Impacts Of Diverse Devices On WiFi Fingerprint Localization

With the rapid development of high-tech and spread of intelligent terminals,the demand for accurate real-time location services is increasingly urgent.Currently mature outdoor localization technology is widely used.In a complex indoor environment,GPS localization technology,due to effects of shading and multipath,is difficult to achieve a satisfying result.While indoor localization technology require to achieve high localization accuracy to meet the requirements.Therefore,indoor localization has become a hot research for its complexity and demands of urgency.WiFi fingerprint-based localization technology is one of the most popular indoor localization methods,which has a very wide range of uses.However,hardware variance of devices always leads to WiFi received signal strength variance in measurement,greatly degrading the accuracy and precision of fingerprint-based indoor localization result.To solve the problem,we propose an unsupervised learning method based on EM algorithm to automatically figure out the transformation function between different devices' received signal strengths.The transformation function we obtain could map tracking device' fingerprints to training dataset,then decreasing variance of different devices' received signal strengths and improving localization system's accuracy.We design a specific algorithm to learn the mapping function between different devices' RSS to calibrate localization device's RSS,and apply the mapping RSSs to a specific location system.Firstly,the solution is proved to be rational theoretically.Then we verify the validity by testing different device pairs.The results show that the algorithm enables the localization error of the mean difference in equipment is reduced by more than 20%.