New Cluster Selection and Fine-grained Search for k-Means Clustering and Wi-Fi Fingerprinting

Preprint version of the paper entitled “k-Means Clustering and Wi-Fi Fingerprinting”, presented in the 2020 International Conference on Localization and GNSS (ICL-GNSS).

Wi-Fi fingerprinting is a popular technique for In- door Positioning Systems (IPSs) thanks to its low complexity and the ubiquity of WLAN infrastructures. However, this technique may present scalability issues when the reference dataset (radio map) is very large. To reduce the computational costs, k-Means Clustering has been successfully applied in the past. However, it is a general-purpose algorithm for unsupervised classification. This paper introduces three variants that apply heuristics based on radio propagation knowledge in the coarse and fine-grained searches. Due to the heterogeneity either in the IPS side (includ- ing radio map generation) and in the network infrastructure, we used an evaluation framework composed of 16 datasets. In terms of general positioning accuracy and computational costs, the best proposed k-means variant provided better general positioning accuracy and a significantly better computational cost –around 40% lower– than the original k-means.