A hybrid indoor positioning solution based on Wi-Fi, magnetic field, and inertial navigation

Abstract

Tracking a pedestrian with a smartphone accurately and in real-time remains a challenging topic in Indoor Positioning Systems. The current computing power of a smartphone is such that it can handle data-intensive applications reasonably well. Additionally, many technologies such as Wi-Fi, Bluetooth, and inertial sensors are already available so no extra hardware cost is needed. However, cost-effective ICs are prone to produce erroneous measurements, resulting in faulty localizations. For instance, Radio Frequency based applications like Wi-Fi and Bluetooth provide accuracy up to 2–3 meters but it requires a lot of effort to maintain the system due to dynamic characteristics of an indoor setting. Additionally, Wi-Fi and Bluetooth ICs in smartphones are usually slow to complete a single scan, which introduces further difficulties when designing real-time tracking applications. Another solution is to use Inertial Measurement Unit (IMU) in a smartphone but accumulative errors due to sensor measurements still remain a problem. In this paper, a hybrid solution is proposed through the exploitation of the unique characteristics of existing technologies and compensating each other's drawbacks. Wi-Fi Fingerprinting is implemented to perform localization when a position of the pedestrian is completely unknown. After narrowing down with the Wi-Fi positioning results, a particle filter, powered by Magnetic Field Fingerprints, is utilized to provide a maintainable and accurate tracking system. Feasibility of the proposed method is demonstrated in an indoor positioning case, where a smartphone device is used throughout the experiment.