Automatic radio map creation in a fingerprinting‐based BLE/UWB localisation system

Abstract

One of the most popular localisation algorithms used in radio localisation systems is fingerprinting. The method has significant advantages: its implementation is easy and since it does not require any information on system infrastructure placement, it can be implemented in already deployed networks. However, it has a significant drawback, which is a high effort needed to create the radio map used for localisation. This study describes a low-effort method for automatic radio map creation intended for use in hybrid Bluetooth low energy/ultra-wideband (BLE/UWB) positioning systems. The proposed method adopts fingerprint crowdsourcing and radio map interpolation approaches. It consists in collecting signatures comprising user location calculated using an UWB subsystem and measured BLE signal levels. The recorded data is used to identify parameters of a log-distance path loss model and a Gaussian process regressor and interpolate a complete radio map. The presented concept was experimentally tested. The results have shown that the method allows for more accurate localisation than conventional received signal strength ranging based solution.