Virtual Sensor for Behavior Pattern Identification in a Smart Home Application

This paper presents an architecture for smart home applications virtualization. The aim of this approach is to enable the separation of the physical sensor infrastructure from the services provisioned to the user, allowing development of more complex applications that can be used in different sensor networks. This also increases flexibility for services that can use data from multiple heterogenous sources in intelligent processing algorithms. In this paper a virtual sensor for identification of temporal relationship pattern in user's behavior was developed using recurrent neural networks. We studied the behavior related to energy consumption, developing an application to display graphically the analysis results. This can increase awareness of the occupant regarding his artificial lighting control habits.