Multi-sensor data fusion methods for indoor activity recognition using temporal evidence theory

Information fusion has been widely used in context aware applications to create situational awareness. In this paper, a multi sensor fusion methodology using temporal evidence theory is proposed for indoor activity recognition. The fusion method develops an incremental conflict resolution method within the Dempster-Shafer (D-S) theory framework. This method has distinct advantages over the proportional conflict resolution technique of D-S theory. The key contribution of this paper lies in introduction of temporal information into the fusion methodology in a multi-sensor environment. The proposed framework is used for activity detection in smart homes. Two smart home datasets are used in the experiments on activity recognition wherein the data is recorded through a series of passive sensors. The experimental results obtained for activity recognition are motivating enough to be useful in applications like assisted living.