Discriminating sensor activation in activity recognition within multi-occupancy environments based on nearby interaction

The number of elderly people has increased significantly in recent decades due to various factors, such as increased life expectancy and improved health services [1]. As more people require care, there is also a need for more caregivers to be involved in this field, which means higher costs and workloads for caregivers. However, while a growing number of older people prefer to stay at home for as long as possible rather than go to a care home [2], most of the population cannot afford the expense of having a carer at home. In this sense, smart environments have emerged as a solution to help elderly people to live safely, comfortably and independently in their homes, while reducing the toll on healthcare systems. The combination and use of different types of sensors to recognise Activities of Daily Living (ALD) is becoming increasingly common [3] and includes vision sensors, audio sensors, wearables, binary sensors, RFID, PIR... under sensor fusion approaches [4].