Research And Design Of Network Trust Evaluation Method Based On Context Correlation Analysis

With the development of trust computing technology, and the raise of higher requirements for network security, the concept of trustworthy network is proposed. With the mechanism of trust computing,trustworthy network is guaranteed by the chain of trust,which starts from the root of trust in the computer, and transfer trust up to the hardware platform, operating system and applications layer by layer. The network is considered trustworthy as long as the user, node, communication and service in the network is trustworthy. However, with the evolution of network technology and application, the scale of network has become larger and larger, and the various behaviors in the network has become much more complicated and space-time related. At the same time, the attacks against network entities have become much more difficult to detect and defend so that there is a great possibility that the network entities may be captured by the attackers. Faced with these challenges, the traditional static trust computing method which is based on control mechanism could have many problems. It cannot evaluate trust dynamically with the change of network contexts and status. To solve this problem, a dynamic trust evaluation mechanism based on behavior and content analysis is proposed in this paper. In this mechanism, a network is considered trustworthy only when the network’s functions operate well. Based on the knowledge of trust model and context-aware systems, this paper mainly focusses on the following two parts.A context-aware trust evaluation method based on behavioral data analysis in distributed network environments is proposed. The method consists of four steps. Firstly, the behavioral data collected by the monitoring sensors are normalized and stored in the behavior record database. Secondly, we propose a correlation analysis algorithm to filter out the irrelevant context properties. Thirdly, all relevant historical records are extracted based on the context similarity, which is calculated based on the relevant context properties obtained from step two. Finally, a trust value prediction algorithm for the network entity is proposed based on aforesaid three steps.An entropy based self-adaptive multi-attribute ranking method for node importance in complex networks is proposed. Firstly, we define four indicators to reflect different characteristics of the network structure. Secondly, we calculate the weight of different indicators based on information entropy theory with the evaluation results of the four indicators in the first step. Finally, the rank result of node importance is obtained by weighted average method with aforesaid steps.By combining the trust and importance of each node in the network, the trust value of the whole network could be evaluated.