Deep residual systolic network for massive MIMO channel estimation by joint training strategies of mixed-SNR and mixed-scenarios

doi:10.23919/JSEE.2024.000056

Journal of Systems Engineering and Electronics ›› 2025, Vol. 36 ›› Issue (4): 903-913.doi: 10.23919/JSEE.2024.000056

• ELECTRONICS TECHNOLOGY • Previous Articles

Deep residual systolic network for massive MIMO channel estimation by joint training strategies of mixed-SNR and mixed-scenarios

Meng SUN(), Qingfeng JING(), Weizhi ZHONG()

College of Astronautics, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China

Received:2023-02-14 Online:2025-08-18 Published:2025-09-04
Contact: Qingfeng JING E-mail:1352269658@qq.com;jing_nuaa@163.com;zhongwz@nuaa.edu.cn
About author:
SUN Meng was born in 1998. He received his Bachelor’s degree from the College of Astronautics, Nanjing University of Aeronautics and Astronautics in 2021. He has been a Master’s student at Nanjing University of Aeronautics and Astronautics since 2021. His research interests are channel estimation and artificial intelligence.E-mail: 1352269658@qq.com

JING Qingfeng was born in 1981. He received his Master and Ph.D. degrees in the School of Electronics and Information Engineering, Harbin Institute of Technology in 2005 and 2009, respectively. He has been working as an associate professor since 2009 in the College of Astronautics, Nanjing University of Aeronautics and Astronautics. His research interests are digital signal processing, satellite communication, and broadband multi-carrier communication.E-mail: jing_nuaa@163.com

ZHONG Weizhi was born in 1980. She received her B.S. and M.S. degrees in communication and information system from Jilin University in 2006, and Ph.D. degree in communication and information system from Harbin Institute of Technology in 2010, respectively. She is an associate professor in the College of Astronautics, Nanjing University of Aeronautics and Astronautics. Her research interests include millimeter wave communication for the fifth generation (5G), massive multiple-input multiple-output technique and beamforming and beam tracking technique.E-mail: zhongwz@nuaa.edu.cn
Supported by:
This work was supported by the National Key Scientific Instrument and Equipment Development Project (61827801).

Abstract

Abstract:

The fifth-generation (5G) communication requires a highly accurate estimation of the channel state information (CSI) to take advantage of the massive multiple-input multiple-output (MIMO) system. However, traditional channel estimation methods do not always yield reliable estimates. The methodology of this paper consists of deep residual shrinkage network (DRSN) neural network-based method that is used to solve this problem. Thus, the channel estimation approach, based on DRSN with its learning ability of noise-containing data, is first introduced. Then, the DRSN is used to train the noise reduction process based on the results of the least square (LS) channel estimation while applying the pilot frequency subcarriers, where the initially estimated subcarrier channel matrix is considered as a three-dimensional tensor of the DRSN input. Afterward, a mixed signal to noise ratio (SNR) training data strategy is proposed based on the learning ability of DRSN under different SNRs. Moreover, a joint mixed scenario training strategy is carried out to test the multi scenarios robustness of DRSN. As for the findings, the numerical results indicate that the DRSN method outperforms the spatial-frequency-temporal convolutional neural networks (SF-CNN) with similar computational complexity and achieves better advantages in the full SNR range than the minimum mean squared error (MMSE) estimator with a limited dataset. Moreover, the DRSN approach shows robustness in different propagation environments.

Key words: massive multiple-input multiple-output (MIMO), channel estimation, deep residual shrinkage network (DRSN), deep convolutional neural network (CNN)

Meng SUN, Qingfeng JING, Weizhi ZHONG. Deep residual systolic network for massive MIMO channel estimation by joint training strategies of mixed-SNR and mixed-scenarios[J]. Journal of Systems Engineering and Electronics, 2025, 36(4): 903-913.

