基于预测原理的回声状态网络在线盲均衡算法

doi:10.11959/j.issn.1000-436x.2020031

通信学报 ›› 2020, Vol. 41 ›› Issue (3): 145-153.doi: 10.11959/j.issn.1000-436x.2020031

基于预测原理的回声状态网络在线盲均衡算法

杨凌,韩琴,程丽,赵傲男,杜娟

兰州大学信息科学与工程学院，甘肃兰州 730000

修回日期:2019-12-20 出版日期:2020-03-25 发布日期:2020-03-31
作者简介:杨凌（1966- ），女，甘肃张掖人，兰州大学副教授、硕士生导师，主要研究方向为机器学习理论与通信系统智能信号处理|韩琴（1996- ），女，山西忻州人，兰州大学硕士生，主要研究方向为储备池计算模型及其在盲均衡中的应用|程丽（1996- ），女，贵州铜仁人，兰州大学硕士生，主要研究方向为极限学习机及其在盲均衡中的应用|赵傲男（1995- ），男，河南周口人，兰州大学硕士生，主要研究方向为神经网络理论及其在盲均衡中的应用|杜娟（1979- ），女，博士，兰州大学讲师、硕士生导师，主要研究方向为非线性电路和系统以及混沌安全通信
基金资助:
甘肃省自然科学基金资助项目(20180322);中央高校基础研究基金资助项目(lzujbky-2019-91)

Online blind equalization algorithm with echo state network based on prediction principle

Ling YANG,Qin HAN,Li CHENG,Aonan ZHAO,Juan DU

Institute of Information Science and Engineering,Lanzhou University,Lanzhou 730000,China

Revised:2019-12-20 Online:2020-03-25 Published:2020-03-31
Supported by:
The Natural Science Foundation of Gansu Province(20180322);The Fundamental Research Funds for the Central Universities(lzujbky-2019-91)

摘要/Abstract

摘要：

针对非线性信道，基于预测原理提出了回声状态网络（ESN）在线盲均衡算法，首先将具有良好非线性映射能力的ESN代替传统的线性预测误差滤波器，并采用递归最小二乘（RLS）算法计算网络的输出权值，使网络预测误差达到最小；然后进行幅值和相位的调整。仿真实验表明，对于16QAM信号，所提算法可以有效降低非线性信道对发送信号产生的畸变。与其他基于预测原理的盲均衡算法相比，所提算法有更低的均方误差值和更快的收敛速度。

关键词: 非线性信道, 预测误差滤波器, 回声状态网络, 递归最小二乘算法

Abstract:

In view of the nonlinear channel,the online blind equalization algorithm with echo state network was proposed based on prediction principle.In the proposed algorithm,the traditional linear prediction error filter was replaced by the ESN with good nonlinear mapping ability,and recursive least square (RLS) algorithm was used to calculate the output weight of the network to minimize the network prediction error.Then,the amplitude and phase were adjusted.Simulation results show that the proposed algorithm can effectively reduce the distortion caused by nonlinear channel to the transmitted signal for 16QAM signal,which has lower mean square error and faster convergence speed in comparison with other blind equalization algorithms based on prediction principle.

Key words: nonlinear channel, predictive error filter, echo state network, recursive least square algorithm

中图分类号:

TN911.5

杨凌,韩琴,程丽,赵傲男,杜娟. 基于预测原理的回声状态网络在线盲均衡算法[J]. 通信学报, 2020, 41(3): 145-153.

Ling YANG,Qin HAN,Li CHENG,Aonan ZHAO,Juan DU. Online blind equalization algorithm with echo state network based on prediction principle[J]. Journal on Communications, 2020, 41(3): 145-153.

图/表 13

图1

图2

表1

表2

表3

表4

表6

表5

图3

图4

图5

表7

Linear-PEF算法、MLP-PEF算法与ESN-PEF算法的复杂度比较"

算法	复杂度
Linear-PEF	O (3L_w+5)
MLP-PEF	$O (2 N_{MLP}^{2} + 6 N_{MLP} + 6)$
ESN-PEF	O(N ³+6N²+11N+5)

