Research On Detector Generating Algorithm Based On Improved Model Of Kim

In recent years, immune-based intrusion detection technology has becomea key research field in intrusion detection system. Its prominent character isthat it can explore natural immune logical theories, mechanisms and principlesfor detecting and reacting to intrusions.At present, the majority of commercialized intrusion detection productsjust adopt the simple template match technique, which can be only adapted forsome simpler attack modes with a high misrepresentation rate and just candetect known attack modes. Whereas immune-based intrusion detection systemcan detect unknown attack modes according to immature information, whichhas great actual significance. Several aspects were probed into as follows.Firstly, on the basis of studying the working mechanism of artificialimmune system completely the paper makes some improvement on existingartificial immune model. The modification includes design of detector,co-stimulation, mutation and evolution of memory strings, making the artificialimmune model easier to realize. The paper also devises a network-basedintrusion detection system according to the improved model. The IDS takesnetwork packets as detecting data resource and has the advantages of detectingunknown intrusion, distributed disposition.Secondly, the paper researched on the detector generating algorithmsapplying in the process of negative selection for the model, which is the keyalgorithm in the system. By the analyses on the existing detector generatingalgorithms, we achieved the improvement of the linear time detector generatingalgorithm, which was widely used at present. The efficiency of the linearalgorithm was improved by eliminating the reluctant detectors, and at the sametime the improved algorithm was ensured to cover the nonself space as much aspossible.Afterwards, on the basis of generating algorithms, analyzing the reason ofdetecting holes. The thesis proposed a kind of new algorithms on computingdetection holes and summarized the detector generating algorithms, at the sametime, pointing out the questions existing in the algorithm.Finally, the initialization detector set scale and the failure probability ofdetector generating algorithm have been tested and the test results indicate thatthe improved algorithm enhanced the maturity and the reliability contrastingwith the former algorithm.A new way was introduced to the fields of computer safety system throughthis thesis. The proposed detector generating algorithms has great potential.