Based Decision-making Associated With Intrusion Analysis And Research

The network technologies develop fast and there are the kinds of the intrusion technologies, so the traditional intrusion detection technologies can't satisfy needs of the intrusion detection. The new intrusion detection technologies must be applied to intrusion detection for adapting the changes.The decision tree has advantages of simple structure, well understanding, high precision and no parameters, so this dissertation applied the decision tree to intrusion detection. In this dissertation, when we select attributes for deciosin tree, we select the attributes whose correlation is the maximal between its father attribute. The experiment proves that it increases the detection ratio.In order to adapt the feature that the structure of the Ad Hoc network changes fast and is various, this dissertation applied association rule to construct anomaly detection system. The anomaly detection system can detect the ways that are not in normal ways.In order to reduce the false alarm ratios, this dissertation used a classifying implement which is construted with decision tree to classify the abnormal rules.The rules putted into the classifying implement may include new attacks, but the classifying implement can't detect all of them. This dissertation introduced multi- criterion pruning with power to improve the ability that decision tree detect new attacks.