Detecting Collusive Agents by Trust Measures in Social IoT Environments: A Novel Reputation Model

In the Internet of Things, smart objects can build multidimensional and context-sensitive network infrastructures potentially rich of social interactions. Smart objects can be associated with software agents to boost social interactions and realizing complex and sophisticated forms of collaboration of objects with both other objects and people. In such a scenario, there exists the possibility to interact with unreliable partners exposing agents to the risks deriving by malicious behaviors. To mitigate these risks, Trust and Reputation Systems can be adopted to provide each agent with appropriate trustworthiness measures about the potential counterparts in order to select the best ones. In this context, our contribution consists of (i) a method to preliminarily identify the best candidates as malicious in order to consider them as pre-untrusted entities and (ii) a novel effective reputation model able to detect collusive malicious agents without introducing collateral effects with respect to the reputation scores of honest agents.