摘要
随着网络的普及以及新的应用如(VoD,VoIP)的出现,宽带网络服务需求(如多媒体、视频业务等)的激增,网络的突发业务流量急剧增加,基于传统模型的流量特征不再适合当前网络流量的分析。由于自相似模型能够更加真实地描述网络流量的特性,成为科研人员研究的热点问题。
本文首先介绍了国内外对自相似性流量的研究现状,包括自相似性的定义、性质、生成方法、描述流量自相似程度的Hurst参数估计方法及多个具有自相似性的网络模型;重点分析了自相似流量的成因、TCP与流量自相似性的相互关系、及其对网络性能的影响。本文的试验分为三部分:
1.校园网自相似性的研究。利用我校网络中心的网络流量监测系统对局域网流量进行采集,通过处理得到不同时段的分组到达过程,采用经验方差时间分析方法,对Hurst参数进行估计。证实校园网流量分组到达过程的统计特征是具有自相似性;实际流量从繁忙状态到空闲状态H值呈递减的趋势;且在繁忙时段,流量采集时间尺度越大,H值也越大。
2.流量自相似性的仿真试验。通过采用FFT-FGN方法在网络模拟器NS2下对生成网络模拟流量进行分析,采用方差时间分析法进行Hurst参数估计方法,并与测量得到的实际流量进行比对。证明仿真流量是具有自相似性的,可以有效地反映真实网络流量特征。在H≦0.7时,方差-时间图方法能够准确地估计H值;但对于H值较高时(H≧0.75)通过方差-时间图分析法得到的H估计值会低于期望值。
3.自相似性对网络性能影响试验。基于FFT-FGN方法生成自相似流量,分析了在不同协议、不同网络负载因素下自相似性对性能中分组丢弃率的影响,并将结果与基于传统泊松模型下的流量作以比对;同时在仿真试验中增加丢包模块,察了分组丢弃率对网络流量自相似程度的影响。随着UDP和TCP负载增加,自相似程度增加时分组丢弃率也呈增长趋势,但是泊松模型下分组丢弃率的幅度相比FGN流量下的分组丢弃率要小,证实传统模型确实存在不能准确描述网络流量特征的缺陷,自相似性对网络性能具有更大的影响。
The popularization and enhancement of new network applications and the demand of broad band service make network traffic increasing rapidly such as multimedia and VoIP etc, so that traditional traffic model character will no longer be adapted to current network traffic. The effective method for studying self-similar traffic is to build models so that models could describe network character more authentic, and be applied to simulation research. The self-similar traffic becomes a hotspot for researchers.In the paper, we introduce research background、 mathematic definition、 character of self-similar traffic, including several common methods of creating self-similar traffic and estimating Hurst parameters, Then we analyze the cause of self-similar traffic appearance, the relation between TCP and self-similar, influence to network performance of self-similar traffic in detail. The paper includes three experimentations:1. The research on the traffic of campus network. We have measured local network traffic with inspect system of network center, we deal with packet arrival process and adopt V/T method to estimate Hurst parameter. We validate that the packet arrival process of local traffic is self-similar;The value of Hurst parameter is decreasing with the traffic state from busyness to idle;In the busy state, the longer time scale of capturing traffic, the higher the value of Hurst parameter.2. Simulation experimentation on self-similar traffic. In order to model self-similar traffic source and make simulation in NS, we use the method of FFT-FGN to synthesize sample path. Comparing with the measured actual traffic based on the same estimating method of Hurst. We validate the effectiveness of the way for creating self-similar traffic. It can depict the traffic character availably. For a true self-similar process, the variance-time plot for a given value of H will coincide with the corresponding line for H = 0.7, but for H ≧ 0.75, the variance-time plot was sometimes lower (i.e. steeper-sloped) than expected.3. Study on influences to network performance of self-similar traffic. Based on the simulation experimentation, we analyze influences self-similar traffic makes on the packet drop rate at different protocols and network loads;then compare the test with
the one on Poission traffic. We also seeing about how the packet drop rate make influence on self-similar character with adding the packet loss module in the simulation. The results are as follows: With the increasing of TCP and UDP loads and the degree of self-similar character, the packet drop rate grows on, further more, this extend of drop rate is higher than the one on Poission model. It shows traditional models have deficiency in describing network character well and truly;and also approve self-similar character has more influences on network performance.
