation (Release 15) (2017)
3GPP Technical Specification (TS) 38.214: NR; Physical layer procedure for data (Release 15) (2017)
Zhao, L., Zhao, H., Zheng, K., Xiang, W.: Massive MIMO in 5G Networks: Selected Applications, 1st edn. Springer, Heidelberg (2018). https://doi.org/10.1007/978-3-319-68409-3
Zheng, K., Zhao, L., Mei, J., Shao, B., Xiang, W., Hanzo, L.: Survey of large-scale MIMO systems. IEEE Commun. Surv. Tutor. 17(3), 1738–1760 (2015)
Wen, C., Shih, W., Jin, S.: Deep learning for massive MIMO CSI feedback. IEEE Wirel. Commun. Lett. 7(5), 748–751 (2018)
Akbari, M., Liang, J., Han, J.: DSSLIC: deep semantic segmentation-based layered image compression. In: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, pp. 2042–2046 (2019)
Huang, C., Nguyen, T., Lai, C.: Multi-channel multi-loss deep learning based compression model for color images. In: IEEE International Conference on Image Processing (ICIP), Taipei, pp. 4524–4528 (2019)
Cai, C., Chen, L., Zhang, X., Gao, Z.: Efficient variable rate image compression with multi-scale decomposition network. IEEE Trans. Circuits Syst. Video Technol. 29(12), 3687–3700 (2019)
Guo, J., Wen, C., Jin, S., Li, G.Y.: Convolutional neural network-based multiple-rate compressive sensing for massive MIMO CSI feedback: design, simulation, and analysis. IEEE Trans. Wirel. Commun. 19(4), 2827–2840 (2020)
Lu, Z., Wang, J., Song, J.: Multi-resolution CSI feedback with deep learning in massive MIMO system. In: IEEE International Conference on Communications (ICC), Dublin, pp. 1–6 (2020)
Qing, C., Cai, B., Yang, Q., Wang, J., Huang, C.: Deep learning for CSI feedback based on superimposed coding. IEEE Access 7, 93723–93733 (2019)
Liu, Y., You, S.D., Chung, C.: Channel estimation method for LTE downlink MIMO transmission. In: IEEE 8th Global Conference on Consumer Electronics (GCCE), Osaka, pp. 719–722 (2019)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, pp. 770–778 (2016)
Chen, T., Guo, J., Jin, S., Wen, C., Li, G.Y.: A novel quantization method for deep learning-based massive MIMO CSI feedback. In: IEEE Global Conference on Signal and Information Processing (GlobalSIP), Ottawa, pp. 1–5 (2019)
Acknowledgment
This work was supported in part by the China Natural Science Funding under Grant 61731004.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering
About this paper
Cite this paper
He, Z., Zhao, L., Luo, X., Cheng, B. (2022). Deep CSI Feedback for FDD MIMO Systems. In: Gao, H., Wun, J., Yin, J., Shen, F., Shen, Y., Yu, J. (eds) Communications and Networking. ChinaCom 2021. Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, vol 433. Springer, Cham. https://doi.org/10.1007/978-3-030-99200-2_28
Download citation
DOI: https://doi.org/10.1007/978-3-030-99200-2_28
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-99199-9
Online ISBN: 978-3-030-99200-2
eBook Packages: Computer ScienceComputer Science (R0)