The overall aim of the project is to develop a model and middleware for the management of, and reasoning about, trust for pervasive systems. The novel trust management model will be centered around the idea of groups of trusted agents. Like in human societies, computational trust is built on top of historical data (past direct experiences and recommendations); highly volatile groups will bring little benefit to the problem of accurate trust prediction, as the building block represented by history will be missing. In this project, we thus wish to:

• characterise the circumstances under which stable groups of trusted agents form and how they evolve;
• design a novel trust model that supports trusted group management and reasoning;
• define metrics to assess the quality of a trust management model, and build a simulation framework that can be used to evaluate models based on these metrics;
• derive a trust management framework from thedeveloped trust model, and use the simulationframework to compare it with others;
• embed the trust management framework into a mobile computing middleware and apply it to a real pervasive computing application (see Utiforo project).

Dr. Licia Capra

This research is funded to the extent of approximately GBP 210,000 by the EPSRC (Sponsor reference: EP/E009190/1)