Activity Recognition as a Service for Smart Home: Ambient Assisted Living Application via Sensing Home

Abstract

An aging population has inspired the marketing of advanced real-time ambient assisted living solutions. However, the accurate collection of behavior information, recognition of human activities and implementation of this scheme in a real living environment is a challenging task. In this paper, we propose a feasible extended framework which is compatible with common devices and protocols. Specifically, we treat activity recognition as a cloud service; thus, the application of home automation, environment and health management and advance warning of risk can be directly applied using this service. Finally, we evaluate our approach and show that it can achieve a theoretical accuracy of 93%, and can be run on cost-effective devices such as BeagleBone.