Automated Modeling and Reasoning Approach for Context in Smart Homes based on Stochastic Analysis of Sensory Data

Abstract

Currently, various smart home platforms have been published, aiming to assist human residents. Context-awareness computing, which enables to recognize a current context and to provide needed services accordingly, is one of the primary keys for the smart home technology. The goal of our research is to automate modeling and reasoning contexts based on stochastic analysis of sensory data. We proposed an automated approach in which sensor data were derived to context models by applying machine learning techniques and statistical analysis such as K-means clustering and Principal Components Analysis. In reasoning process, a context, which was closest to the current situation, was discovered by applying computional methods for similarity. The entire process is fully automated and shows as good results as non-automated approach does.