基于混沌理论与改进回声状态网络的网络流量多步预测

doi:10.11959/j.issn.1000-436x.2016053

摘要/Abstract

摘要：

网络流量预测是网络管理及网络拥塞控制的重要问题，针对该问题提出一种基于混沌理论与改进回声状态网络的网络流量预测方法。首先利用0-1混沌测试法与最大Lyapunov指数法对不同时间尺度下的网络流量样本数据进行分析，确定网络流量在不同时间尺度下都具有混沌特性。将相空间重构技术引入网络流量预测，通过C-C 方法确定延迟时间，G-P算法确定嵌入维数。对网络流量时间序列进行相空间重构之后，利用一种改进的回声状态网络进行网络流量的多步预测。提出一种改进的和声搜索优化算法对回声状态网络的相关参数进行优化以提高预测精度。利用网络流量的公共数据集以及实际数据进行了仿真，结果表明，提出的预测方法具有更高的预测精度以及更小的预测误差。

关键词: 网络流量, 混沌, 回声状态网络, 时间尺度, 预测

Abstract:

Network traffic prediction was an important problem of network management and network congestion control.In order to solve this problem,a network traffic prediction method based on chaos theory and improved echo state net-work was proposed.Firstly,network traffic sample with differe time scale were analyzed by 0-1 test algorithm for chaos and maximum Lyapunov exponent,the calculation results show that the network traff has chaotic characteristics in different time scale.The phase space reconstruction technique was introduced for the prediction of network traffic,the delay time was determined through the C-C method,the embedding dimension was determined through the G-P algo-rithm.Network traffic time series was processed with phase space reconstruction,the multi-step prediction of network traffic was achieved by an improved echo state network.In order to improve the prediction precision,the key dynamic reservoir and prediction parameters of echo state network were optimized by an improved harmony search algorithm.Through the simulation on public and actual network traffic data,the results verify the proposed prediction method has higher prediction accuracy and smaller prediction error.

Key words: network traffic, chaos, echo state network, time scale, prediction

田中大,李树江,王艳红,王向东. 基于混沌理论与改进回声状态网络的网络流量多步预测[J]. 通信学报, 2016, 37(3): 55-70.

Zhong-da TIAN,Shu-jiang LI,Yan-hong WANG,Xiang-dong WANG. Network traffic multi-step prediction based on chaos theory and improved echo state network[J]. Journal on Communications, 2016, 37(3): 55-70.