引文
[1] Leland W, Taqqu M, Willinger W, et al. On the Self-Similar Nature of Ethemet Traffic (Extended Version). IEEE/ACM Transactions on Networking, 1994,2(1): 1-15
[2] Paxson V, Floyd S. Wide Area Traffic: The Failure of Poisson Modeling. IEEE/ACM Transactions on Networking, 1995,3:226-244
[3] Erramilli A, Roughan M, Veitch D, et al. Self-Similar Traffic and Network Dynamics. in: Proc of The IEEE, 2002,90(S):800-819
[4] Willinger W, Taqqu M, Sherman R, et al. Self-Similarity Through High-Variability: Statistical Analysis of Ethernet LAN Traffic at the Source Level. IEEE/ACM Transactions on Networking, 1997,5(1): 71-86
[5] Crovella M, Bestavros A. Self-Similarity in World Wide Web Traffic: Evidence and Possible Causes. IEEE/ACM Transactions on Networking, 1997, 5 (6): 835-846
[6] Likhanov N, Tsybakov B. Analysis of an ATM Buffer with Self-Similar ("Fractal") Input Traffic. in: Proc of IEEE INFOCOM'9S, 199S, 98S-992
[7] Norms I. A Storage Model with Self-Similar Input. Queueing Systems, 1994, 16:387-396
[8] Grossglauser M, Bolot J.C. On the Relevance of Long-range Dependence in Network Traffic. in: Proc of ACM SIGCOMM'96, 1996,15-24
[9] Feldmann A, Gilbert A.C, Willinger W. Data Networks as Cascades: Investigating the Multifractal Nature of Internet WAN Traffic. Computer Communications Review, 1998,28(4): 42-58
[10] Park K, Kim G,Crovella M. On the Relationship Between File Sizes, Transport Protocols, and Self-Similar Network Traffic. in: Proc of IEEE International Conference on Network Protocols, 1996, 171-180
[11] Sikdar B, Vastola K.S. The Effect of TCP on the Self-Similarity of Network Traffic. in: Proc of the 35th Conference on Information Sciences and Systems.Baltimore, 2001,188-195
[12] Veres A, Boda M. The Chaotic Nature of TCP Congestion Control. in: Proc of IEEE INFOCOM'2000, 715-1723
[13] Matthew Roughan, Darry Veitch, and Patrice Abry, On-line Estimation of Parameters of Long-Range Dependence, IEEE GLOBE-COM'98, Sydney Australia, Nov. 1998, p3716-3721
[14] Michael Jiang, Analysis of Wireless Data Network Traffic.2002 http://www.ensc.sfu.calpeoplefaculty/ljiljalcnl/presentations/michael/michael_thesis/index.htm
[15] 许都,李乐民.网络中业务流的自相似性与线性AR模型[J].电子学报,1999,27(4):6-10.
[16] 王振龙.时间序列分析.北京:中国统计出版社,2000.
[17] Tsybakov B and Georganas N D. Self-similar processes in communications networks. IEEE Transactions on Information Theory, Sep. 1998, 44:1713-1725
[18] Beran J. Statistic for Long-Memory Processes. New York: Chapman&Hall, 1994
[19] Paxson V. Fast, approximate synthesis of fractional Gaussian noise for generating self-similar network traffic. Computer Communication Review, 1997, 5(3): 5-18
[20] Jan Beran, Robert Sherman, Tagqu M S, Walter Willinger, Long-Range Dependence in Variable-Bit-Rate Video Traffic. IEEE Transactions On Communications, 1995, 43 (2/3/4)
[21] Taqqu M. S. Teverovsky V. Willinger W. Estimators for Long Range Dependence: an Empirical Study. Fractals, 1995,3(4): 785-798,
[22] Wornell G W, Oppenbein A V. Estimation of fractal signals from noisy measurements using waveless. IEEE Trans on Signal Processing, 1993, 40(3): 611-623
[23] B.B. Mandelbrot and M. S. Taqqu. Robust R/S analysis of long-run serial correlation. In Proceedings of the 42nd Session of the International Statistical Institute, Manila, 1979, 48: 69-99.
