Research And Implementation Of WiFi Indoor Location Based On Location Fingerprint

With the development of mobile smart devices and wireless techniques,location based services(LBS)become more and more popular in our daily lives.In typical LBS scenarios such as global navigation,military search and rescue,position information plays the main role.The global position system(GPS)is mostly applied in open areas.However,due to the blockage by various obstacles,the satellite signals may be severe attenuated,or totally blocked in complicated environments,such as dense canyons and indoor environments.Therefore,wireless network localization becomes an alternating solution in such scenarios.Recently,the WiFi techniques have been developed rapidly,for its obvious advantage of wide coverage,high convenience,low cost and complexity.In this dissertation,we mainly try to perform optimization and improvements of WiFi localization.The main works and contributions can be concluded as follows.First,we analyze the statistic distribution of the received signal strength indication(RSSI)in the reference points.The disadvantages of the existing fingerprint database based on RSSI means are illustrated.We then present a new scheme of constructing the position fingerprint database based on the RSSI of both time series and direction.Second,we show that the starting point selection in the k-means clustering is important,or a local optimum can be obtained.Therefore,we present an affinity propagation method,to handle the problem that the Euclidean distance of RSSI between the anchor and target points deviates from the real distance.A double clustering algorithm is proposed to filter the reference points that affect the accuracy of location.Due to the time and complexity consuming during the offline database formulation,we try to reduce the system complexity using the widely applied compressed sensing(CS)method.The data collected in a small amount of reference point to restore the integrity of the fingerprint database.Third,based on the theoretical algorithms presented above,we try to develop a WiFi-based indoor localization system accordingly.The localization system is implemented on a platform of Android,which is a popular operation system on mobile devices.Both data collection client and localization server are designed and implemented.Experimental results show that the achieved localization performance is able to meet most indoor LBS applications,with the advantage that no additional hardware is required.