Research Of P2P Detection Based On Genetic Neural Network

P2P traffic detection is motivated operators need to improve the quality andefficiency of its network operations. Because a large number of P2P file downloadingand video applications exist, allows P2P traffic may occupy more than60%of availablebandwidth, has seriously affected the conduct of business operators and their networkquality of service improved. To do this, operators need to detect P2P traffic, in order toachieve the control of P2P traffic. Of course, not to sealing of P2P traffic, but only takeup too much bandwidth, or charge to remind the user, or by planning into a limit thebandwidth of P2P traffic in order to achieve control of P2P traffic to monitor them.Practical cases, the P2P detection of real-time the giant traffic under the problemhas not been resolved, this paper proposes using secondary detection methods,innovative use of half-open connections as a one-detection feature, screening traffic,then screeningtraffic using deep packet inspection, in order to solve a giant real-timetraffic. And use a neural network based on genetic algorithm to find the threshold ofhalf-open connection features.The thesis designs and implements a use for detection of P2P traffic detectiontechnology based on the half-open connection behavior characteristics of P2P trafficdetection system.Based on the half-open connection behavioral characteristics andapplications through the use of genetic algorithm-based neural network technology, totry to resolve the lack of performance of the large flow environment under real-timedetection of P2P traffic, compared with the traditional detection system, the completesystem in network trafficcase to ensure recognition accuracy rate of more than80%andlower overhead.