Format of Original
Journal of Mobile Multimedia
Collaborative effort to share resources is a significant feature of pervasive computing environments. To achieve secure service discovery and sharing, and to distinguish between malevolent and benevolent entities, trust models must be defined. It is critical to estimate a device's initial trust value because of the transient nature of pervasive smart space; however, most of the prior research work on trust models for pervasive applications used the notion of constant initial trust assignment. In this paper, we design and implement a trust model called DIRT. We categorize services in different security levels and depending on the service requester's context information, we calculate the initial trust value. Our trust value is assigned for each device and for each service. Our overall trust estimation for a service depends on the recommendations of the neighbouring devices, inference from other service-trust values for that device, and direct trust experience. We provide an extensive survey of related work, and we demonstrate the distinguishing features of our proposed model with respect to the existing models. We implement a healthcare-monitoring application and a location-based service prototype over DIRT. We also provide a performance analysis of the model with respect to some of its important characteristics tested in various scenarios.