• Login
  • Register
  • Search

Research on air interface performance based on AI

Wei Zhu, Danhuai Zhao, Xiaolin Bi, Xiaoqiu Xu


with the continuous development of artifi cial intelligence, the excellent nonlinear processing ability of deep neural network
shows important application potential in wireless air interface, which is of great significance to the research of 6G technology. As the
wireless environment becomes more and more complex, the performance of traditional wireless air interface is more and more unable to
meet. How to improve the performance of wireless air interface has become an important issue. In this paper, an enhancement method based
on deep neural network is proposed for CSI prediction, channel estimation and polar code decoding, and its performance is verifi ed by


deep learning; CSI forecast; Channel estimation; Polar code

Full Text:


Included Database


[1]Rp-212708New Sid on ai/ml for NR air interface, moderator (Qualcomm) [s], 3GPP TSG ran94e, electronic meeting, 2021: 1-22

[2]W. Jiang, H. D. schottenNeural network based channel prediction and its performance in multi antenna systems[c], 2018 IEEE 88th vehicular Technology

Conference (VTC fall), 2018: 1-6

[3]W. Jiang, H. D. schottenRecurrent neural network based frequency domain channel prediction for broadband communications[c]2019 IEEE 89th vehicular

Technology Conference (vtc2019 spring), 2019: 1-6

[3]S. Hochreiter, J. schmidhuberLong short term memory[J]Neural computing, 1997, 9 (8): 1735-1780

[4]M. Honkala, D. korpi, J. M. J. huttunenDeeprx: fully revolutionary deep learning receiver[J]IEEE Transactions on wireless communications, 2021, 20 (6):


[5]M. E. morocho cayamcela, H. Lee, and w. limMachine learning for 5g/b5g mobile and wireless communications: potential, limitations, and future

direction[J]IEEE access, 2019, 7 (1): 137184-137206

[6]Technical specifi cation group radio access network; NR; Physical channels and modulation, document TS 38.211 version 16.1.0 release 16[s], 3GPP, ETSI,

Sophia Antipolis Cedex, France, 2020: 1-136

[7]Z. Chang, Y. Wang, H. Li, and Z. wangComplex CNN based equalization for communication signal[c]IEEE 4th int. conf. signal image process (icsip),

2019: 513-517

[8]H. He, S. Jin, C. K. Wen, F. Gao, G. y. Li, and Z. xuModel driven deep learning for physical layer communications[c]IEEE wireless commun, 2019, 26 (5):

77 – 83

[9]F. CholletXception: deep learning with profound separable confl icts[c]IEEE conf. computeVisPattern recognizit(CVPR), 2017: 1800-1807

[10]E. ArikanChannel polarization: a method for constructing capacity achieving codes for symmetric binary input memoryless channels[J]IEEE Transactions

on information theory, 2009, 55 (7): 3051-3073

[11]B. Yuan and K. K. parhiEarly stopping criteria for energy effi cient low latency belief propagation polar code decoders[J]IEEE transSignal process, 2014,

62 (24): 6496-6506

[12]J. Yang, C. Zhang, H. Zhou, et al. pipelined belief propagation polar decoders[c]2016 IEEE International Symposium on circuits and systems (ISCAS),

2016: 413-416

[13]W. Xu, Z. Wu, Y. L. ueng, et al. improved polar decoder based on deep learning[c]2017 IEEE International Workshop on signal processing systems (SIPS),

2017: 1-6

[14]N. Doan, S. A. Hashemi, e. n. mambou, t. tonnellier, and W. J. grossNeural belief dissemination coding of CRC polar concatenated codes[c]IEEE int.

conf. commun(ICC), 2019: 1-6

DOI: http://dx.doi.org/10.18686/modern-management-forum.v6i11.6654