Design protocol and performance analysis of indoor fingerprinting positioning systems

Abstract

Location fingerprinting is a technique widely suggested for indoor positioning. Given specific positioning requirements, this paper provides methods for setting up the network elements such that those requirements can be met by the location fingerprinting method. In particular, the paper aims to optimize indoor fingerprinting systems such that the positioning performance gets close to the optimal performance indicated by the lower bound of the system. The Weiss–Weinstein bound (WWB) and Extended Ziv–Zakai bound (EZZB) are suggested for indoor environments, as they are shown to have superior predictive performance for this application. The effects of the number and geometry of access points (APs), the number and spatial arrangement of reference points (RPs), and the number of signal strength samples taken per location are presented, both through simulations and analytical lower bound estimates. The impact of the path-loss exponent, the standard deviation of the signal strength measurement, and size of the operating area are also investigated. These theoretical/simulation estimates are also assessed using experimental data. By utilizing these tools, a system designer is able to set appropriate parameters to optimize the compromise between positional accuracy and the costs associated with the setting up of the fingerprinting measurements database.