Fuzzy reputation-based trust model

We present PATROL-F (comPrehensive reputAtion-based TRust mOdeL with Fuzzy subsystems) as a comprehensive model for reputation-based trust incorporating fuzzy subsystems to protect interacting hosts in distributed systems. PATROL-F is the fuzzy version of our previous model PATROL, and aims at achieving a truly unique model incorporating various concepts that are important for the calculation of reputation values and the corresponding decisions of whether or not to trust. Among the incorporated concepts are direct experiences and reputation values, the credibility of a host to give recommendations, the decay of information with time based on a dynamic decay factor, first impressions, and a hierarchy of host systems. The model also implements the concepts of similarity, popularity, activity, and cooperation among hosts. In addition, PATROL-F's fuzzy subsystems account for humanistic and subjective concepts such as the importance of a transaction, the decision in the uncertainty region, and setting the result of interaction. We present simulations of PATROL-F and show its correctness and reliability.