The Construction And Updating Of Multi-sensors Based Indoor Navigation Database

With the coming of smart life, the popularity of mobile intelligent terminal has driven the rise of location-based services market. Meanwhile, the background of big data highlights the importance of ubiquitous mapping. The ubiquitous mapping means the creation and use of geographic information or map for the relationship between the environment and human cognitive activities. It emphasizes the natural characteristics of information to the people, the environment and social relevance. As the ubiquitous mapping emphasizes the ubiquitous nature of time and space, indoor positioning is an important part of ubiquitous mapping. Outdoor Location based services (LBS) have been developed for decades, and the Global Navigation Satellite Systems (GNSS) nowadays is able to basically meet the application requirements. However, indoor positioning has not been effectively resolved. For the current status of consumer applications, this paper presented a navigation algorithm which combines the sensors information of WiFi, Bluetooth Low Energy (BLE), gyroscope, accelerometer and magnetometer.So far the majority of consumer-grade indoor positioning techniques (WiFi fingerprinting, magnetic matching etc.) are based on navigation databases. The quality of the navigation database directly determines the positioning accuracy and reliability. Take the case of WiFi fingerprint positioning, in order to ensure the quality of the database the surveyors need to spend a lot of time to build the database and have to repeat this process to update it, which is the main limitation of the generalization of this method. Therefore, the crowdsourcing based database updating method is the trend of development. To achieve the automatic updates of the navigation database, accurate trajectories are needed to be chosen from the massive user data. This paper proposes a quality evaluation model of the pedestrian navigation data which provides the rules of data filtering. The main novation points list below:(1) It provides a WiFi/BLE/Pedestrian Dead Reckoning (PDR)/Magnetic Matching (MM) integration indoor positioning method, which makes full use of each technique and uses the multiple quality control mechanisms to improve the accuracy, availability and robustness of the indoor positioning system.(2) The proposed smoothing algorithm combining the forward and backward PDR navigation solutions.(3) It presents an anchor points and PDR solutions based navigation database updating method, which adopts the daily data of the users to generate indoor navigation database.(4) It analyses the main factors that influence the quality of the crowdsourced navigation database:motion mode, navigation time and sensor bias. Also, it provides the corresponding solutions to reduce the impacts.(5) It introduces an quality evaluation model of the pedestrian navigation data. Based on this model the program can choose the desired trajectory automatically.