Research On A Paralleled Feedback Structure Of IDS

With the development of high speed network technology, the analyses speed of the centralization IDS falls highly behind data flowing speed on the internet. This situation leads to under-reporting problem which would impact detective veracity and validity. It was for the moment that IDS should set primary task to advance velocity and efficiency. Aiming at solving these problem, the thesis adduces a paralleled feedback structural IDS model. It divides high speed data packets flow into several smaller streams, which are sent to one clusters of computer. They can be analyzed respectively on this computer clusters. The system can adjust streams divided to each detection modular through the medium of dynamic feedback allocation mechanism, and the burden balance algorithm guarantees to hold intrusive information fully. By this way, the bottle neck of the data packets flowing on the internet can be abolished. A test demonstrates the system can advances total detective performance.