关键词: 网络流量势能, 分布式拒绝服务攻击, 时间序列, 流量检测, 支持向量机, DARPA 数据集

Abstract: This paper introduces the current situation and development of Distributed Denial of Service(DDoS) attack, and proposes a flow potential energy analysis model based on time sequence,constructs sequence of network flow potential energy. It uses Auto Regression(AR) model to fit multi-dimensional parameter vector and describes the stability of network flow in unit time,and employs Support Vector Machine(SVM) based method to classify and train the target network flow character parameter vector,gains the best-matched network data flow potential energy set and final achieves accurate description of different DDoS attacks. It uses DARPA dataset,IXIA 400 network test machine and other softwarehardware fundamentals to construct a real and analysis of the value network,validates the network flow potential energy analysis model based on the constructed network. Analysis and contrasts of the key indicators include DDoS detection accuracy, recognition rate, etc. Experimental results show that the method has higher detection precision and comprehensive better detection quality to DDoS.

Key words: network flow potential energy, Distributed Denial of Service ( DDoS ) attack, time sequence, flow detection, Support Vector Machine(SVM), DARPA dataset