Possibility-based trust for mobile wireless networks

Abstract

Managing trust in mobile networks necessitates dealing with uncertainty, resulting from incomplete, duplicate or conflicting evidence. In this work, possibility theory is used to deal with uncertainty, and a trust model is developed that includes the network nodes’ integrity, their competency and that of the connections among them. Evidence obtained by direct observation (‘passive’) is integrated with that solicited from other nodes (‘active’) to generate possibility distributions, or trust profiles, that can be combined, updated and discounted. Simulations of mobile ad hoc networks show that the inclusion of these trust profiles in routing decisions conserves network resources and limits data exposure by directing packets over reliable links through uncompromised nodes. Possibility theory enables the information gain provided by different forms of evidence to be evaluated, and this is used to explore the benefits of different sources of active trust evidence. It is seen that passive evidence is not sufficient to maintain trust profiles with low uncertainty, and that active trust evidence obtained by authentication requests provides higher informational value than that received using reputation solicitation.