Using Context Information to Improve Computation of Trust in Ad Hoc Networks

Trust and reputation mechanisms play important roles in large open environments. In order to have more precise trust evaluation, trust models try to gather as much information as possible. The most used source of information is interaction history between agents in the system. Besides, some trust models consider additional information; context is one of them. However, there is no consensus over what is considered as the context and how it is modelled. In our work, some of the conditions under which an interaction happens may influence the perception of its result; these conditions are what we consider as the context. In this paper, we present how we consider context information to improve a Bayesian-network based trust model. We show through simulations that it helps agents in obtaining a more precise trust evaluation in dynamic environments such as ad hoc networks.