Dissertation > Excellent graduate degree dissertation topics show

Research on Neural Network Ensemble and Its Application to the Identification of P2P Traffic

Author: DingLing
Tutor: LiuDanPing
School: Chongqing University
Course: Communication and Information System
Keywords: P2P technology Network control P2P traffic identification Neural Network Ensemble
CLC: TP393.02
Type: Master's thesis
Year: 2011
Downloads: 60
Quote: 1
Read: Download Dissertation

Abstract

P2P technology is more and more widely, but the most important problem of the application is to consume a large amount of network bandwidth, and bring copyright, security and spam issues. P2P traffic monitoring and management in order to improve the performance of network services. P2P traffic identification is a key technology in the monitoring and management system. Research and efficient method of P2P traffic identification is important. From home and abroad reported that P2P traffic identification method based on neural network can achieve better results; generalization capability of the system but the method is not high, the subject of the use of neural network integration to solve this problem. This article innovation mainly lies in the following three aspects: First, the proposed adaptive Genetic Algorithm Neural Network Ensemble method, this method can effectively improve the generalization ability of learning systems; integrated application of neural network in RBF and FUZZY ARTMAP neural network integration, and the integration of these two neural network used in the identification of P2P traffic; build simulation models, the integration of these two neural network P2P traffic identification method simulation, and compare these two neural network integrated P2P traffic identification method. We first compared the BP, RBF and FUZZY ARTMAP three neural network simulation results show that the three types of neural networks have a high recognition rate of P2P traffic, but, RBF neural network training time best, BP neural network worst . Integrated RBF and FUZZY ARTMAP neural network for P2P traffic identification, the results show that the two neural network integration sacrifice some training time and recognition time based on the rate of identification of P2P traffic has greatly improved. We finally optimize the neural network ensemble, the results showed that, based on the simple average of the RBF neural network ensemble average recognition rate of 99.15%, the average recognition rate than before optimization to improve by 1.5%, and the average training time 2.5872s average recognition time is 0.9174s; FUZZY ARTMAP neural network ensemble average recognition rate of 99.48%, compared to optimize the average recognition rate increased by 0.6%, the average training time 6.2824s, 1.4193s average recognition time. From the results on RBF neural network integrated training time and recognition time is less, to meet the real-time requirements, identify better recognition rate below FUZZY ARTMAP neural network ensemble. If you use the software, the best RBF neural network integrated solutions; If you are using a hardware solution, FUZZY ARTMAP neural network integration is the best idea.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Computer network > General issues > Computer network architecture and design
© 2012 www.DissertationTopic.Net Mobile