Design And Implement Of Traffic Identificationand Control System On VoIP And P2P

In recent years the technology of broadband Internet rapidly developed in China,the topology of the network became more complex,the network carries more and more rich business,so the network traffic grows fast.Especially VoIP and P2P applications emerge constantly,the number of users rapidly grows,thus the two kinds of traffic occupies most of the network bandwidth,at the same time some illegal applications take into the Internet and threat the network security.So it becomes a hot research spot about detecting and controlling the network traffic reliably and effectively,analyzing in real time and accurately on all kinds of business carried by the network,excavating capacity of network resources all aspects,effectively reduce the cost of network information transmission,and providing foundation basis for the network development planning and the business adjustment optimization.Based on reading and study the predecessors' research about VoIP protocol,P2P protocol,traffic identification technology,control technology,matching algorithm,machine learning,this paper designs and implements a network traffic monitoring system for VoIP and peer-to-peer(P2P)traffic.This system innovatively combines the AC match algorithm with the naive bayesian classification algorithm to identify the network traffic together.This paper take some VoIP and P2P typicle traffic as an example to design and use the system,shows the effectiveness and accuracy of the system.This paper's work mainly includes the following aspects:1.This paper reads a lot of related references at home and abroad,analyzes the predecessors' research achievements about traffic identification technology,traffic control technology,and compares the advantages and disadvantages of them.Then this paper in-depth studies their research on AC matching algorithm,naive bayesian algorithm and machine learning.2.This paper analyzes in detail the AC matching algorithm,further studies the matching principle and implementation steps.Moreover this paper optimizes the AC engine,makes a pretreatment after loading the rules library and flows into the system,which is named as separately processing graph.So the separately processing graph effectively reduces the scale of the AC automata,accelerating matching.For the traffic without apparent load characteristics in application layer this system uses the machine learning method of naive bayesian classification algorithm,which can identify the traffic.By designing the system,this system can automatically separate the training step and application step for machine learning according to the "training" attribute.This system combines the AC engine and the FB engine to identify most of the network traffic.3.This paper analyzes the VoIP protocol and P2P protocol,and takes QQ,Skype and thunder applications as an example to study their traffic,and use the system to identify,control the flows and test some related parameters.4.The paper designs and implements separately all function modules.The system contains rule library module,database module,traffic identification module,flow control module,log module,traffic statistics module.Each module describes about the design process,frame structure,and content of realizing.After completing the system designing and realizing this paper proceeds the functional test and performance test and shows the test results in the form of tables and charts,and makes a simple analysis for the test results.In the end,this paper summarizes all the work,the problems existing in the system and possible improvements.