基扩展模型下基于LSTM神经网络的时变信道预测方法

doi:10.12305/j.issn.1001-506X.2022.09.33

摘要/Abstract

摘要：

针对高速移动正交频分复用系统, 提出了一种基扩展模型(basis expansion model, BEM)下基于长短期记忆(long short-term memory, LSTM)神经网络的时变信道预测方法。为了降低传统BEM的建模误差, 根据高速移动环境中不同列车在相同位置处的无线信道具有强相关性的特点, 首先基于历史时刻的信道状态信息获取最优的基函数, 并利用该基函数对信道进行建模。然后, 通过LSTM神经网络对信道基系数进行线下训练与线上预测来获取未来时刻信道信息, 大大降低了计算复杂度。在线下训练中, 将网络的逼近目标设置为信道估计值, 而不是理想的信道信息, 以增强预测模型的实用性。仿真结果表明, 相比现有方法, 新方法的计算复杂度较低, 且预测精度较高。

关键词: 高速移动, 长短期记忆神经网络, 基扩展模型, 时变信道预测

Abstract:

For high-speed mobile orthogonal frequency division multiplexing system, a time-varying channel prediction method based on long short-term memory (LSTM) neural network under basis expansion model (BEM) is proposed. To reduce the modeling error of the traditional BEM, according to the strong correlation characteristics of the wireless channel at the same location for the different trains in a high-speed mobile environment, the optimal basis function is obtained by the channel sate information of historical time, and it is used to model the channel. Then, the channel information at the future time is obtained by the offline training and online prediction of the channel base coefficient via LSTM neural network, which greatly reduces the computational complexity. In offline training, to enhance the practicality of the prediction model, the channel estimation, rather than the ideal channel information, is set to the approximation objective of the network. The simulation results show that the proposed method has lower computational complexity and better prediction accuracy comparing with the existing methods.

Key words: high-speed mobile, long short-term memory (LSTM) neural network, basis expansion model (BEM), time-varying channel prediction

中图分类号:

TN929.5

聂倩, 杨丽花, 呼博, 任露露. 基扩展模型下基于LSTM神经网络的时变信道预测方法[J]. 系统工程与电子技术, 2022, 44(9): 2971-2977.

Qian NIE, Lihua YANG, Bo HU, Lulu REN. Time-varying channel prediction method based on LSTM neural networks under basis expansion model[J]. Systems Engineering and Electronics, 2022, 44(9): 2971-2977.

