An Evidential Fusion Network based context reasoning for smart media service

For effective smart media service, a reliable and confidential context recognition is required to prepare and react properly. However, it is difficult to achieve a higher confidence level for several reasons. First, raw data from multiple sensors have different degrees of uncertainty. Second, generated contexts can indicate conflicting results, even though they are acquired by simultaneous operations. In this paper, we demonstrate an Evidential Fusion Network (EFN) based context reasoning for smart media service. For this we conduct the context classification and state-space based context modelling. Then, we perform the static evidential fusion process (SEFP) to obtain a higher confidence level of contextual information. It processes sensor data with an evidential form based on the Dezert-smarandache theory (DSmT). The execution with proposed example scenario demonstrates that the DSmT approach based on PCR5 rule performs better than the DST approach based on Dempster's rule.