Dissertation > Excellent graduate degree dissertation topics show

Distributed Real-Time Detection System Research and Implementation of P2P

Author: ZhouZhiZuo
Tutor: LiZhiTang
School: Huazhong University of Science and Technology
Course: Information Security
Keywords: Distributed Real time Peer network Behavior recognition Passive packet capture Traffic Filtering
CLC: TP393.02
Type: Master's thesis
Year: 2011
Downloads: 15
Quote: 0
Read: Download Dissertation

Abstract

In recent years, increasing broadband users and \Studies have shown that in addition to WEB class applications, in which the main bandwidth from P2P (peer to peer networks), including P2P downloading and P2P streaming media categories. In order to ensure the fairness of network resource utilization, identify and control P2P traffic has become a focus of research in the field traffic identification. However, with the popularity of 10-Gigabit Ethernet, a common network monitoring system and intrusion detection system, has become increasingly difficult to meet the needs of high-volume environments. Experience has shown that the low-end server performance has been unable to adapt to the passive Gigabit Ethernet packet capture environment, because CPU cycles and bus bandwidth has reached a bottleneck. Fast network based detection system under estimate the efficiency of algorithms than hardware factors more likely to become system bottleneck. Despite today's P2P detection and classification algorithms endless, but the vast majority of the efficiency of P2P behavior recognition algorithms are just stuck in the \Through the pre-Gigabit Ethernet environment on passive packet capture, and found a better solution, and propose a way to capture 10Gbps line speed or greater flow of distributed behavior-based detection of P2P identification system, and without using special hardware. Subsequently elaborated large flow of data processing environment critical - flow filter, and compare the existing traffic filtering policy, proposes a new dynamic based on the identified node changing traffic flow filtration methods to effectively solve the duplicate identification known nodes problem. In the P2P behavior recognition algorithms to improve recognition accuracy, while the industry was first proposed to improve the efficiency of P2P behavior recognition program, through the algorithm memory footprint and improved statistical methods, behavior recognition algorithm to achieve the P2P real-time processing. Based on the above approach, design and implementation of distributed real-time P2P behavior recognition system, through a set of low-speed interfaces (such as 1Gbps interfaces), to capture and analyze high-speed interfaces (such as 10Gbps interfaces) flow data. System is divided into data interface layer, data logic layer and data aggregation layer three parts. In the system implementation part, explain, including no lock ring buffer memory shared between processes, optimization of busy waiting, nodes, flow hash algorithm, including a variety of optimization techniques in the implementation of distributed systems while improving overall performance. Currently a prototype system has been put into trial operation, the average recognition rate of 95%.

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