Developing an Improved Fingerprint Positioning Radio Map using the K-Means Clustering Algorithm

Abstract

Recently, with the development of Wi-Fi technology and the increase of mobile devices, location-based services that provide user location have drawn much attention. One of the most utilized methods for an indoor-based location acquisition system is the fingerprinting matching method, which estimates the user's location by analyzing the strength of the Wi-Fi signal. This system, however, suffers from the RSS variance problem in which the signal strength is unstable due to environmental factors. Therefore, it is crucial to collect stable sample records, which can be achieved by collecting signal samples over a sufficient period and averaging them. However, this is not the most suitable solution since signal strengths tend to be reliant on the device used and the time that it was measured. Eventually, sampled signals tend to form groups of clusters with respect to their obtained attributes. In this paper, we propose a more accurate radio map-generating algorithm by finding out the optimal number of clusters and applying the K-means clustering algorithm. This process generates a more precise radio map than the average sampling model.