Localization of Elderly People using WiFi: A Comparative Study on Parametric versus Kernel-based Approaches
Résumé
Localization is an important issue in indoor environments to assist elderly people in distress. In this paper, we propose a zoning-based localization approach using the WiFi technology. The idea is to equip each person with a WiFi-compatible mobile sensor in the form of a bracelet, a watch, or a medallion. The objective becomes to localize the mobile sensor, thus determining the zone where the elderly person resides. This paper serves as a comparative study of parametric vs kernel-based approaches. In the former approach, we fit the WiFi signals of each zone into a parametric distribution, while in the latter approach, we model the WiFi signals using the kernel density estimation. Experimental results show an advantage of kernel-based approach in terms of overall localization accuracy. The parametric approach should be investigated first as it is less complex. However, if signals fail to fit to one of the parametric distributions with a defined significance value, a kernel-based approach should be considered.