A Formal Model for Security Analysis of Trust and Reputation systems

摘要

Trust and reputation systems (TRSs) are used as decision making criterion in many modern systems. In these systems normally a set of agents cooperate with each other to achieve system/own goals based on ethical norms of the system. The security of a TRS against malicious agents who try to bypass the ethical norms of the system can be evaluated using simulation or verification techniques in which both of them need to have formal models of TRSs. There are a few works who tried to present formalisms for modeling TRSs; however most of them are either unsophisticated or improper for checking security properties. In this paper we introduce a formal model of an agent interaction system along with the underlying TRS which is capable of formalizing the attacker behaviors. The presented formalism is suitable to be used in simulation or verification schemes for analyzing the security and robustness of TRSs. To demonstrate the sufficiency and capabilities of our model, eBay reputation system as a famous TRS is formalized with the presented model.