Figures/Tables 17

Fig 1

Fig 2

Fig 3

Fig 4

Table 1

Table 2

Table 3

Table 4

Table 5

Fig 5

Table 6

Fig 6

Table 7

Fig 7

Fig 8

Fig 9

Fig 10

References 23

1	ALKHATEEB A, EL AYACH O, LEUS G, et al Channel estimation and hybrid precoding for millimeter wave cellular systems. IEEE Journal of Selected Topics in Signal Processing, 2014, 8 (5): 831- 846. doi: 10.1109/JSTSP.2014.2334278
2	GAO Z, HU C, DAI L L, et al Channel estimation for millimeter-wave massive MIMO with hybrid precoding over frequency-selective fading channels. IEEE Communications Letters, 2016, 20 (6): 1259. doi: 10.1109/LCOMM.2016.2555299
3	WANG Y C, XU W, ZHANG H, et al Wideband mmWave channel estimation for hybrid massive MIMO with low-precision ADCs. IEEE Wireless Communications Letters, 2018, 8 (1): 285- 288.
4	ZHOU Z, FANG J, YANG L X, et al Low-rank tensor decomposition-aided channel estimation for millimeter wave MIMO-OFDM systems. IEEE Journal on Selected Areas in Communications, 2017, 35 (7): 1524- 1538. doi: 10.1109/JSAC.2017.2699338
5	GHAUCH H, KIM T, BENGTSSON M, et al Subspace estimation and decomposition for large millimeter-wave MIMO systems. IEEE Journal of Selected Topics in Signal Processing, 2017, 10 (3): 528- 542.
6	QIN Z, YE H, LI G Y, et al Deep learning in physical layer communications. IEEE Wireless Communications, 2019, 26 (2): 93- 99. doi: 10.1109/MWC.2019.1800601
7	HE H, WEN C K, JIN S, et al Deep learning-based channel estimation for beamspace mmWave massive MIMO systems. IEEE Wireless Communications Letters, 2018, 7 (5): 852- 855. doi: 10.1109/LWC.2018.2832128
8	WANG T, WEN C K, JIN S, et al Deep learning-based CSI feedback approach for time-varying massive MIMO channels. IEEE Wireless Communications Letters, 2018, 8 (2): 416- 419.
9	HE H, JIN S, WEN C K, et al Model-driven deep learning for physical layer communications. IEEE Wireless Communications, 2019, 26 (5): 77- 83. doi: 10.1109/MWC.2019.1800447
10	KANG J M, CHUN C J, KIM I M Deep-learning-based channel estimation for wireless energy transfer. IEEE Communications Letters, 2018, 22 (11): 2310- 2313. doi: 10.1109/LCOMM.2018.2871442
11	DONG P, ZHANG H, LI G Y. Machine learning prediction based CSI acquisition for FDD massive MIMO downlink. Proc. of the IEEE Global Communications Conference, 2018. DOI: 10.1109/GLOCOM.2018.8647328.
12	JEON Y S, HONG S N, LEE N. Blind detection for MIMO systems with low-resolution ADCs using supervised learning. Proc. of the IEEE International Conference on Communications , 2017. DOI: 10.1109/ICC.2017.7997434.
13	SOLTANI M, POURAHMADI V, MIRZAEI A, et al Deep learning-based channel estimation. IEEE Communications Letters, 2019, 23 (4): 652- 655. doi: 10.1109/LCOMM.2019.2898944
14	DONG P, ZHANG H, LI G Y, et al Deep CNN-based channel estimation for mmWave massive MIMO systems. IEEE Journal of Selected Topics in Signal Processing, 2019, 13 (5): 989- 1000. doi: 10.1109/JSTSP.2019.2925975
15	JIANG P, WEN C K, JIN S, et al Dual CNN-based channel estimation for MIMO-OFDM systems. IEEE Trans. on Communications, 2021, 69 (9): 5859- 5872. doi: 10.1109/TCOMM.2021.3085895
16	MARINBERG B, COHEN A, BEN-DROR E, et al. A study on MIMO channel estimation by 2D and 3D convolutional neural networks. Proc. of the IEEE International Conference on Advanced Networks and Telecommunications Systems, 2020. DOI:10.1109/ANTS50601.2020.9342797.
17	GE L J, GUO Y C, ZHANG Y, et al Deep neural network based channel estimation for massive MIMO-OFDM systems with imperfect channel state information. IEEE Systems Journal, 2021, 16 (3): 4675- 4685.
18	ZHANG J, MA X L, QI J, et al Designing tensor-train deep neural networks for time-varying MIMO channel estimation. IEEE Journal of Selected Topics in Signal Processing, 2021, 15 (3): 759- 773. doi: 10.1109/JSTSP.2021.3051490
19	ZHAO M H, ZHONG S S, FU X Y, et al Deep residual shrinkage networks for fault diagnosis. IEEE Trans. on Industrial Informatics, 2019, 16 (7): 4681- 4690.
20	DAHLMAN E, PARKVALL S, SKOLD J. 5G NR: the next generation wireless access technology. Cambridge: Academic Press, 2020.
21	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition. Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770−778.
22	CHEN L, DAI X R, ZHANG X F Joint angle and frequency estimation for linear array: an extended DOA-matrix method. Journal of Systems Engineering and Electronics, 2022, 33 (4): 887- 895. doi: 10.23919/JSEE.2022.000086
23	SUN S L, LIU S, WANG J, et al Joint polarization and DOA estimation based on improved maximum likelihood estimator and performance analysis for conformal array. Journal of Systems Engineering and Electronics, 2023, 34 (6): 1490- 1500. doi: 10.23919/JSEE.2023.000048

