Research Of The Trustworthiness Of Nodes' Behaviors In The Distributed Networks

In order to ensure the safety of all the terminal devices accessed to networks which are complex, heterogeneous and dynamic, how to make networks trustworthy needs to be researched.With the rise of distributed networks, the research of trusted distributed networks becomes important. Trustworthinesses of the nodes in the network is the foundation of the trustworthiness of the network. Trustworthinesses of the nodes include the trustworthiness of nodes' identities and trustworthiness of nodes'behaviors. Traditional security mechanisms such as authentication and authorization can assure the trustworthiness of nodes'identities, but can not tell whether the behaviors of the nodes are trustworthy. This paper focuses on ensuring the trustworthinesses of the nodes'behaviors in the network. The main research contents and the results of this paper are as follows:Trust and Reputation Management System is an important means to solve the security challenges that cannot be solved by the encryption and authentication based traditional security solutions. However it is difficult to make comparision between different models. To deal with this problem, this paper researches on a approach of unified evaluation criteria for trust models' defense capability:abstracting the kernel of trust model, building a common secure trust model and then designing a secure trust model simulator.The performance of a trust and reputation management system may decline in potentially adversarial environments, such as a network in which there are malicious server and malicious recommender. In this paper, we listed a number of security threats applicable in the field of trust and reputation management by analyzing the behaviors of the malicious nodes. And then we proposed statistical anomalies detection algorithms to cope with these threats even in the networks which has a great proportion of malicious nodes. The reliability of the information used by trust and reputation management system can be ensured with these algorithms.Traditional trust and reputation models do not pay enough attention to nodes'deceit in recommendation, so their reputation evaluation may be affected by malicious nodes' collusion. This paper presents a trust and reputation model named EAraTRM, which is based on ant colony algorithm. Node recommendation behaviors analysis and adaptive mechanism to malicious nodes density were added into reputation evaluation of EAraTRM, in order to overcome the shortage of traditional models. Simulation experiments show that EAraTRM can restrain the collusion of malicious nodes, and give more accurate reputation evaluation results, even when 90% nodes in a network are malicious.The nodes in the distributed networks must be robust enough to most security attacks, otherwise they might be controlled by malicious users, and the data or services they provide might be tampered. This requires the operation systems or the applicaions of the nodes have no known vulnerabilities. Patch package management systems (PMS) can fix the vulnerabilities that might be utilized by attackers. The main contributionsof this paper are proposing a new model about the threats faced by PMS nowadays. Analyzed the common attacks which could be appliled on PMS and the defense machnisms to these attacks, this paper defines a set of empirical security design principles of PMS for the first time. In addition, secure PMSs are designedand implemented in accordance with these principles on different operating systems.