Summary: | 碩士 === 國立虎尾科技大學 === 電機工程研究所 === 102 === With the development of embedded systems and the advance of communication network technologies, positioning applications and creative value-added applications has on smart mobile devices was proposed continuously, such as entertainment, tools, and learning. The positioning results is not good in an indoor environment, because of the lack of GPS signal or the poor signal. Nowadays, many solutions used Wi-Fi, G-Sensor or Gyro for position. However, the G-Sensor and Gyro have the problem of cumulative errors, it is just suitable for static environments and it has to be regularly calibrated to improve the positioning accuracy. Heterogeneous smart mobile devices have different features, such as different wireless module, antenna, etc. Radio frequency is vulnerable to environmental impact. Therefore, using Wi-Fi signal strength of methods will be challenged on the issue of positioning accuracy. This thesis implements an indoor navigation system for smart mobile devices. It combines two wireless signals, i.e., Wi-Fi RSSI (master information), and Bluetooth RSSI (slave information). The Location Fingerprinting method. is design and implemented in our system. In addition, the enhanced Neighboring Weighted Positioning method is implemented to increase the position accuracy and the wide-area positioning. In order to manage the system easily and reduce the power consumption of mobile devices, the thesis designs and implements the position computing server. The position information is managed and computed centrally. After training the fingerprint database, the server will deal with position computing task with web service. The mobile devices will received the calculated location information from the server. Moreover, a navigation app using our position method is also design. In the experiments, the system using Wi-Fi and Bluetooth wireless signals with Neighboring Weighted Positioning method is better than that only using Wi-Fi wireless signal with Five-point Weighted Positioning method. The average position error is less than 5 meters and the standard deviation is less than 2 meters. In future, the system architecture and positioning technology can apply to both of indoor and outdoor navigation applications, e.g. department stores, shopping malls, museums, amusement parks, etc.
|