TEMPORAL MODELING OF HUMAN ACTIVITY IN SMART HOMES
Muhammad Raisul Alam, Mamun Bin Ibne Reaz, Mohd Alauddin Mohd Ali, Salina Abdul Samad · Informacije MIDEM - Journal of Microelectronics Electronic Components and Materials · 2011
Recognition of human activity is a potential challenge to design an effective smart home. The paper proposed a novel algorithm to recognize activities of daily living (ADL) of the resident. It provides analysis and mathematical modeling of temporal intervals of the event. The opposite entity states are used to extract the pattern of event sequence. Each extracted episode represents a distinct task of the resident. Result shows that, the algorithm can successfully identify 135 unique tasks of different lengths with temporal characteristics. The analysis confirms that temporal pattern follows normal distribution which can be modeled by Gaussian function.