图/表 7

图1

表1

图2

表2

图3

图4

表3

参考文献 30

1	CHEN J Q , GE X H , NI Q . Coverage and Handoff analysis of 5G fractal small cell networks[J]. IEEE Trans.on Wireless Communications, 2019, 18 (2): 1263- 1276. doi: 10.1109/TWC.2018.2890662
2	陈中康. 基于深度学习的高铁移动通信信道预测和信号检测研究[D]. 南京: 南京邮电大学, 2020.
	CHEN Z K. Research on deep learning-based channel prediction and signal detection in high-speed rail mobile communication scenarios[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2020.
3	WEI J , HANS D S . Deep learning for fading channel prediction[J]. IEEE Open Journal of the Communications Society, 2020, 1, 320- 332. doi: 10.1109/OJCOMS.2020.2982513
4	SHARMA P , CHANDRA K . Prediction of state transitions in Rayleigh fading channels[J]. IEEE Trans.on Vehicular Technology, 2007, 56 (2): 416- 425. doi: 10.1109/TVT.2007.891421
5	DONG Z X, ZHAO Y S, CHEN Z H. Support vector machine for channel prediction in high-speed railway communication systems[C]//Proc. of the IEEE MTT-S International Wireless Symposium, 2018.
6	TANG Q, LONG H, YANG H J, et al. An enhanced LMMSE channel estimation under high speed railway scenarios[C]//Proc. of the IEEE International Conference on Communications Workshops, 2017: 999-1004.
7	LIAO R F , WEN H , WU J S , et al. The Rayleigh fading channel prediction via deep learning[J]. Wireless Communications and Mobile Computing, 2018, doi: 10.1155/2018/6497340
8	WEI J, HANS D S. Recurrent neural networks with long short-term memory for fading channel prediction[C]//Proc. of the IEEE 91st Vehicular Technology Conference, 2020.
9	WEI J, HANS D S. A deep learning method to predict fading channel in multi-antenna systems[C]//Proc. of the IEEE 91st Vehicular Technology Conference, 2020.
10	YANG J, LI L, ZHAO M J. A blind CSI prediction method based on deep learning for V2I millimeter-wave channel[C]//Proc. of the IEEE 28th International Conference on Network Protocols, 2020.
11	LIU G Q , XU Y , HE Z J , et al. Deep learning-based channel prediction for edge computing networks toward intelligent connected vehicles[J]. IEEE Access, 2019, 7, 114487- 114495. doi: 10.1109/ACCESS.2019.2935463
12	KIM H , KIM S , LEE H , et al. Massive MIMO channel prediction: Kalman filtering Vs. machine learning[J]. IEEE Trans.on Communications, 2021, 69 (1): 518- 528. doi: 10.1109/TCOMM.2020.3027882
13	ZHNG Y X , ZHANG J H , YU L . Cluster-based fast time-varying MIMO channel fading prediction in the high-speed scenario[J]. IEEE Access, 2019, 7, 148692- 148705. doi: 10.1109/ACCESS.2019.2946881
14	MA L, XIAO F, LI M Y. Research on time-varying sparse channel prediction algorithm in underwater acoustic channels[C]//Proc. of the 3rd International Conference on Electronic Information Technology and Computer Engineering, 2019: 2014-2018.
15	LIAO Y, SUN G D, SHEN X F, et al. BEM-based channel estimation and interpolation methods for doubly-selective OFDM channel[C]//Proc. of the IEEE International Conference on Smart Internet of Things, 2018: 70-75.
16	SON W S, HAN D S. Analysis on the channel prediction accuracy of deep learning-based approach[C]//Proc. of the International Conference on Artificial Intelligence in Information and Communication, 2021: 140-143.
17	MADHUBABU, THAKRE A. Long-short term memory based channel prediction for SISO system[C]//Proc. of the International Conference on Communication and Electronics Systems, 2019.
18	JOO J , PARK M C , HAN D S , et al. Deep learning-based channel prediction in realistic vehicular communications[J]. IEEE Access, 2019, 7, 27846- 27858. doi: 10.1109/ACCESS.2019.2901710
19	RAMAKRISHNA S, PRIYATAMKUMAR. A comprehensive study of modeling doubly selective channel using basis expansion techniques[C]//Proc. of the International Conference on Wireless Communications, Signal Processing and Networking, 2016: 1822-1824.
20	WANG Y, WANG X Y, LONG K. Fast-varying channel estimation method based on basis expansion models in IEEE 802.16e systems[C]//Proc. of the 5th International Conference on Bio Medical Engineering and Informatics, 2012: 1462-1466.
21	程露. 基于未来高速移动场景的OFDMA系统中时变信道估计方法研究[D]. 南京: 南京邮电大学, 2020.
	CHENG L. Research on time-varying channel estimation method in OFDMA system for future high-speed mobile scenarios[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2020.
22	WANG X Y, WANG G P, SUN J, et al. Channel estimation with new basis expansion model for wireless communications on high speed railways[C]//Proc. of the IEEE 83rd Vehicular Technology Conference, 2016.
23	WANG X Y , WANG G P , FAN R F , et al. Channel estimation with expectation maximization and historical information based basis expansion model for wireless communication systems on high speed railway[J]. IEEE Access, 2018, 6, 72- 80. doi: 10.1109/ACCESS.2017.2745708
24	SHENG Z C, FANG Y, WANG C. A BEM method of channel estimation for OFDM systems in high-speed train environment[C]//Proc. of the International Workshop on High Mobility Wireless Communications, 2013: 6-9.
25	黄锦锦. 基于基扩展模型的高移动性信道估计算法研究[D]. 福州: 福州大学, 2018.
	HUANG J J. Research on high mobility channel estimation algorithm based on basis expansion model[D]. Fuzhou: Fuzhou University, 2018.
26	CAREEM M A A , DUTTA A . Real-time prediction of non-stationary wireless channels[J]. IEEE Trans.on Wireless Communications, 2020, 19 (20): 7836- 7850.
27	SASAKI, KUNO N, NAKAHIRA T, et al. Deep learning based channel prediction at 2-26 GHz band using long short-term memory network[C]//Proc. of the 15th European Confe-rence on Antennas and Propagation, 2021.
28	YUAN J , NGO H Q , MATTHAIOU M . Machine learning-based channel prediction in massive MIMO with channel aging[J]. IEEE Trans.on Wireless Communications, 2020, 19 (5): 2960- 2973. doi: 10.1109/TWC.2020.2969627
29	LEMAYIAN J P, HAMAMREH J M. Recurrent neural network-based channel prediction in mMIMO for enhanced performance in future wireless communication[C]//Proc. of the International Conference on UK-China Emerging Technologies, 2020.
30	赵四方, 马启原, 李铁楠. 基于LSTM网络的短波天波信道分析及预测[J]. 舰船电子工程, 2019, 39 (12): 65- 70.
	ZHAO S F , MA Q Y , LI T N . Analysis and prediction of sky wave channel based on LSTM network[J]. Ship Electronic Engineering, 2019, 39 (12): 65- 70.

