An Evidence-Based Mobility Prediction Agent Architecture

One of the major challenges in wireless environments is the provision of Quality of service (QoS) guarantees that different applications demand considering the highly dynamic nature of these environments. User mobility prediction represents a key factor for providing a seamless delivery of multimedia applications over wireless networks. Most of the existing approaches for mobility prediction presume that users travel in a-priori known pattern with some regularity; an assumption that may not always hold (e.g., a tourist in a foreign city). This paper presents a novel architecture of a mobility prediction agent (MPA) that accurately performs mobility prediction using knowledge of user’s preferences, goals, and spatial information without imposing any assumptions about the availability of his movements history. Using concepts of evidential reasoning of Dempster-Shafer’s theory, the MPA captures the uncertainty of the user’s navigation behavior by gathering pieces of evidence concerning different groups of candidate future locations. These groups are then refined to predict the user’s future location when evidence accumulate using Dempster rule of combination.