Bluetooth low energy indoor localization for large industrial areas and limited infrastructure

Abstract

Systems using Bluetooth Low Energy (BLE) communication are widely used in various industrial applications, which benefit from the efficient low-power operation and pervasive availability of BLE transceivers in a large number of devices and sensors. It is no different in Smart Factory and Smart Farming settings, where BLE systems are already used for asset monitoring, management, tracking, and localization. The existing BLE-based localization systems aim at high accuracy and precision using radio propagation models and multilateration, or radio fingerprinting. Both methods use the received signal strength indicator (RSSI) measurements and the dependency of RSSI on the distance between the transmitter and the receiver. Because RSSI measurements are highly inaccurate and susceptible to radio propagation phenomena, high localization accuracy is only possible if the anchor devices are densely deployed, and the propagation conditions are stable. Unfortunately, these requirements do not hold in industrial systems where radio propagation is complex, the number of anchors is limited, and propagation conditions change because the environment is dynamic. Therefore, localization methods for industrial applications have to balance system complexity (e.g., anchor density, energy cost) and localization accuracy so that practical benefits can be attained. This paper presents a set of localization algorithms that require limited infrastructure, have low complexity, and can provide valuable location information at low costs. The proposed algorithms were verified in a real-life Smart Farming application where BLE is used to monitor well-being of farm animals. The animal localization was integrated with the already existing system and proved to operate reliably despite system-level constraints and varying propagation conditions — interference, signal attenuation, and unavailability of raw radio signal measurements. The proposed approaches allow localizing animals in the cowshed of 1600m2, using only 10 anchors with an average positioning error below 8m. The algorithms are based on signal strength measurement, so they can be applied to other radio technologies.