The Research Of Distributed Ids Based On Improved BP Neural Network

Along with the coming of the information society, Enterprises,governments, other organizes and personals have become dependent on the computer network system more and more. It is imperative to protect these systems against being attacked. IDS (Intrusion Detection System) has become the most important part of the network security, Distributed IDS is the developed trend in local and massive network.The article research on the class method standard of the intrusion and the intrusion detection ,give a systematic research on IDS, widely sum up the application and the research of the IDS based on neural network, propose the NAIDM (the distributed Neural Agent-based Intrusion Detection Model), discuss the implement of the model . the article deeply realize and research its structure design, algorithm improvement, emluator test, aiming at the core of the static intelligent Agent based on neural networkAmong the NAIDM, many different network source nodes are characterized with different intrusive types,so these nodes need collecting and coding different charactersistic vector, training and learning different intelligent Agent based on neural network . The neural network of the intelligent Agent can liken or unliken, in this article the author propose a kind of ideology that design and train different intelligent based on neural network, discuss the mutual help and collaboration of many intelligent Agent.In order to design and achieve an intelligent Agent module, author investigate BP algorithm and its improvement, study and test the data collection and vector distilling on important network nodes, the emluator test obtain satisfying result after designing and training.