Research On Techniques Of DDoS Attacks And Defense

The network is more and more widespread and important in the human society life. Work and life have been closely connected by Internet. However, with the development and popularity of Internet, network security has become a hot issue. And different kinds of means of attacks emerged endlessly, among which DoS (Denial of Service) attacks become one of the common network attack techniques by the characteristics, such as extensive area, strong concealment, simpleness and efficiency, hard to defense and great destroy, etc. Especially, DDoS (Distributed Denial of Service) attacks are greatly threatening Internet, since their greater destroy. The losing was immeasurable while under such attack. So it is a very important target in the network security field to establish more effective defense mechanism against DDoS attack.Firstly, the principle and means of Dos and DDoS attacks are analyzed, and the some kinds of DDoS attacks are discussed. The current situation of the research of the technology of detection, defense of DDoS attacks is studied. Attack tree is adopted to model the Distributed Denial of Service attack. Then, Object-Z language, a formal depict attack language, is used to depict the sub-term of the attack tree model in detail to guide the analysis, detection and defense of the DDoS attack. In succession, form the view of ISP (Internet Service Provider) domain, adopting the technology of mobile agent and the idea of integrated method, a DDoS defense model based on mobile agent is exploringly put forward. The elements of the mobile agent in defense model are designed in detail. The technology of mobile agent enable the defense model itself has the ability to defense DDoS attacks. And the idea of integrating flow monitoring, traffic filtering and traceback schemes breaks through the limitation of the single defense method. Subsequently, the definite ability against DDoS attacks of the model is proved by theories analyzing. Moreover, mobile agent programs are developed to prove the feasibility of using mobile agent in the model. Finally, the future research work is presented.