A multi-strategy Bayesian model for sensor fusion in smart environments

Sensor fusion became a powerful scheme to recognize the daily life activities in smart homes. This paper proposed a multi-strategy approach to overcome the challenges of accuracy and efficiency. We design a model to integrate k-Nearest Neighbor (kNN, k=5) technique and Bayes classifier for recognizing the activities of daily living. There are three stages of this model. The first stage is used to reduce the search space by discovering the useful regions. A Bayes classifier is utilized in the second stage to refine the degree of beliefs. The confidence values have been denoted by the output of the Bayes classifier. Finally, max rule has been applied to fuse confidence values. The proposed model has been evaluated on five different types of activities from Place Lab dataset (PLIA1). We compare our Multi-strategy approach with the Naive Bayes Classifier and get 9% higher accuracy and 186 ms faster execution time.