Multinomial Subjective Logic Based Extended Trust Propagation Model

With the development and popularity of network technology, the traditional computingmodel does not meet users’ requirement. The network computing model of distributed system(such as P2P, cloud computing, grid computing, Ad Hoc network)has gradually taken theplace of the traditional computing model and quickly became the mainstream of networkdevelopment. Distributed network is characterized by its openness, dynamic property andwithout center. Entities’s behavior is characterized by uncertainty and is affected by thoseenvironmental things. Meanwhile, application systems lack the restraint mechanism onentities’s behavior. These resulted in the lack of trust between the entities. The securitydemand of distributed system is not satisfied with the traditional means of network security.Therefore, the security issue of distributed network has become a focus of the higherresearchers.Based on the character that with distributed network, which is a social one, the human isthe active part in social activity, at the same time, trust is the core of human relationships, theeffective way of solving the security issue of distributed network is to build a trust mechanism,that is, to build and manage trust relationships of entities. The trust propagation is a keyproblem in trust model. Due to the subjectivity and uncertainty of trust, structuring efficientand precise trust propagation model is still a challenging and very significant issue.Subjective logic trust model that is provided by J sang based on Beta distributionfunction of posterior probabilities of binary events. Trust propagation operator of this modelaiming at the binomial opinion, neither makes any assumptions about the reliability of trustsources, nor considers environmental factors, with some limitation. In light of this, this paperproposes a multinomial subjective logic based extended trust propagation theoretical modelthat is composed by the trust fusion and trust transitivity. Trust fusion is the base of trustpropagation. Making the fused opinion more accurate and more in line with the intuitivejudgment of human, reputation-based fusion operator of multinomial opinions takes effects ofthe reputation value and observation circumstance into consideration, puts forward relatedconcepts, and gives reputation-based fusion operator of multinomial independent opinions,dependent opinions and partially dependent opinions, according to the different time at whichtrust evidence is attained. Trust transitivity is the manifestation of trust propagation andmultinomial subjective opinions is a general form of trust expression. This paper interprets multinomial opinion transitivity as meaning that transitivity process is binomial logicalopinion and transitivity information is multinomial opinion, giving trust transitivity formulaof multinomial opinion and proving related theorem. Finally, the adaptability and accuracy ofthe model are put to the proof by example analysis.