参数	数值
OFDMA符号长度(N)	128
循环前缀长度	16
载波频率/GHz	2.35
子载波间隔/kHz	15
列车速度/(km/h)	500
信道模型	5径莱斯信道
莱斯因子	5
基函数个数(Q)	4
隐藏层神经元数目(D₁、D₂)	10
小批量大小	64
神经元损失概率	0.2
最大迭代次数	250

训练样本数	训练时长/s
100	58.105 428
500	317.138 123
1 000	651.253 307
2 000	1 312.550 167

预测方法	复杂度(FLOPs)
文献[7]	获取历史信息	$O\left(22 N_p+N L \log _2(N L)\right) $	6 670
	线下训练	$ O\left(2 U\left(N D_1+D_1 D_2+D_2 N\right)\right)$	5.32×10⁶
	线上测试	$ O\left(2\left(N D_1+D_1 D_2+D_2 N\right)\right)$	5.32×10³
文献[9]	获取历史信息	$O\left(22 N_p+N L \log _2(N L)\right) $	6 670
	线下训练	$\begin{gathered}O\left(2 U L \left(4\left(N\left(D_1+D_2\right)+\left(D_1+D_2\right)^2+\left(D_1+D_2\right)\right)+\right.\right. \\\left.\left.4\left(\left(D_1+D_2\right)^3+\left(D_1+D_2\right)^4+\left(D_1+D_2\right)^2\right)+\left(\left(D_1+D_2\right)^2 N+N\right)\right)\right)\end{gathered} $	7.368 48×10⁹
	线上测试	$ \begin{gathered}O\left(2 L \left(4\left(N\left(D_1+D_2\right)+\left(D_1+D_2\right)^2+\left(D_1+D_2\right)\right)+\right.\right. \\\left.\left.4\left(\left(D_1+D_2\right)^3+\left(D_1+D_2\right)^4+\left(D_1+D_2\right)^2\right)+\left(\left(D_1+D_2\right)^2 N+N\right)\right)\right)\end{gathered}$	7.368 48×10⁶
新方法	获取历史信息	$O\left(N^2+4 L Q N_P+2 L Q\left(N_P-1\right)\right) $	20 184
	线下训练	$ \begin{gathered}O\left(2 U L \left(4\left(Q L\left(D_1+D_2\right)+\left(D_1+D_2\right)^2+\left(D_1+D_2\right)\right)+\right.\right. \\\left.\left.4\left(\left(D_1+D_2\right)^3+\left(D_1+D_2\right)^4+\left(D_1+D_2\right)^2\right)+\left(\left(D_1+D_2\right)^2 N+N\right)\right)\right)\end{gathered}$	1.456 416×10⁹
	线上测试	$\begin{gathered}O\left(2 L \left(4\left(Q L\left(D_1+D_2\right)+\left(D_1+D_2\right)^2+\left(D_1+D_2\right)\right)+\right.\right. \\\left.\left.4\left(\left(D_1+D_2\right)^3+\left(D_1+D_2\right)^4+\left(D_1+D_2\right)^2\right)+\left(\left(D_1+D_2\right)^2 N+N\right)\right)\right)\end{gathered} $	1.456 416×10⁶

[1]	张心宇, 刘源, 宋佳凝. 基于LSTM神经网络的短期轨道预报[J]. 系统工程与电子技术, 2022, 44(3): 939-947.
[2]	张捷, 杨丽花, 聂倩. 新型的基于堆栈式ELM的时变信道预测方法[J]. 系统工程与电子技术, 2022, 44(2): 662-667.
[3]	赵宇, 陈西宏, 薛伦生. OFDM/OQAM系统中基于BEM信道模型的盲载波频偏估计算法[J]. 系统工程与电子技术, 2016, 38(6): 1435-1439.