图/表 28

图2

图1

表1

图3

图4

图5

表2

图6

图7

图8

图9

图10

图11

表3

表4

图12

图13

图14

图15

表5

表6

图16

图17

表7

表8

图18

图19

表9

参考文献 43

[1]	QU H , MA W T , ZHAO J H . Prediction method for network traffic based on maximum correntropy criterion[J]. China Commu ications, 2013(1): 134-145.
[2]	王升辉，裘正定 . 结合多重分形的网络流量非线性预测[J]. 通信学报， 2007,28(2): 45-50. WANG S H , QIU Z D . Network traffic nonlinear prediction combined with mutifractal[J]. Journal on Communications, 2007,28(2): 45-50.
[3]	LANER M , SVOBODA P , RUPP M . Parsimonious fitting of long-range dependent network traffic using ARMA models[J]. EEE Communications Letters, 2013,17(12): 2368-2371.
[4]	ZHOU D D , CHEN S L , DONG S . Network traffic prediction based on ARIMA model[J]. International Journal of Computer Science Issues, 2012,6(9): 106-111.
[5]	WANG J . A process level network traffic prediction algorithm based on ARIMA model in smart substation[C]// 2013 IEEE Int Conf on Signal Processing,Communications and Computing. Washington,DC. c2013:1-5.
[6]	袁小坊，陈楠楠，王东，等. 城域网应用层流量预测模型[J]. 计算机研究与发展， 2009,46(3): 434-442. YUAN X F , CHEN N N , WANG D , et al. Traffic prediction models of traffics at application layer in metro area network[J]. Journal of Computer Research and Development, 2009,46(3): 434-442.
[7]	REN X Y , YANG Y , ZHANG J F , et al. Parameter estimation and application of time-varying FARIMA model[J]. International Journal of Advancements in Computing Technology, 2011,3(3): 89-94.
[8]	陈晓天，刘静娴 . 改进的基于小波变换和 FARIMA 模型的网络流量预测算法[J]. 通信学报， 2011,32(4): 153-157. CHEN X T , LIU J X . Network traffic prediction based on wavelet transformation and FARIMA[J]. Journal on Communications,2011, 2011,32(4): 153-157.
[9]	孟庆芳，陈月辉，冯志全 . 基于局域相关向量机回归模型的小尺度网络流量的非线性预测[J]. 物理学报， 2013,62(15): 150509. MENG Q F , CHEN Y H , FENG Z Q , et al. Nonlinear predict n of small scale network traffic based on local relevance vector machine regression model[J]. Acta Phys Sin, 2013,62(15): 150509.
[10]	LIU X W , LI H , CHEN L . An improved forecasting algorithm for wireless network traffic[J]. IEICE Electronics Express, 2011,8(12): 916-922.
[11]	唐舟进，彭涛，王文博 . 一种基于相关分析的局域最小二乘支持向量机小尺度网络流量预测算法[J]. 物理学报， 2014,63(13): 130504. TANG Z J , PENG T , WANG W B . A local least square support vector machine prediction algorithm of small scale network traffic based on correlation analysis[J]. Acta Phys Sin, 2014,63(13): 130504.
[12]	田中大，高宪文，石彤 . 用于混沌时间序列预测的组合核函数最小二乘支持向量机[J]. 物理学报， 2014,63(16): 160508. TIAN Z D , GAO X W , SHI T . Combination kernel function squares support vector machine for chaotic time series prediction[J]. Acta Phys Sin, 2014,63(16): 160508.
[13]	Li X B . RBF neural network optimized by particle swarm optimization for forecasting urban traffic flow[C]// 3rd Internatio l Symposium on Intelligent Information Technology Application. c2009:124-127.
[14]	QIAN F . A extreme learning machines approach for accurate estima-tion of large-scale IP network traffic matrix[J]. Journal of Computa-tional Information Systems, 2012,8(2): 755-762.
[15]	WANG J S , WANG J K , ZHANG M Z . Prediction of internet traffic based on Elman neural network[C]// Chinese Control and Decision Conference. c2009:1248-1252.
[16]	赵艳平，姜子运 . 基于EKF的神经网络在网络流量预测中的应用[J]. 兰州交通大学学报， 2009,28(6): 23-25. ZHAO Y P , JIANG Z Y . Network traffic prediction based on EKF neural network[J]. Journal of Lanzhou Jiaotong University, 2009,28(6): 23-25.
[17]	CAO J H , LIU Y , DAI Y . Network traffic prediction based on error advanced DGM(1,1) model[C]// International Conference on Wireless Communication Networking and Mobile Computing. c2007:6353-6356.
[18]	PENG Y , LEI M , LI J B , et al. A novel hybridization of ho state networks and multiplicative seasonal ARIMA model for mobile com-munication traffic series forecasting[J]. Neural Compu ing and Appli-cations, 2012:1-8.
[19]	WANG P , ZHANG S Y , CHEN X J . VoIP traffic capacity est mation of hybrid network based on finite user timely refuse mode 5th In-ternational Conference on Wireless Communications,Networking and Mobile Computing. c2009:1-4.
[20]	CHANG B R , TSAI H F . Improving network traffic analysis by fore-seeing data-packet-flow with hybrid fuzzy-based model prediction[J]. Expert Systems with Applications, 2009,36(2): 6960-6965.
[21]	温祥西，孟相如，马志强，等. 小时间尺度网络流量混沌性分析及趋势预测[J]. 电子学报， 2012,40(8): 1609-1616. WEN X X , MENG X R , MA Z Q , et al. The chaotic analysis and trend prediction on small-time scale network traffic[J]. Acta Electronic Sinica, 2012,40(8): 1609-1616.
[22]	LIU X W , FANG X M , QIN Z H , et al. A short-term forecasting algorithm for network traffic based on chaos theory and SVM[J]. Journal of Network and Systems Management, 2011,19(4): 427-447.
[23]	JAEGER H , HAAS H . Harnessing nonlinearity:predicting chaotic systems and saving energy in wireless communication[J]. Science, 2004,304(5667): 78-80.
[24]	XUE Y , JIA L S , TENG W Z , et al. Long-range correlations in vehicu-lar traffic flow studied in the framework of Kerner's three-phase theory based on rescaled range analysis[J]. Communications in Nonlinear Science and Numerical Simulation, 2015,22(1-3):285-296.
[25]	MANDELBORT B B . How long is the coast of Britain statistical self-similarity and fractional dimension[J]. Science, 1967,1556(3775): 636-678.
[26]	GOTTALD G A , MELBOURNE I . On the implementation of 0-1 test for chaos[J]. SIAM Journal on Applied DynamicalSystems, 2009,8(1): 129-145.
[27]	WOLF A , SWIFT J B , SWINNEY H L , et al. Determining Lyapunov exponents from a time series[J]. Physica D, 1985,16:285-317.
[28]	李新杰，胡铁松，等. 0-1测试方法的径流时间序列混沌特性应用[J]. 水科学进展， 2012,32(6): 875-882. LI X J , HU T S , et al. Application of the 0-1 test algorithm for chaos to runoff time series[J]. Advances in Water Science, 2012,32(6): 875-882.
[29]	KHATIBI R , SIVAKUMAR B , GHORBANI M A , et al. Investigating chaos in river stage and discharge time series[J]. Jour l of Hydrolo-gy, 2012,(414-415):108-117.
[30]	KIM H S , EYKHOLT R , SALAS J D . Nonlinear dynamics dela times,and embedding windows[J]. Physica D, 1999,127(1): 48-60.
[31]	张洪宾，孙小端，贺玉龙 . 短时交通流复杂动力学特性分析及预测[J]. 物理学报， 2014,63(4): 040505. ZHANG H B , SUN X D , HE Y L . Analysis and prediction of complex dynamical characteristics of short-term traffc flow[J]. Acta Phys.Sin, 2014,63(4): 040505.
[32]	GRASSBERGER P , PROCACCIA I . Measuring the strangeness f strange attractors[J]. Physica D, 1983,9:189-208.
[33]	ONGENAE F , VAN L S , VERSTRAETEN D , et al. Time series classi-fication for the prediction of dialysis in critically ill patients using echo state networks[J]. Engineering Applications of Artific l Intelligence, 2013,26(3): 984-996.
[34]	SHENG C Y , ZHAO J , LIU Y , et al. rediction for noisy nonlinear time series by echo state network based on dual estima ion[J]. Neuro-computing, 2012(82): 186-195.
[35]	LI D C , HAN M , WANG J , et al. Chaotic time series prediction based on a novel robust echo state network[J]. IEEE Trans on Neural Networks and Learning Systems, 2012,23(5): 787-797.
[36]	OZTURK M C , XU D M , PRINCIPE J C . Analysis and design of echo state networks[J]. Neural Computation, 2007,19(1): 111-138.
[37]	彭宇，王建民，彭喜元 . 基于回声状态网络的时间序列预测方法研究[J]. 电子学报， 2010,38(z1): 148-154. PENG Y , WANG J M , PENG X Y . Researches on time series prediction with echo state networks[J]. Acta Electronic Sinica, 2010,38(z1): 148-154.
[38]	GEEM Z W , KIM J H , LOGANATHAN G V . A new heuristic optimization algorithm:harmony search[J]. Simulations, 2001,76(2): 60-68.
[39]	欧阳海滨，高立群，邹德旋 , 等. 和声搜索算法探索能力研究及其修正[J]. 控制理论与应用， 2014,31(1): 57-65. OUYANG H B , GAO L Q , ZOU D X , et al. Exploration ability study of harmony search algorithm and its modification[J]. Control Theory and Applications, 2014,31(1): 57-65.
[40]	MAHDAVI M , FESANGHARY M , DAMANGIR E . An improved harmony search algorithm for solving optimization problems[J]. Applied Mathematics and Computation, 2007,188(2): 1567-1579.
[41]	ALATAS B . Chaotic harmony search algorithms[J]. Applied Mathe-matics and Computation, 2010,216(9): 2687-2699.
[42]	DAS S , MUKHOPADHYAY A , ROY A , et al. Exploratory power of the harmony search algorithm:analysis and improvements for global numerical optimization[J]. IEEE Transactions on Systems,Man,and Cybernetics,Part B:Cybernetics, 2011,41(1): 89-106.
[43]	彭宇，王建民，彭喜元 . 储备池计算概述[J]. 电子学报， 2011,39(10): 2387-2396. PENG Y , WANG J M , PENG X Y . Survey on reservoir computing[J]. Acta Electronic Sinica, 2011,39(10): 2387-2396.

