Artificial Neural Network Technology Application In Intrusion Detection System

Network openness makes it vulnerable targets, the whole society depends oncomputer network is increasing also makes network crime has become more and moreserious.so network crime is also becoming more and more rampant. So, networksecurity turns into the hot spot in Computer science research. The intrusion detectionsystem is a normal scheme in computer network security, which collects informationaccording to the computer network or some keys of the computer system, and theintrusion detection system analyzes the information in order to discover whether there isthe behavior against the safe strategy and the evidence of being attacked in the networkor system or not.Based on the model of the intrusion detection system which was givenby predecessor.However, the existing intrusion detection methods cannot detect all intrusion,intrusion detection rate is not high, detection algorithm of single and comprehensive useof various means of detection. The purpose of this study is to explore an effectivemethod for detecting, from the history of network audit data for normal behavior andabnormal behavior rule, extraction of normal and abnormal patterns of behavior, to ahigher level of data analysis, thus completing the single packet pattern matching cancomplete the type of attack detection. Final completion of a neural network basednetwork intrusion detection system design, so that the correct rate and availability in arelatively ideal range, and in a specific environment to achieve.Based on existing intrusion detection system model, This thesis researched anddesigned a neural network based intrusion detection system, and tested is on a campusnetwork platform. The majority of research is as follows:(1) From the current situation of network security, analysis of the need for intrusiondetection technology, introduces the neural network technology, intrusion detectionsystem described the basic principle, research status and future development trend.(2) Application of BP neural networks to the intrusion detection technique,structure based on neural network intrusion detection system, system includes dataextraction, decision module six, a detailed analysis of the neural network classifierconstruction.(3) To validate the system performance, a good deal of experimental data was prepared. Intrusion simulation has been done in inner network and the intrusion datawas analyzed deeply. The system was trained and tested by both normal data andintrusion data that have been captured.(4) Experimental results are collected and analized, which show that the neuralnetwork system has a high detection rate and low false alarm rate to known data schema,and has a good capacity to recognize unknown data schema too.Finally, the paper summarizes the research done, and the subject of further workare discussed.