Mobility Profiling for Mobile Trajectory Based Services

There has been extensive research and development on Location Based Services (LBS). The location in this context is either, an exact mobile location or a large area, cell, or several cells, referred to as a location area (LA). The focus is on the present location of the mobile user. The focus of our research is on the mobility profile of a mobile user in the form of spatiotemporal distribution, i.e., a traveled trajectory. The profile consists of the mobility of each day of the week. A variety of personalized services can be built on each user's profile, hence the birth of another class of mobile services; we named it Mobility Trajectory Based Services (MTBS). All of the location estimation techniques use air resources extensively. With the implementations of MTBS, the location estimation signaling takes place only during profiling. There are two major challenges in profiling the mobility of a mobile user: (1) location estimation error, and (2) location estimation polling frequency. The required estimated location accuracy is different for different LBS. Some services require an accurate location, whereas some would work with cell ID information. We investigate the suitability of location estimation technologies for mobility profiling. We propose an adaptive polling method, such that the polling is not so frequent as to overburden the air resources unnecessarily, and it is not so sparse that it leads to an ambiguous and erroneous mobility profile. The polling method includes a new map-matching scheme based on the previous location, map topology, and travel time from one intersection to the other. To deal with the difficult problem of identifying candidate road segments within the feasible area, we propose geodesic candidacy conditions. We use point-to-line map-matching technique to translate the raw erroneous location points onto a road on the map. Instead of correcting each location point, our focus is on matching the estimated trajectory to the actual one, selecting from a set of possible trajectories from a GIS (Geographical Information System) map. Such maps are available for most of the developed world. So, from raw, moderately polled location points with time stamps, we shall form the spatiotemporal trajectory for a mobile user. A set of such trajectories for all days of the week would form a mobility profile, on which a plethora of MTBS could be provisioned using the proposed MTBS architecture and provisioning protocols.