Application Of Naive Bayes Classification In DPDK-based DDoS Defense System

Distributed Denial of Service(DDoS)is developed by DoS attacks,mainly for the effectiveness of information security(secure,integrity,validity).Usually,multiple controlled zombie hosts simultaneously attack the victim host,occupy-ing the computing resources and effective bandwidth of the victim host,thereby achieving the purpose of preventing the victim host from providing normal ser-vices.The existing defense methods for DDoS attacks are many,but Due to the shortcomings of the TCP/IP(Transmission Control Protocol/Internet Protocol)protocol,DDoS attacks cannot be completely eliminated.At the same time,the large and full function of traditional protective equipment has caused the cost of hardware to cost a lot of costs.With the development of hardware and software and the maturity of machine learning technology,the defense of machine learn-ing technology in DDoS attacks.It plays an increasingly important role.There-fore,this topic will focus on the defense system that uses machine learning tech-nology to build DDoS attacks.Based on a large number of domestic and foreign literatures,this paper proposes a defense system based on DPDK(Data Plane Development Kit)and Na(?)ve Bayes algorithm.In the research process,through the experimental analysis,the decision tree(Decision Tree),Support Vector Machine(Support Vector Machine),Na(?)ve Bayes algorithm,Artificial Neural Network,Fuzzy logic(Fuzzy)were compared.Logic)and other machine learning techniques,through repeated experiments,have proved that the Na(?)ve Bayesian algorithm has better effects in dealing with DDoS attacks than other machine learning techniques.In turn,the research system combines machine learning technology with the traditional DDoS defense meth-od designed by DPDK to build a new defense system that is more in line with the development trend and effectively responds to the new DDoS attack.Compared to traditional DDoS hardware protection devices,this system has several ad-vantages:1.Reduce production costs.The protection system runs on a common server.Compared with the expensive cost of traditional protection equipment,the new system can greatly save production costs without affecting the defensive perfor-mance,and is easier to promote.2.High availability and portability of the system.The system can be inte-grated into the link system without changing the existing network topology,which is more convenient for maintenance and expansion,and reduces operation and maintenance costs.3.The integration of Na(?)ve Bayesian algorithm and traditional defense strategy.Taking advantage of machine learning technology,it plays a good role in protecting the DDoS attack method,which is not easy to be logically identified and new.At the same time,it gives full play to the advantages of the stability of traditional defense methods,and gives more problems in parameter adjustment in machine learning technology.Response time to improve protection.The defense system and method proposed in this paper have been proved to be able to cope with many traditional DDoS attacks,and also show good defense effects for some new attack methods.The features of convenient expansion and transplantation also have a good application prospect.