[24] B. B. Mandelbrot, J. R, Wallis. Computer Experiments with Fractional Gaussian Noises. Water Resources Research, 1969,5:228-267
[25] J. Geweke and S. Porter-Hudak. The Estimation and Application of Long Memory Time Series Models. Journal of Time Series Analysis, 1983,4:221-238
[26] Murad S. Taqqu and V. Teverovsky. Semi-Parametric Graphical Estimation Techniques for Long-Memory Data. Proc. Athens Conference on Applied Probability and Time Series Analysis, 1994, 2:420-432
[27] P. Abry and D. Veitch. Wavelet analysis of long-range-dependent traffic. IEEE/ACM Transactions on Networking, 1998, 44(1):2-15.
[28] P. Abry and D. Veitch. A Wavelet Based Joint Estimator of the Parameters of Long-Range Dependence. IEEE/ACM Transactions on Information Theory, April, 1999.
[29] P. Abry, L. Delbeke and P. Flandrin. Wavelet-Based Estimator for the Self-Similarity Parameter of a Stable Process. IEEE-ICASSP-99, Phoenix (AZ), 1999.
[30] M. S. Taqqu, W. Willinger and R. Sherman. Proof of a Fundamental Result in Self-Similar Traffic Modeling. Computer Communication Review, 1997, 27: 5-23.
[31] D. R. Cox. Long-Range Dependence: A Review in Statistics: An Appraisal. Proceedings 50th Anniversary Conference, Iowa State Statistical Library, H. A. David and H. T. David, editors, Iowa State University Press, pp. 55-74, 1984.
[32] Feldmann A, Gilbert A.C, Willinger W, et al. The Changing Nature of Network Traffic: Scaling Phenomena. Computer Communication. 1998,28:5-29
[33] 马皓,关于网络流量自相似特性的研究.ppt,北大网络实验室.2002年10月
[34] 赵佳宁,李忠诚,基于模拟的网络流量自相似现象分析,计算机科学 2001,28(11):57-61
[35] Park K, Kim G, Crovella M. On the Effect of Traffic Self-Similarity on Network Performance. In: Proc of SPIE International Conference on Performance and Control of Network Systems, 1997,989-996
[36] 孙海荣,李乐民.具有自相似特性的业务及其对ATM网络性能的影响.电路与系统学报.1997年02期.
[37] M. Paulekar and A. M. Makowski. Tail probabilities for a multiplexer with self-similar traffic. Proc. IEEE INFOCOM, pp. 1452-1459, 1996.
[38] D. Clark, S. Shenker, and L. Zhang. Supporting real-time applications in an integrated services packet network: architecture and mechanism. In Proc, SIGCOMM '92, Aug, 1992, pp. 14-26.
[39] D. Yates, 1. Kurose, D. Towsley, and M. Hluchyj. On per-session end-to-end delay distributions and the call admission problem for real-time applications with QOS requirements. In Proc, SIGCOMM'93, Sept. 1993, pp. 2-12.
[40] Smith K. Effects of Self-Similar Traffic on Queueing Systems. In: Proc of CMG' 94,1994,1236-1247
[41] Fiorini P, Ding Y, Lipsky L. On Modeling and Characterizing Self-Similar Data Traffic. in: Proc of CMG '98, 1998, 874-885
[42] Paxson V. Fast Approximation of Self-Similar Network Traffic. ACM SIGCOMM Computer Commurtication Review, 1997,27(5): 5-18
[43] Erramilli A, Narayan O, Willinger W. Experimental Queueing Analysis with Long-Range Dependent Packet Traffic. IEEE/ACM Transaction on Networking, 1996, 4(2): 209-223
[44] Mandelbrot B. Ness J. Fractional Brownian motion. Fractional noise and Applications. SIAM Review, 1968, 10(4):422-437
[45] Lau W.C, Erramilli A., Wang J. L., et al. Self-Similar Traffic Generation: The Random Midpoint Displacement Algorithm and Its Properties. in: Proc of the IEEE International Conference on Communications, Seattle, 1995,1:466-472
[46] P Flandrin. Wavelet Analysis and Synthesis of Fractional Brownian Motion. IEEE Transactions on Information Theory, 1992, 38 (2):910 -917
[47] Ganger C, Joyenn R.An introduction to long-memory time series models and fractional differencing. Time Series Anal, 1980.1(1):5-29
[48] Hosking J.Fractional differencing. Biometrika, 1981, 68(1): 165-176.
