Optimization And Realization Of Indoor Localization Technology Based On WLAN Fingerprint

The increasing demand for location based services of indoor environment promotes the indoor localization technology receives much attention.In many studies,WLAN fingerprint technology is becoming popular,due to the simple implementation and no need to deploy additional equipment.However,the WLAN signals are always affected by reflection,scattering and shadowing from interference of obstacle in indoor environment.Furthermore,the database size and construction cost also deteriorate the system performance.In this dissertation,we present a system based on dynamic sensing strategy to optimize the indoor localization technology.The main parts of this system are a three-axis accelerometer of mobile device and WLAN hotspots environment.The main content of this dissertation is described as follows:Firstly,in view of unstable WLAN signal strength,this paper evaluates the impact of interference factors by simulation experiment,and adopts the improved Kalman filter algorithm to preprocess the fingerprint due to traditional filter cannot restrain quickly when signals jump suddenly.Secondly,in order to solve cost problem caused by over site investigation and low performance problem from large scale of database in offline stage,a quick method is proposed to build slight database efficiently,which only allows same kind of WLAN router to be deployed in the environment and records distance between routers and reference points rather than large coordinates of deployed reference points.Therefore,this method could avoid interference of equipment differences and reduces scale of database in the meantime.Thirdly,in consideration of the low accuracy in online stage,this paper proposes a dynamic sensing strategy to calibrate coordinates.For one thing,the strategy collects three-axis accelerometer readings in different motion states,by which a Bayesian model could be established in offline stage.For another,if the model determines that user is static,it records current coordinate is same as the last one.Once the model determines that the user moves,it would use K-NN algorithm to get coordinate and calculate the distance between last coordinate and current one.More importantly,the strategy should adjust current coordinate according to distance threshold and step information which is updated dynamically.Fourthly,for the proposed optimization technique based on WLAN fingerprint in indoor localization system,we design and implement a prototype system on Android platform.The experiment result shows its average position error is 0.7 meters which could meet the needs of users,reduce training cost and have real-time response speed.All these features prove that this work is significant and easy to promote.