Scenario parameter	Statistical compoment	UMi-street Canyon
Scenario parameter	Statistical compoment	LoS	NLoS	O2I
DS	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	0.24log₁₀ (1+$ {f}_{c} $)−7.14	−0.24log₁₀ (1+$ {f}_{c} $)−6.83	−6.62
DS	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	0.38	0.16log₁₀ (1+$ {f}_{c} $)+0.28	0.32
ASD	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	1.25
ASD	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{D}} $	0.41	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	0.42
ASA	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{A}} $	−0.08log₁₀ (1+$ {f}_{c} $)+1.73	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	1.76
ASA	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{A}} $	0.014log₁₀ (1+$ {f}_{c} $)+0.28	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	0.16
ZSA	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{Z}\mathrm{S}\mathrm{A}} $	−0.1log₁₀ (1+$ {f}_{c} $)+0.73	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	1.01
ZSA	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{Z}\mathrm{S}\mathrm{A}} $	−0.04log₁₀ (1+$ {f}_{c} $)+0.34	−0.05log₁₀ (1+$ {f}_{c} $) + 1.21	0.43
SF/dB	$ {\sigma }_{{\mathrm{SF}}} $	4	6	7
K/dB	$ {\mu }_{K} $	9	−	−
K/dB	$ {\sigma }_{K} $	5	−	−

Scenario parameter	Statistical component	UMa
Scenario parameter	Statistical component	LoS	NLoS	O2I
DS	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	−6.955-0.0963log₁₀ ($ {f}_{c} $)	−6.28-0.204log₁₀ ($ {f}_{c} $)	−6.62
DS	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	0.66	0.39	0.32
ASD	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{D}\mathrm{S}} $	1.06-0.1114log₁₀ ($ {f}_{c} $)	1.5-0.1144log₁₀ ($ {f}_{c} $)	1.25
ASD	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{D}} $	0.28	0.28	0.42
ASA	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{A}} $	1.81	2.08-0.27log₁₀ ($ {f}_{c} $)	1.76
ASA	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{A}\mathrm{S}\mathrm{A}} $	0.20	0.11	0.16
ZSA	$ {\mu }_{\mathrm{l}\mathrm{g}\mathrm{Z}\mathrm{S}\mathrm{A}} $	0.95	−0.3236log₁₀ (1+$ {f}_{c} $) + 1.512	1.01
ZSA	$ {\sigma }_{\mathrm{l}\mathrm{g}\mathrm{Z}\mathrm{S}\mathrm{A}} $	0.16	0.16	0.43
SF/dB	$ {\sigma }_{{\mathrm{SF}}} $	4	6	7
K/dB	$ {\mu }_{K} $	9	−	−
K/dB	$ {\sigma }_{K} $	5	−	−

Layer	Type	Number of filters	Size of filers	Activation function
1	Input	−	−	−
2	Conv	64	3×3	ReLU
3−10	RSBU.conv	64	3×3	ReLU
11	Output	2$ Q $	3×3	tanh

Devices and software	Environment
Operating platform	Windows 11
Calculation software	Python 3.8
Deep learning framework	Keras
GPU	CUDA 12.1-based RTX3070

Approach	SNR = −10 dB	SNR=0 dB	SNR=10 dB
LS	9.538	0.954	0.0954
MMSE	0.803	0.230	0.0372
DRSN	0.321	0.0916	0.0313

Deep residual systolic network for massive MIMO channel estimation by joint training strategies of mixed-SNR and mixed-scenarios

