Sensor Data Fusion Using DSm Theory for Activity Recognition under Uncertainty in Home-Based Care

Reliable contextual information of remotely monitored patients should be generated to prevent hazardous situations and to provide pervasive services in home-based care. This is difficult for several reasons. First, low level data obtained from heterogeneous sensors have different degrees of uncertainty. Second, generated contexts can be corrupted or conflicted even if they are acquired by simultaneous operations. In this paper, we utilize Dezert-Smarandache theory (DSmT) as an evidence fusion approach to reduce ambiguous or imperfect information then to get higher belief levels in the data fusion process of contextual information. To analyze the improvement of DSmT fusion process, we compare DSmT with Dempster-Shafer theory (DST) using PCR5 rule of combination and Dempster's rule of combination respectively.