Short paper: Time-dependent power load disaggregation with applications to daily activity monitoring

In this paper we explore the possibility of inferring activities of daily life (ADLs) from aggregate power load signatures of people's homes, which has many applications including e-healthcare. Such power load data are available from smart meters that will be widely deployed in many countries by utilities or customers, creating an infrastructure at the forefront of the Internet of Things (IoT). The main contribution of this work is a time-dependent factorial hidden Markov model to extract behaviour related features linked with individual appliance usage. The results show that the introduced time-dependent structure can improve the performance while also provide a probability distribution related to ADLs. These results further provide a promising indication of appliance usage connotations of e-health, and a foundation for further research.