Design And Implementation Of An Accurate Wi-Fi Positioning System On Android Platform

Presently GPS (Global Positioning System) has basically met the needs of people for outdoorpositioning, while in the indoor environment, because of obstacles blocking, indoor positioningcannot rely on GPS. There exist several indoor positioning technologies: A-GPS technology, sensorpositioning, ultrasonic positioning technology, UWB technology, and Wi-Fi technology and so on.Because Wi-Fi positioning technology has wide coverage, high information transmission speed,lower cost etc. it gets more and more attention.Through making use of the ubiquitous Wi-Fi signalsand widely-adopted smart phones, this thesis designed and implemented an indoor positioningsystem based on fingerprints positioning algorithm to improve the indoor positioning accuracy.This thesis first to the existing various existing wireless locating method and indoor positioningtechnologies are introduced and compared. Especially, in comparison with Wi-Fi base schemes, theweak points of Bluetooth and RFID based indoor positioning schemes are summarized, which servethe main motivation of this thesis.Second, Wi-Fi signal propagation environment and characteristics of Wi-Fi fingerprints aredeeply investigated, in which the impact of traffic of human walking is demonstrated. Andmoreover, Wi-Fi propagation model and Wi-Fi fingerprint based indoor positioning technologies areanalyzed, which explicitly illustrates the advantage of Wi-Fi fingerprint technology.Then, the prototype of Wi-Fi fingerprint based indoor positioning is designed and implementedon Android platform, which adopts the architecture of single client. Specifically, the prototype iscomposed of two processes: offline gathering of Wi-Fi fingerprints, and online localization. Firstly,the offline maps and fingerprint database are designed. And then, through comparing the detectedWi-Fi signals of being localized point with fingerprint database, KNN algorithm is used to obtainthe specific position. Finally, select a number of scenarios, to achieve the above fingerprint Wi-Fipositioning system on the Android platform.Finally, considering the high fluctuation of Wi-Fi signal in indoor environment, the traditionalKNN clustering algorithm used in the prototype is simply improved. In detail, in the renewedprototype, the average mean of localization results in the latest three times is intentionally used asthe final location, and moreover, the weights are those three localization results are heuristicallychosen. The experimental results show that the localization accuracy in the renewed prototype canbe improved in comparison with the original prototype.