Multi-layered System Design for Classifying Activities of Daily Living Saif OKOUR, Anthony MAEDER, Jim BASILAKIS School of Computing, Engineering &Mathematics, University of Western Sydney, Australia Abstract. Our overall aim in this research is to develop a health smart home design that can monitor activities of its occupants. Our design includes three stages: system physical layers and software, fusion of multi-layered data, and classification of activities of daily living. We review prior work, discuss design and implementation issues, and describe our validation approach. Keywords. Smart home, telehealth, remote monitoring, personal location Introduction This paper gives an overview of a health smart home system we have under development and the approach taken to monitor the activities of its occupants, with two main objectives: (a) to investigate how to construct a system for smart home applications based on multi- layered wireless sensor networks, to fuse location and motion data from different sources, and determine personal motion and location by using localization techniques on the data. (b) to utilise the above inputs to perform classification of major activities of daily living (ADLs), and variations to a subject’s performance to detect whether the activity is normal. This topic addresses a growing telehealth area: remote monitoring of subjects to give carers better knowledge of health status and so to apply better healthcare management techniques. 1. Smart Homes The “smart home” concept [1] refers to a residence which is equipped with technology to facilitate monitoring of residents, and promote independence and increase residents’ quality of life. The technology is integrated into the infrastructure of the residence and does not require training of or operation by the residents. Where the smart home objectives are specifically aimed at supporting health of residents, this is termed a “health smart home”. 1.1 Health Smart Homes To date the development of health smart homes can be divided into two generations of practices based on health services: the social practice of telesurveillance for safety purposes and the medical practice of remote monitoring for medical supervision purposes [2]. Remote monitoring is used widely in the health sector to solve issues such as fall detection and prevention, chronic disease management, diminished mobility, and rehabilitation [3]. Different health smart home projects have been developed to provide solutions for particular health issues. Most of the early activity monitoring systems in health smart homes used pressure sensors to identify and monitor patterns of mobility and location. EMMA provided environmental monitor/movement alarms [4] which detected movement by using pressure mats under the carpets and a vibration detector on the bed. The first telemonitoring health smart home which measured mobility and detected the activity type was presented by [5]. It used magnetic switches which were placed on the doors, infrared