An efficient sensing approach using dynamic multi-sensor collaboration for activity recognition

This paper presents an efficient sensing approach for activity recognition using multi-sensor fusion. The main achievement of the approach is to accurately recognize the human activity with the minimum body sensor usage through the use of dynamic sensor collaboration. The Naïve Bayes Classifier is adopted as the classification engine due to not only its easy implementation but also the advantages for multi-sensor fusion. The sensor selection is based on the real-time assignment information value of each sensor node. The platform is composed of a base station and a number of sensor nodes. The base station is used to assign the real-time information value for each sensor node, and fuse the chosen sensor data.