Design And Implementation Of Wifi-Based Localization Engine Software

The explosive growth of mobile devices makes the demand for mobile positioning and navigation is increasing, as the the "last mile" of outdoor positioning, the indoor positioning gains more and more people’s attention. But because the indoor environment is complex and indoor positioning also has more stringent accuracy requirements, so there is still not perfect indoor positioning technology to use. Therefore, experts proposed a variety of indoor positioning technology solutions, each of the technology has its applications scenarios and advantages Because WiFi-based indoor positioning technology has wide coverage, information transmission speed to achieve low cost and has become the focus of attention of the people studied.This thesis analyzes the key technologies of currently WiFi indoor positioning, anda WiFi positioning engine software has beendesigned and implemented based on this. Currently WiFi-based indoor positioning technology have two methods: positioning method based on propagation model and positioning method based onradio fingerprints matching. Positioning method based on propagation model try tofinda propagation model between AP and RSSI value. But because the effects of multipath propagation of wireless signal and theobstacles blockingin the communication process, it does not exista certain propagation model between the AP and RSSI value, so this indoor positioning method does not get good results. The positioning method based onradio fingerprints matching require expert to collect RF signal in pre-selected position, and build aradio map between the signal strength and the position(i.e. offline training). When positioning, a mobile terminal collect a real time RSSI vector, and then find the most similar fingerprint in the radio fingerprint database established in the offline phase, and position corresponding tothe fingerprint will be the estimated position (i.e. positioning online). Currently, all indoor positioning algorithms based on radiofingerprints matching basically depends on the specific RSSI value, due to the heRSS variance over time and device diversity,we have to takes much time to collect RF signals for training and the positioning accuracy still not very satisfactory. For these reasons, this thesis introduces another positioning methodbased on fingerprint matching algorithm, which is no longer dependent on the specific RSSI value, but by establishing a relationship between the AP and the AP, and use the relationship for positioning calculation, in general this relationship for different devices are relatively stable, so a good solution to this positioning algorithm is a good solution to thedevice diversityproblem, and has good robustness. Based on this algorithm, a robustindoor WiFi-based positioning engine softwarehas been designed and implemented. The experimental result proved that the positioning engine software provides consistent accuracy of3m around.