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

Share this article

Figures/Tables 17

References 23

Related Articles 15

Recommended Articles

Metrics

Comments

Training strategy	SNR distribution
DRSN-1	N_{−10 dB}:N_{−5 dB}:N_{0 dB}:N_{5 dB}:N_{10 dB}=1:1:1:1:6
DRSN-2	N_{−10 dB}:N_{−5 dB}:N_{0 dB}:N_{5 dB}:N_{10 dB}=1:1:1:1:16
DRSN-3	N_{−5 dB}:N_{0 dB}:N_{10 dB}=1:1:198

[1]	Yan LIU, Jianxin YI, Xianrong WAN, Yunhua RAO, Caiyong HAO. Channel estimation in integrated radar and communication systems with power amplifier distortion [J]. Journal of Systems Engineering and Electronics, 2024, 35(5): 1098-1108.
[2]	A. Khoso IMRAN, Xiaofei ZHANG, Hayee Shaikh ABDUL, A. Khoso IHSAN, Ahmed Dayo ZAHEER. Low-complexity signal detection for massive MIMO systems via trace iterative method [J]. Journal of Systems Engineering and Electronics, 2024, 35(3): 549-557.
[3]	Siting LYU, Xiaohui LI, Tao FAN, Jiawen LIU, Mingli SHI. Deep learning for fast channel estimation in millimeter-wave MIMO systems [J]. Journal of Systems Engineering and Electronics, 2022, 33(6): 1088-1095.
[4]	Yumeng SU, Hongyuan GAO, Shibo ZHANG. Energy-efficient resource management for CCFD massive MIMO systems in 6G networks [J]. Journal of Systems Engineering and Electronics, 2022, 33(4): 877-886.
[5]	Yasong LUO, Shengliang HU, Chengxu FENG, Jijin TONG. Power optimization algorithm for OFDM underwater acoustic communication using adaptive channel estimation [J]. Journal of Systems Engineering and Electronics, 2019, 30(4): 662-671.
[6]	Ruoyu ZHANG, Honglin ZHAO, Jiayan ZHANG, Shaobo JIA. Hybrid orthogonal and non-orthogonal pilot distribution based channel estimation in massive MIMO system [J]. Journal of Systems Engineering and Electronics, 2018, 29(5): 881-898.
[7]	Jiansheng HU, Zuxun SONG, Shuxia GUO, Qian ZHANG, Dongdong SHUI. Sparse channel recovery with inter-carrier interference self-cancellation in OFDM [J]. Journal of Systems Engineering and Electronics, 2018, 29(4): 676-683.
[8]	Xu Wang, Tao Yang, and Bo Hu. Low-complexity fractional phase estimation for totally blind channel estimation [J]. Journal of Systems Engineering and Electronics, 2015, 26(2): 232-240.
[9]	Xiaofei Zhang and Dazhuan Xu. Blind channel estimation for multiple antenna OFDM system subject to unknown carrier frequency offset [J]. Journal of Systems Engineering and Electronics, 2014, 25(5): 721-727.
[10]	Donghua Chen and Hongbing Qiu. Mobile channel estimation for MU-MIMO systems using KL expansion based extrapolation [J]. Journal of Systems Engineering and Electronics, 2012, 23(3): 349-354.
[11]	Peng Xu, Jinkuan Wang, and Feng Qi. Improved H-infinity channel estimator based on EM for MIMO-OFDM systems [J]. Journal of Systems Engineering and Electronics, 2011, 22(4): 572-578.
[12]	Yang Zhang, Jiandong Li, and Lihua Pang. Hybrid pilots assisted channel estimation algorithm for MIMO-OFDM systems [J]. Journal of Systems Engineering and Electronics, 2010, 21(5): 721-728.
[13]	Xiaoping Zhou, Yong Fang, and Min Wang. Compressed sensing based channel estimation for fast fading OFDM systems [J]. Journal of Systems Engineering and Electronics, 2010, 21(4): 550-556.
[14]	Maoge Xu and Yaoliang Song. Bayesian sequential state estimation for MIMO-OFDM systems [J]. Journal of Systems Engineering and Electronics, 2010, 21(1): 148-153.
[15]	Pang Jiyong, Li Jiandong, Lu Zhuo, Zhao Linjing & Chen Liang. Optimal training sequences for MIMO systems under correlated fading [J]. Journal of Systems Engineering and Electronics, 2008, 19(1): 33-38.