| Summary: | The growing interest in location-based services (LBS), due to the demand for its application in personal navigation, billing and information enquiries, has expedited the research development for indoor positioning techniques. The widely used global positioning system (GPS) is a proven technology for positioning, navigation, but it performs poorly indoors. Hence, researchers seek alternative solutions, including the concept of signal of opportunity (SoOP) for indoor positioning. This research planned to use cheap solutions by utilizing available communication system infrastructure without the need to deploy new transmitters or beacons for positioning purposes. Wi-Fi fingerprinting has been identified for potential indoor positioning due to its availability in most buildings. In unplanned building conditions where the available number of APs is limited and the locations of APs are predesignated, certain positioning algorithms do not perform well consistently. In addition, there are several other factors that influence positioning accuracy, such as different path movements of users and different Wi-Fi chipset manufacturers. To overcome these challenges, many techniques have been proposed, such as collaborative positioning techniques, data fusion of radio-based positioning and mobile-based positioning that uses sensors to sense the physical movement activity of users. A few researchers have proposed combining radio-based positioning with vision-based positioning while utilizing image sensors. This work proposed integrated layers of positioning techniques, which is based on enhanced deterministic method; Bayesian estimation and Kalman filter utilising dynamic localisation region. Here, accumulated accuracy is proposed with distribution of error location by estimation at each test point on path movement. The error distribution and accumulated accuracy have been presented in graphs and tables for each result. The proposed algorithm has been enhanced by location based calibration with additional QR calibration. It allows not only correction of the actual position but the control of the errors from being accumulated by utilizing the Bayesian technique and dynamic localisation region. The position of calibration point is determined by analysing the error distribution region. In the last part, modification on Kalman filter step for calibration algorithm did further improve the location error compared to other deterministic algorithms with calibration point. The CDF plots have shown all developed techniques that provide accuracy improvement for indoor positioning based on Wi-Fi fingerprinting and QR calibration.
|
|---|
