Design And Implementation Of P2P Traffic Identification System Based On Ant Colony Optimization And Support Vector Machine

In recent years,with the rapid development of the network,the peer-to-peer(P2P)technique has been widely used and taking increasing percentages of the network flow since its emergence in 1990s.The development of P2P has benefited a lot to the customers' convenient use,however,it also produces a lot of negative impact to the network management and quality.Hence,it is critical to effectively control and manage the P2P traffic and the prerequisite of which is how to categorize P2P trafficefficiently,accurately and quickly.However,the current available traditional traffic categorizing technologies,such as the onesbased on static port,application signatures and deep packet inspection,couldn't work well due to its limitations to the protocol encryptionand port hopping.Thus,developing a new identification method based on the machine learning has attracted a lot attention in this area.Basically,the P2P traffic identification is a classification problem for 2 classes in nature.Also,the statistical decision classification method of machine leaning and data mining enable us to treat the process of P2P traffic identification as the process of determining the statistical features of P2P traffic and choosing the appropriate classification method.The selection and classification of the traffic flow featureshave a big influence on the accuracy of the categorizing results.Thus we could convert the problem into seeking the optimal method for P2P traffic identification to some extent.The objective of this paper is to design and develop a P2P traffic identification system based on ant colony algorithm and support vector machine(SVM).The paper mainly includes the following work:1.Investigatethe current available P2P traffic identification technologies and make a comparison of the machine learningin the P2P traffic identification application and the features of traditional P2P traffic identification techniques.Also,a knowledge gap was identified and the research objectives were drawn based on the currentdomestic and international research and future trend of P2P traffic identification method development.2.Make a detail introduction of the relevant theoreticalbackground of thekey technologies in this paper including SVM,SVM categorizing algorithm and feature selection methods etc.Also,a novel algorithm based on the ant colony was proposed for the traffic features selection,which could not only improve the accuracy of the identification,but also avoid the extra work load due to the need for collecting more traffic features.This is very important for the next step work.3.Analyzethe demand,design outlines as well as how to realize the design details of the P2P identification system developed in this paper.Every sub-module of the system was explained in detail and finally a P2P system based on ant colony algorithm was realized.4.A testing platform was constructed and experiment data were collected to test the performance and functionality of the system.Also,the experimental results were carefully analyzed and discussed.The results showed that the P2P traffic identification system designed in our work could overcome the limitations of traditional P2P traffic identification technologies.A higher accuracy was realized too,which gives the method potential practical application in the future.