表7

图6

参考文献 19

[1]	SATO Y . A method of self-recovering equalization for multilevel amplitude-modulation systems[J]. IEEE Transactions on Communications, 1975,23(6): 679-682.
[2]	DING Z , KENNEDY R A . On the (non)existence of undesirable equilibria of Godard blind equalizers[J]. IEEE Transactions on Signal Processing, 1992,40(10): 2425-2432.
[3]	LEE W , CHEUN K . Convergence analysis of the stop-and-go blind equalization algorithm[J]. IEEE Transactions on Communications, 1999,47(2): 177-180.
[4]	SHARMA V , RAJ V N . Convergence and performance analysis of Godard family and multimodulus algorithms for blind equalization[J]. IEEE Transactions on Signal Processing, 2005,53(4): 1520-1533.
[5]	XUE W , YANG X N , ZHANG Z Y . A stop and go blind equalization algorithm for QAM signals[C]// 2010 6th International Conference on Wireless Communications Networking and Mobile Computing (WiCOM). Piscataway:IEEE Press, 2010: 1-4.
[6]	马思扬, 彭华, 王彬 . 适用于稀疏多径信道的稀疏自适应常模盲均衡算法[J]. 通信学报, 2017,38(1): 149-157.
	MA S J , PENG H , WANG B . Sparse adaptive normal mode blind equalization algorithm for sparse multipath channels[J]. Journal on Communications, 2017,38(1): 149-157.
[7]	赵太飞, 刘龙飞, 王晶 . 无线紫外光散射通信中的改进 CMA-FSE盲均衡算法[J]. 通信学报, 2019,40(3): 102-108.
	ZHAO T F , LIU L F , WANG J . An improved CMA-FSE blind equalization algorithm in wireless ultraviolet light scattering communication[J]. Journal on Communications, 2019,40(3): 102-108.
[8]	SLOCK D T M , . Blind fractionally-spaced equalization,perfect-reconstruction filter banks and multichannel linear prediction[C]// IEEE International Conference on Acoustics,Speech,＆ Signal Processing. Piscataway:IEEE Press, 1994:585.
[9]	FERRARI R , PANAZIO C M , ATTUX R R F ,et al. Unsupervised channel equalization using fuzzy prediction-error filters[C]// IEEE Workshop on Neural Networks for Signal Processing. Piscataway:IEEE Press, 2003: 869-878.
[10]	CAVALCANTE C C , MONTALVAO F J R , DORIZZI B ,et al. A neural predictor for blind equalization of digital communication systems[C]// Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing,Communications,and Control Symposium. Piscataway:IEEE Press, 2000: 347-351.
[11]	XIE N , LEUNG H . Blind equalization using a predictive radial basisfunction neural network[J]. IEEE Transactions on Neural Networks, 2005,16(3): 709-720.
[12]	WADA C , CONSOLARO D M , FERRARI R ,et al. Nonlinear blind source deconvolution using recurrent prediction-error filters and an artificial immune system[C]// Proceedings of the 8th International Conference on Independent Component Analysis and Signal Separation. Berlin:DBLP, 2009: 371-378.
[13]	CHEN L , YANG L , DU J ,et al. An extreme learning machine architecture based on volterra filtering and PCA[J]. IEICE Transactions on Information and Systems, 2017,(11): 2690-2701.
[14]	BOCCATO L , LOPES A , ATTUX R ,et al. An extended echo state network using Volterra filtering and principal component analysis[J]. Neural Networks:the Official Journal of the International Neural Network Society, 2012,32(2): 292-302.
[15]	马涛 . 最简储备池回声状态网络及其应用研究[D]. 兰州:兰州大学, 2013.
	MA T . Minimum reserve pool ESN and its application[D]. Lanzhou:Lanzhou University, 2013.
[16]	MAASS W,NATSCHL?GER T , MARKRAM H . Real-time computing without stable states:a new framework for neural computation based on perturbations[J]. Neural Computation, 2002,14(11): 2531-2560.
[17]	JAEGER H , HAAS H . Harnessing nonlinearity:predicting chaotic systems and saving energy in wireless communication[J]. Science, 2004,304(5667): 78-80.
[18]	SETH S , OZTURK M C , PRINCIPE J C . Signal processing with echo state networks in the complex domain[C]// 2007 IEEE Workshop on Machine Learning for Signal Processing. Piscataway:IEEE Press, 2007: 408-412.
[19]	曾娟, 王颖, 李晓娜 ,等. 基于频域迭代判决反馈均衡的低复杂度FTN接收机[J]. 通信学报, 2017,38(4): 190-198.
	ZENG J , WANG Y , LI X N ,et al. Low-complexity FTN receivers based on frequency domain iterative decision feedback equalization[J]. Journal on Communications, 2017,38(4): 190-198.

储备池规模				MSE/dB
储备池规模	SNR=0 dB	SNR=5 dB	SNR=10 dB	SNR=15 dB	SNR=20 dB	SNR=25 dB	SNR=30 dB	SNR=35 dB
N=5	9.58	4.81	-0.34	-8.31	-18.92	-27.71	-29.12	-29.42
N=10	9.24	4.58	-0.32	-8.71	-19.23	-27.72	-29.20	-29.45
N=20	9.05	4.72	-0.36	-8.51	-19.48	-27.31	-29.15	-29.41
N=50	9.61	4.46	-0.41	-8.42	-19.15	-25.37	-28.01	-28.35

储备池规模	迭代次数/次
N=5	2 000～2 050
N=10	2 000～2 050
N=20	2 000～2 050
N=50	5 500～5 600

激活函数				MSE/dB
激活函数	SNR=0 dB	SNR=5 dB	SNR=10 dB	SNR=15 dB	SNR=20 dB	SNR=25 dB	SNR=30 dB	SNR=35 dB
tanh(?)	9.58	4.81	-0.34	-8.31	-18.92	-27.71	-29.12	-29.42
sinh(?)	9.15	4.69	-0.12	-7.81	-18.95	-26.82	-28.51	-28.61
tribas(?)	9.21	4.72	-0.18	-7.71	-15.88	-21.31	-24.85	-25.41

激活函数	迭代次数/次
tanh(?)	2 000～2 050
sinh(?)	2 000～2 050
tribas(?)	5 000～5 100

读出层函数	迭代次数/次
线性读出函数	2 000～2 050
非线性读出函数	2 000～2 050

基于预测原理的回声状态网络在线盲均衡算法

Online blind equalization algorithm with echo state network based on prediction principle

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 19

相关文章 1

Metrics

推荐阅读 0