| The Mobile Internet has bred a huge number of valuable services, one of which is Location Based Service (LBS) that is extending from outdoor to indoor environment, which contributes to the increasing attention of WiFi based indoor localization system in recent years. Nevertheless, the disadvantages WiFi signal has restricted the further improvement of its positioning performance.This paper proposes a solution that combines WiFi and Sensor technology (hereinafter called "WiFi-Sensor") intending to obtain complementary advantages of both technologies. The integrated algorithm consists of scene positioning algorithm using WiFi fingerprints and dead-reckoning using sensors. On the one hand, by analyzing some characteristics of WiFi Received Signal Strength (RSS) and the influence they exert on the accuracy of localization, an averaging method using sliding window concept is proposed to reduce location drift. On the other hand, location trails starting at the latest average location would be predicted by dead-reckoning. The solution is of high-performance and better continuity without adding extra hardware facilities. Besides the combination solution, a map auto matching algorithm based on AP-MAP similarity, a map and database caching strategy based on timestamp and a map engine based on vector graphics are also proposed in this paper in order to optimize the user experience by solving series of problems practically. Furthermore, the paper offers detailed design of WiFi-Sensor localization system involving infrastructure deployment, software architecture and data storage structure. Lastly, the practicability and availability of the combination are verified by algorithm simulation, and the prototype system using terminal-server pattern is implemented.The combination algorithm proposed in this paper shows obvious superiority compared with the other two solutions, and the system architecture design is reasonable. Experimental results show high value and potential for many other application areas. |
PDF Full Text Request