Research On Key Technologies Of Network Security Management Based On Artificial Intelligence

With the development of network technology and the progress of information science, the importance of network security and management has already become a global unavoidable problem. The researches on network anomaly detection and trust management model based on artificial intelligent theory have become not only the tastes of scientists but also the interests of governments and forces. The government and industrial communities of many countries and regions are so keen on researches on network anomaly detection and trust management model based on artificial intelligent theory that they have invested a large amount of money on corresponding research. Therefore the progress of the researches on network anomaly detection and trust management model based on artificial intelligent theory will not only promote the development of science and technology but also influence the national powers and defenses.This dissertation makes researches on some issues of anomaly detection method and trust management model based on artificial intelligent theory. The main contributions of this dissertation are summarized as follows:(1) A novel anomaly intrusion detection method based on CSA (Clonal Selection Algorithm)-based unsupervised fuzzy clustering in order to solve the problems of recent anomaly detection method with low detection rate and low false positive rate, and k-means algorithm which is particularly sensitive to initialization and fall easily into local optimization was presented. This method can quickly obtain the global optimal clustering with a clonal operator which combines evolutionary search, global search, stochastic search and local search, then detect abnormal network behavioral patterns with a fuzzy detection algorithm. Simulation results on the data set KDD CUP99 show that this method can efficiently detect unknown intrusions with lower false positive rate and higher detection rate.(2) Two new combination rules based on conjunctive and complementary pooling criterion, which were used to solve the combination problem of conflict evidence of information fusion and the drawbacks of Dempster rule and improved rule of combination was presented. First, the advantages and disadvantages of Dempster rule of combination as well as several improved combination approaches proposed so far were analyzed. Then, a new mass function based on proportional belief criterion of conjunctive and complementary belief was combined. Finally, the results of numerical examples show that the proposed approach of combination can not only maintain the advantages of original Dempster rule of combination, but also efficiently solve and make up the Zadeh counterintuitive contradiction, loss of majority opinion, robustness and fairness of original and improved Dempster rules of combination.(3) A novel trust model for peer-to-peer network based on proportional conjunctive and complementary pooling criterion based evidence combination rule, which was used to solve the problems of not effectively defending against several kinds of attacks issued by malicious peers, aggregating inconsistent recommendation and dealing with uncertainty of information in the reputation based peer-to-peer trust model was presented, in which the uncertainty degree between trust relationships was considered, and more valid and effective fusion results based on preprocessing and combination of recommendation evidence were developed. The experimental results show that compared to some current trust models, the proposed model has advantages in increasing the successful transaction rate and evaluating the trust degree of peers. Moreover, the proposed model is more robust and advanced in defending against denigrate, collusive and strategic attacks.