Agent-based Movement Analysis and Location Prediction in Cellular Networks

The huge quantity of individual location data recorded by mobile phones or mobile networks inspires research questions regarding new and intelligible services on behalf of mobile users. Creating a system for location prediction and following user habits is a sizeable and challenging task. We present a system for movement analysis and location prediction of individual mobile users. The goal of movement analysis is to determine which locations are significant to the user and to classify locations according to user-specific regularities. The analysis is performed by multi-agent system in which individual agent performs a chosen set of functions. The prediction system consists of two predictors which work in parallel, based on the type of user-specific locations. Prediction system is intended to estimate the future location of a user device and provide it to user-authorized applications and services. The proposed system is evaluated using real movement data recorded as part of the MIT Reality Mining Project.