Research Of IDS Based On The Improved Levenberg-Marquardt Algorithm

With the continuous development of computer networks, the global informationhas become the major trend of human development. But because the computernetwork has joint form diversity, the uneven distribution of the terminal and networkopenness, connectivity and other characteristics, resulting in network vulnerableagainst the attacks, the security of online information and confidentiality is a criticalissue. For the military command automation network, banking, e-commerce andother transmission of sensitive data to the computer network system, the networkinformation security and the security is very important. The network securitymeasure should be able to deal with all kinds of threats and vulnerabilities, so as toensure that the network of information confidentiality, integrity and availability.Based on the neural network research, the intrusion detection system detectionmechanism for learning, the study focused on the BP neural network algorithm andLevenberg-Marquardt algorithm, and using Levenberg-Marquardt algorithm, the BPneural network optimization algorithm, combined with Levenberg-Marquardtalgorithm, through research and comparison, the existing intrusion detection systemsbased on the optimization Levenberg-Marquardt algorithm and one based on theoptimization algorithm of the model. The final draft of the" algorithm" and" model"simulation, verification of the Levenberg-Marquardt algorithm and the model isfeasible and effective.First of all, the research from the intrusion detection system and neural networktheory, the intrusion detection system is analyzed. The basic structure, function,in-depth study of the intrusion detection system working mechanism, analysismethod, and the study of intrusion detection system the direction of futuredevelopment and evolution trend. At the same time, the principle of neural networkare discussed, on the BP neural network algorithm and Levenberg-Marquardtalgorithm, the algorithm combines the advantages of two, form the improved Levenberg-Marquardt algorithm, and applies the improved Levenberg-Marquardtalgorithm applied to intrusion detection system, based on the existing intrusiondetection systems, forming the improved model of intrusion detection system, andgives the system structure, introduces the improved intrusion detection system eachmodule and the work flow of each module.Finally, the improved intrusion detection system simulation experiment,experimental results are given, through the analysis of the simulation results, thedescribed earlier detection algorithm and the improved system results, conclusion.