Applying Clustering to a Framework for Generating Trust

This paper addresses the issue of trust within the ad hoc context. Several models which claim to model trust are evaluated and a trust framework is then devised which bases itself on clustering technology. Our model aims at providing trust information about originally unknown nodes while making optimum use of computational capacity, which can be quite scarce in pure ad hoc networks. The use of trust data to generate relationships between nodes is therefore strongly favoured to applied cryptography, which generally involves intensive resource consumption. The method proposed also draws on statistical derivations to propose a condition of normality while attempting to provide definition to behaviour.