Multi person tracking and querying with heterogeneous sensors

Tracking the location of a user is considered to be the most fundamental step for creating a context aware application such as activity monitoring in an assistive environment. The problem becomes very challenging, if the number of people involved in the scenario is larger than one. The reason is that any multi-person environment such as a hospital demands a simultaneous identification and localization mechanism, thus making the system extremely complex. In this dissertation, we present a novel, less-intrusive system that uses RFID and sensors deployed at various locations of an assistive apartment to continuously track and identify every person in a multi-person assistive environment. In addition, the system stores the large scale spatio-temporal sensor data into a common repository and provides a flexible query interface to track the history of the patient. The visualization tool embedded to the system helps the therapists to remotely monitor a person present in a scene in near real time. Such a visualization gives a very good indication about a person/patient's activity and behavior in the assistive environment as well. The system also incorporates the metadata mapping of the large amount of stored data so that a doctor/therapist can query about a patient records without even knowing all the schemas stored in the repository.