A survey of parametric fingerprint-positioning methods

Abstract

The term fingerprint-based (FP) positioning includes a wide variety of methods for determining a receiver’s position using a database of radio signal strength measurements that were collected earlier at known locations. Nonparametric methods such as the weighted k-nearest neighbor (WKNN) method are infeasible for large-scale mobile device services because of the large data storage and transmission requirements. In this work we present an overview of parametric FP methods that use model-based representations of the survey data. We look at three different groups of parametric methods: methods that use coverage areas, methods that use path loss models, and methods that use Gaussian mixtures. Within each group we study different approaches and discuss their pros and cons. Furthermore, we test the positioning performance of several of the analyzed approaches in different scenarios using real-world WLAN indoor data and compare the results to those of the WKNN method.