Research On The Theory And Applied Technology Of Network Intrusion Tolerance

Intrusion tolerance is a rising network security technology which combines the cryptography and the fault tolerance. The concern of intrusion tolerance is not how to detect or prevent intrusion, but how to mask or restrain intrusion. Its goal is to continue to guarantee the confidentiality and integrality of data as well as the usability of service when the network system has been intruded.There are many unresolved problems in intrusion tolerance although the corresponding theory and technology have been studied extensively by the domestic and foreign academia. In this dissertation, the modeling, analysis of performance, scheme of secret sharing, and design method of the intrusion tolerant system are studied. The main contributions are listed as follows:(1) A model of distributed intrusion tolerant system based on Petri net is proposed through summarizing and analyzing the work states and the state transition property of intrusion tolerant system. The detailed description process of the model to the intrusion tolerant system is given, and the advantages of the model on description performance compared to other existing models are analyzed. On this basis, a quantitative model of intrusion tolerant system is proposed through utilizing the knowledge of game theory. The income functions of the intrusion tolerant system and the intruder are designed, and the Nash equilibrium of the game system is achieved through analyzing the optimum strategies of the two game sides.(2) The performance of intrusion tolerant system is analyzed from three aspects: security, security situation, and resource cost. First, a more comprehensive and more feasible computational method of security for intrusion tolerant system is proposed through the multi-dimensional definition for security and the decomposition for intrusion. Then, a method which can evaluate the security of intrusion tolerant system dynamically and timely is proposed through designing the three evaluation indexes: the confidentiality of data, the integrality of data, and the usability of service. Lastly, the resource cost of intrusion tolerance is quantified. The concepts of continuance cost and response cost in intrusion tolerant system are proposed, and some methods and notes of reducing the resource cost are analyzed through simulation.(3) The access structure of secret sharing in intrusion tolerance is studied. First, a secret shaing scheme which can meet the requirements of both the (t, n) threshold and the adversary structure is proposed basing on the existing (t, n) threshold schemes and the knowledge of set theory. The scheme can distribute the shadows of shared secret according to the requirements of threshold and adversary structure directly. Then, a dynamic proactive secret sharing scheme which can be applied to general access structure is proposed through introducing the concept of adversary structure in the existing (t, n) threshold scheme. The scheme can change its access structure dynamically when its shadows are renewed. Lastly, a threshold signature scheme based on the general access structure is proposed through combining the secret sharing scheme based on the general access structure and the (t, n) threshold signature scheme. The applied range of the scheme is wider compared to the existing (t, n) threshold signature schemes.(4) An intrusion tolerant system based on neural networks is proposed through analyzing the two attributes of resource and control. The corresponding security mechanisms on different work states and the work process of the system are given. On this basis, a more feasible intrusion tolerant system based on trust-based access control is proposed through applying the access control to the design of intrusion tolerant system, and the intrusion tolerant performance of the system is analyzed in detail.Finally, the main contributions of this dissertation are summarized, and the shortcomings and future work are pointed out.