MATE: Mobility and Adaptation with Trust and Expected-utility

Mobility is one of the defining charact eristics of many pervasive environments. As pervasive devices move, they experience new contexts (e.g., new network boundaries or new peering devices) to which they may adapt by loading new software components or by updating existing ones. Unfortunately, loading and updating code introduces security problems, particularly because components may be from untrusted sources with which devices happen to be networked at the time of need. Thus, a pervasive device must be able to distinguish between trustworthy and untrustworthy peering devices so that they load software components only from trustworthy ones. To achieve this, the device may run a software agent that comprises a trust mode l and a decision-making model: the former manages information about peering devices' tr ustworthiness upon which the latter decides appropriate actions, e.g., whether to load software or ask the user for further guidance. We here propose a decision-making model named MATE, which works as follows. It lists potential risks depending on the device's context and running application. When it has to decide whether to load software from another device, MATE assigns occurrence probabilities to those risks in two ways. In the first approach, the probabilities depend on the peering device's trustworthiness (i.e., the output of the trust model). In the second approach, MATE computes the probabilities using a Bayesian formalization that works without an additional (trust) model. After computing those probabilities, MATE applies the expected-utility decision rule to decide whether or not it is in its node interests to load a particular component.