A Dynamic Trust Model Based on Recommendation Credibility in Grid Domain

Abstract

Grid is the next generation of Internet, and it has attracted the attentions of research communities recently. In practical applications, the nodes in the grid domain need to transact with other strange nodes to acquire the services provided by those nodes. Recently, trust has been recognized as an important factor for Grid security. And it is important for them to make a trust evaluation for those strange nodes before transactions, because some nodes are dishonest. Currently, the method to solve this problem is to establish an effective trust mechanism in grid domain based on reputation. However, there are still some disadvantages in existence in some cases, as the dynamicity of the grid environment, although it can be risk-averse to some extent. Like some malicious nodes contribute high trust value through numerous good interaction behaviors at first, and then change to attack. In view of those problems and disadvantages, this paper proposes a dynamic grid trust model based on recommendation credibility, called DyGridTrust (Dynamic Grid Trust Model). This model can distinguish honest and dishonest recommendation and adjust the weight of trust evaluation dynamically. In addition, this trust model is able to obtain the trust value more objectively, so as to enhance the security of the grid system. Finally, in this paper, those perspectives have been confirmed by simulations, which are developed by Matlab R2008a.