Context-Aware Profiles

Abstract

A spatial proximity based spontaneous interaction approach for "shared nothing" peer-to-peer systems is proposed. Mobile peers that come spatially close to each other wirelessly exchange self describing profiles encoded in an XML dialect. Each peer individually conducts a similarity analysis of the semistructured data contained in the received profile, and, in case of the similarity exceeding a predifined threshold, notifies an application to start executing. Profiles are organized hierarchically, allowing for an efficient interaction check based on peer roles in the first step, and a more detailed analysis of peer preferences in the second step. A novel context-aware profile description language (PPDL) has been developed, expressing peer preferences with respect to the particular situation (like time, geoposition, owner, environmental conditions, etc.) in which the interaction among peers is attempted. The context of a peer is modelled symbolically within its profile, so that sensors can verify the actual context and use the context specific interpretation of the profile prior to similarity analysis. A fully functional software framework has been developed implementing context-aware profiles. Similarity metrics and profile matching strategies are presented and assesed. In a flexible manufactoring systems applications scenario we demonstrate its potentials in the context of wireless peer ensembles.