| Location based services (LBS) are being widely applied in various fields nowadays. As the key prerequisite to realize LBS, indoor positioning techniques have become a hot research topic in both industry and academia. Among them, the WiFi based indoor positioning technology has attracted much attention due to its simplicity and high scalability. However, several limitations in deployment costs and positioning accuracy still exist in regards to this method.Therefore, on the basis of an in-depth analysis of existing positioning technologies, this paper carries out the following studies.Firstly, based on the spatial correlation of wireless signals during transmission, this paper proposes a Kriging interpolation-based method for artificially constructing location fingerprints. The proposed method significantly reduces the heavy workloads for building the fingerprint database, and further overcomes the drawback of lower positioning accuracy of boundary points. Compared with the traditional positioning technologies, the overall positioning accuracy of the proposed method improves by around 30%.Secondly, combining with pedestrian dead reckoning (PDR) based on inertial sensors embedded in smartphones, this paper implements a location algorithm based on multi-source information fusion via particle filtering. The positioning error is shown to be within 1 m.Finally, according to the proposed algorithm in this paper, we design and implement the prototype of a WiFi based indoor positioning system, and perform extensive experiments in a real indoor environment. The experimental results show that the presented localization algorithms evidently improve the positioning accuracy, while reduce labor costs in the process of system deployment.In conclusion, the multi-source information fusion algorithm based on Kriging interpolation and particle filtering is of great significance in both theories and applications. |
PDF Full Text Request