發(fā)布時間：2018-03-19 23:32

  本文選題：流量監(jiān)測　切入點：流量預測　出處：《西安電子科技大學》2014年碩士論文　論文類型：學位論文

【摘要】：隨著互聯(lián)網(wǎng)的快速發(fā)展,網(wǎng)絡(luò)新型應用逐漸豐富,網(wǎng)絡(luò)規(guī)模不斷增大。網(wǎng)絡(luò)流量監(jiān)測與預測技術(shù)作為增強網(wǎng)絡(luò)控制性的有效技術(shù),不僅能夠獲取網(wǎng)絡(luò)流量數(shù)據(jù),而且對網(wǎng)絡(luò)的監(jiān)督管理、服務(wù)質(zhì)量、安全管理、故障檢測、容量規(guī)劃等都有很重要的影響。SNMP作為一種簡單網(wǎng)絡(luò)管理協(xié)議,已經(jīng)廣泛應用于各種網(wǎng)絡(luò)管理系統(tǒng)。但其輪詢MIB節(jié)點會產(chǎn)生大量的管理報文,這對網(wǎng)絡(luò)帶寬和處理能力提出更高的要求,且只支持集中式管理。而RMON協(xié)議能很容易地解決這些問題,RMON協(xié)議體系包括RMON1和RMON2標準,與RMON1相比,RMON2能夠周期性監(jiān)控更豐富的流量信息。本文實現(xiàn)了基于RMON2協(xié)議的網(wǎng)絡(luò)流量監(jiān)測系統(tǒng),并在此基礎(chǔ)上對兩種神經(jīng)網(wǎng)絡(luò)的流量預測算法進行性能仿真。主要工作如下：1.對于流量監(jiān)測,介紹了RMON協(xié)議的標準和工作方式,重點論述了RMON1和RMON2之間的區(qū)別。對于流量預測,基于BP神經(jīng)網(wǎng)絡(luò)模型和小波神經(jīng)網(wǎng)絡(luò)模型,推導了兩種網(wǎng)絡(luò)算法的步驟,經(jīng)理論分析可得出小波神經(jīng)網(wǎng)絡(luò)具有更好的流量預測性能。2.分析了RMON2系統(tǒng)的需求并提出了該系統(tǒng)的總體設(shè)計。首先將系統(tǒng)總體劃分成子模塊,主要包括系統(tǒng)的零層、一層和二層分解模塊。然后,由一層模塊和二層模塊的運行設(shè)計完成整個系統(tǒng)方案的設(shè)計。最后,使用C語言實現(xiàn)了RMON2系統(tǒng)。3.構(gòu)建RMON2系統(tǒng)后,分別從功能測試、性能測試、規(guī)格測試、組合壓力測試和兼容性測試等方面對RMON2系統(tǒng)進行測試,測試結(jié)果表明該系統(tǒng)功能穩(wěn)定。4.利用一種TCL腳本語言完成了RMON2自動化工具,給出了RMON2自動化工具監(jiān)測網(wǎng)絡(luò)流量的方法,并分析了自動化測試的優(yōu)缺點。5.使用RMON2自動化工具對網(wǎng)絡(luò)運營商的設(shè)備進行流量監(jiān)測,周期性采樣接口上行和下行流量值,并以此數(shù)據(jù)作為后續(xù)流量的預測樣本,分別對BP神經(jīng)網(wǎng)絡(luò)和小波神經(jīng)網(wǎng)絡(luò)的流量預測模型進行性能仿真實驗。仿真實驗表明,兩種預測模型都能很好地對網(wǎng)絡(luò)流量進行預測,且在同一仿真條件下,小波神經(jīng)網(wǎng)絡(luò)的流量預測算法可獲得更小的預測誤差。
[Abstract]:With the rapid development of the Internet, the new network applications are becoming more and more abundant and the network scale is increasing. As an effective technology to enhance the network control, the network traffic monitoring and forecasting technology can not only obtain the network traffic data. Moreover, SNMP has a very important influence on network supervision and management, quality of service, security management, fault detection, capacity planning and so on. SNMP is a simple network management protocol. It has been widely used in various network management systems, but its polling of MIB nodes will produce a large number of management packets, which puts forward higher requirements for network bandwidth and processing capability. The RMON protocol can easily solve these problems, including RMON1 and RMON2 standards. Compared with RMON1, RMON2 can monitor more abundant traffic information periodically. A network traffic monitoring system based on RMON2 protocol is implemented in this paper. The main work is as follows: 1. For traffic monitoring, the standard and working mode of RMON protocol are introduced, and the difference between RMON1 and RMON2 is emphasized. Based on BP neural network model and wavelet neural network model, the steps of two network algorithms are deduced. Through theoretical analysis, it can be concluded that wavelet neural network has better flow prediction performance. 2. The demand of RMON2 system is analyzed and the overall design of the system is put forward. Firstly, the whole system is divided into sub-modules, including the zero layer of the system. The first layer and the second layer decompose the module. Then, the operation design of the first layer module and the second layer module completes the design of the whole system scheme. Finally, the RMON2 system. 3. After constructing the RMON2 system, the function test and the performance test are carried out, respectively. The RMON2 system is tested in the aspects of specification test, combined stress test and compatibility test. The test results show that the system functions stably .4.Using a TCL script language to complete the RMON2 automation tool, This paper presents a method of monitoring network traffic with RMON2 automation tools, and analyzes the advantages and disadvantages of automatic testing. Using RMON2 automation tools to monitor the traffic of network operators' equipment, periodically sampling the upstream and downlink traffic value of the interface. The simulation experiments on the traffic prediction models of BP neural network and wavelet neural network show that the two models can predict the network traffic very well. Under the same simulation condition, the traffic prediction algorithm of wavelet neural network can obtain smaller prediction error.
【學位授予單位】：西安電子科技大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.06

【參考文獻】

相關(guān)碩士學位論文 前1條

1 楊麗紅;軟件測試與可靠性研究[D];四川大學;2006年

，

本文編號：1636554