数据集	Hurst指数	分维数
T_10ms	0.662 1	1.510
T_100ms	0.870 0	1.149
T_1s	0.607 8	1.645
T_10s	0.633 2	1.579
T_1min	0.610 2	1.639
T_10min	0.700 4	1.428
T_30min	0.690 3	1.449
T_1hour	0.638 2	1.567
联通实际数据	0.807 2	1.239

网络流量数据集	渐进增长率
T_10ms	0.998 3
T_100ms	0.998 4
T_1s	0.997 7
T_10s	0.997 0
T_1min	0.998 1
T_10min	0.997 6
T_30min	0.995 8
T_1hour	0.998 8
联通实际数据	0.986 2

数据集	延迟时间t	嵌入维数m
T_10ms	2	10
T_100ms	4	4
T_1s	1	21
T_10s	3	6
T_1min	1	7
T_10min	4	3
T_30min	2	5
T_1hour	3	4
联通实际数据	4	9

网络流量数据集	最大Lyapunov指数
T_10ms	0.106 5
T_100ms	0.802 2
T_1s	0.013 6
T_10s	0.405 1
T_1min	0.162 6
T_10min	0.603 6
T_30min	0.723 0
T_1hour	0.897 7
联通实际数据	0.213 2

算法	平均适应度	最佳适应度	标准差	成功率
HS	15.152 5	10.542 1	2.565 2	20%
EHS	6.562 6	4.120 2	1.230 1	50%
IHS	0.123 6	0.095 2	0.046 3	100%