Dissertation > Excellent graduate degree dissertation topics show

Research on the Stream Data Mining in Network Traffic Analysis

Author: LaiJun
Tutor: LiShuangQing
School: Chongqing University
Course: Computer Software and Theory
Keywords: Network Traffic Analysis Stream data mining Association rules Multi-level model
CLC: TP393.06
Type: Master's thesis
Year: 2011
Downloads: 101
Quote: 0
Read: Download Dissertation

Abstract

With the rapid development of Internet, Web-based applications become increasingly diverse types of enterprises have adopted the network to expand their business scope. New emerging network applications made more demands on network bandwidth, network performance concern. Network resources and services in order to alleviate the contradiction between quality, through in-depth analysis of network traffic, and thus rational allocation of limited network resources has become a hot issue of current research. Traditional network traffic analysis mainly uses the methods of mathematical statistics, can not adapt to a variety of complex applications in today's network situations. To this end, this paper combines the characteristics of network flow data, stream data mining theory into network traffic analysis which. Complex network flow data is often a variety of network applications in the course of a process of integration, a variety of applications from which to explore the links to find the frequent appearance of a variety of network traffic will be able to intuitively grasp the operation of the network, so as to network management and provide the basis for resource allocation. This paper completed the following tasks: (1) the formation of network traffic patterns and characteristics, to discuss the difference between a typical network flow and classification methods. Analysis of the current network traffic analysis methods the main characteristics, focusing on the flow of data mining theory in network traffic analysis in the feasibility of the application. (2) analysis of a variety of frequent itemset mining algorithms and multi-level association rules mining technology for the traditional algorithm in time and space through the large complex is not suitable for stream data mining insufficient integration attenuation window sliding window with the advantages of both, made a Dictionary-based order prefix tree species LOP-Tree algorithm for mining frequent itemsets STFWFI, effectively reducing the frequent itemset mining of time and space complexity. Proposed based on the statistical distribution of the node weight calculation method SDNW, improve the network flow nodes valuation accuracy. Based on these frequent itemsets mining results, the use of multi-level association rule mining method for network traffic associated with the application of rules and network traffic load correlation analysis. (3) on the Windows platform using network development kit Winpcap for network flow data acquisition and processing. Weka data mining platform based on open-source model to achieve the core mining algorithms, and complete the prototype system. By mining frequent network flow, generate consistent analysis of multi-level association rules target to calculate the load between various types of network traffic correlation coefficient, which verifies the validity of theoretical methods papers.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Computer network > General issues > Computer networks, test , run
© 2012 www.DissertationTopic.Net Mobile