[49] Huang C C, Devestikiotis M, Lambadaris I, et al. Modeling and simulation of self-similar variable bit rate-compressed video aunifiedapproach. ACM Computer Communication Review, 1995, pp.114-125.
[50] Huang C C, Devestikiotis M, Lambadaris I, et al. Fast simulation for self-similar in ATM networks. ICC'95 [C]. New York, USA, 1995.pp. 438-444.
[51] Adas A. Traffic Models in Broadband Networks. IEEE Communications Magazine, 1997, 35(7): 82-89.
[52] Garrett M W, Willinger W. Analysis, Modeling and Generation of Self-similar VBR Video Traffic. ACM SIGCOMM, 1994, pp.269-280.
[53] 张连芳,薛飞,王宙等.自相似网络业务的一个FARIMA模型.计算机研究与发展,2000,37(9):1138-1144.
[54] Mondragon R J. A model of packet traffic using a random wall model. Int J of Bif and Chaos. 1999, 9(7): 1382-1392
[55] 蔡弘,陈贤民,李衍达.自相似业务模型--通信网络突发业务建模的新方法.通信学报,1997,18(11):51-59.
[56] A Erramilli, R P Singh,P Pruthi. An Application of Deterministic Chaotic Maps to Model Packet Traffic. Queue Systems and their Application, 1995, (5): 521-529
[57] Nagarajan K, Zhou G T. Self-similar traffic sources: modeling and real-time resource allocation. IEEE on Statistical Signal Processing, 2001, 11(8):74-77
[58] Wornell G W. A Karhun-Loeve-like expansion for 1/f process. IEEE Transactions on Information Theory, 1990, 36(4): 859-861
[59] Wornell G W. Wavelet-based representations for the 1/f family of fractal process. Proceedings of the IEEE, 1993, 81(10): 1428-1450
[60] Roughan R, Veitch D, Abry P. Real-time estimation of the parameters of long-range dependence. IEEE/ACM Transactions on Networking. 2000, 8(4): 467-478
[61] Haslett J. Space-time modeling with long-memory dependence: assessing Ireland's wind power resource. Applied Statistics. 1989. 38(1):1-50
[62] 孙清华、孙昊.随即过程-内容、方法与技巧.华中科技大学出版社,2003
[63] Beran J. Estimation, Testing and Prediction for Self-Similar and Related Processes. Ph.D. dissertation, ETH, Zurich, 1986.
[64] Program for synthesizing self-similar processes, FFT-FGN-C 1.2[CP/OL]. http://ita.ee.1bl.gov/html/contrib/fft_fgn_c.html.
[65] Peha J. Protocols can make traffic appear self-similar. In Proc of the 1997 IEEE/ACM /SCS Comm. Networks and Distributed System. Modeling and Simulation Conf, 1997, 1: 47-52
[66] Vexes A, Hungary B and Boda M.The Chaotic Nature of TCP Congestion Control [J]. iProceedings of IEEE INFOCOM' 2000. 3
[67] Willinger W, Paxson V and Taqqu M S. Self-Similarity and Heavy-Tails: Structural Modeling of Network Tramc. In A Practical Guide to Heavy Tails: Statistical Techniques and Applications [J]. Vexes A, Hungary B and Taqqu M S, editom ISBN 0-8176-395 l-9.Birkh auser. Boston, 1998