基于人工神经网络的入侵检测系统的研究与实现-计算机应用技术专业论文.docxVIP

贵州大学硕士学位论文英文摘要 贵州大学硕士学位论文 英文摘要 Research and Implementation of Intrusion Detection System-based on Artificial Neural Network Abs订act Intrusion detection is active security-defence technology which protect the network resoul℃e$，it is the effective supplement to such information security measu∞s as”firewall”。’’ data encryption”，’’access control”，etc，，and be used for identifying and responding to the malicious using behavior the computer and network ltsource．Intrusion detection system not only Can detect the attacks from outside network，but also Can supervise unauthorized behavior of inside network at the same time． Seeing that the traditional intrusion detection techniques exit various difficulties(for instance， it relies mostly feel and experience to meusUlt behavior,it only depict attack characteristic by some regular army,ere．)，the paper proposes kind of modeling selwme of real—time intrusion detection based BP artificial neural network．First，the paper analyzes the header structure of data p∞ket in皿TCP,UDP and ICMP under Ipv4．Then，it especially discusses the implementation principle of various comlTlon network刮挑k8．Finally,utilizing the local network ofthe laboratory,according to some networl【柏ck it designs and implements simple real-time intrusion detection system based BP artificial neural network under Linux (kemel edition 2．4．20-81． The system adopts the development toolkit—Libpoap fPacket Capture Library)specially designed to use in the data monitoring application program to implement network data gather． Thus．血e system meet such demands弱larger flow network and 10wer lusing p锄eket rate，ere．． Utilizing pcap_loop0 function among the library,the system inputs network datmn got to catch into call-back function．According to the implementation principle of network anack it COents to the characteristic attribute value ofdata packets ofrelevant connection in specific time window(2 seconds in the system)，Combining these characteristic values．the system forms the necessar