A pedestrian tracking algorithm using grid-based indoor model

Pedestrian tracking is widely required by location-based services, e.g. indoor navigation, mobile advertising, and guidance for emergency response, etc. But indoor localization and tracking are still challenging due to the complexity of indoor environments and low positioning accuracy or/and precision. This paper presents an indoor pedestrian tracking approach that utilizes indoor environment constraints in the form of a grid-based indoor model to improve the localization of a WiFi-based system. The indoor space is subdivided into grid cells with a specific size and corresponding semantics. The algorithm recursively computes the location probability over these cells based on the indoor model and magnetometer measurements on a mobile phone. Our experiments prove that the proposed tracking approach can compensate for tracking errors such as improper locations, wrong headings and jumps between consequent locations, which significantly enhance the tracking performance.