Research Of Unsupervised Indoor Localization Technology Based On Fusion Of PDR,WiFi And BLE

In recent years,with the rapid development of wireless communication technologies and smart phones,the research on Indoor Positioning System(IPS)has become a hot area.Although many teams have achieved very high positioning accuracy,there is still no one localization system that can be commercially deployed in various indoor environments.Among the indoor positioning technologies,Pedestrian Dead Reckoning(PDR)method,fingerprint localization method and localization method based on propagation model are commonly used.PDR technology can provide continuous position estimation,but it needs to provide initial position and there will be serious cumulative error problem with time;Fingerprint positioning is widely used because it can use the signals of wireless routers in the environment.However,the fingerprint method requires a lot of time and labor cost to construct the fingerprint database,and it needs regular updating of the fingerprint database to ensure accuracy.The signal propagation model method is greatly influenced by the Not Line of Sight(NLOS)in the indoor environment.The research in this paper is to fuse multi-sensor data to provide an unsupervised positioning algorithm.The research focused on the fusion algorithm of key technologies such as fingerprint database self-building,suppression of fingerprint localization drift,and suppression of PDR cumulative error.The main research content is as follows:1.Self-building and updating algorithm of fingerprint database.The traditional fingerprinting method requires a large amount of manpower and time cost in the off-line phase to collect Wi-Fi data intensively,and the localization method of the fingerprinting method is sensitive to environmental changes,and it needs to manually update the fingerprint database on a regular basis.In recent years,many researches have used the fingerprint database updating algorithm.However,this type of algorithm defaults to trusting user feedback and cannot accurately filter the erroneous data.We proposed a Wi-Fi fingerprint database auto-building and updating algorithm fusing crowdsourcing(AUAFC).During the off-line phase,we collected data at very little landmarks,between landmarks we analysed the data which gathered from user's smartphone and provide basic location service.To reduce the boundary of the landmark recognition,the AUAFC proposed the algorithm to shrink the boundary of landmark.When the server receives the user's feedback,it will use a dynamic clustering algorithm to extract the credible data and update the database or seed new landmark to the database.2.An indoor positioning algorithm that incorporates smart phone sensors and iBeacon(SuLoc).Since the fingerprint method requires a labor and time cost to build a fingerprint database,this article attempted to use the PDR method to provide a localization service.And after the fusion of landmark correction algorithms based on iBeacon,we proposed the SuLoc algorithm.Due to the serious error accumulation problem in PDR method,this paper improved the landmark correction algorithm based on iBeacon landmarks and proposed a reliability location algorithm,the algorithm effectively solves the problem of repeat correction in the threshold correction algorithm.For the problem of step size and direction estimation in PDR method,this paper integrates the particle filter algorithm to correct the step length and the direction of motion attributes on the premise of combining indoor map data.Experiments show that it can effectively correct the error caused by the sensor error and the user's holding posture.3.A multi-sensor fusion unsupervised indoor localization algorithm(UILoc).In the SuLoc algorithm,the initial position is determined based on the propagation model method.The result is affected by the fluctuation of the BLE signal and there is a large positioning error.In this paper,UILoc is proposed,which can build a fingerprint database and combine with the propagation model method and fingerprint positioning to provide more accurate positioning results.In the process of experimental analysis,this paper implements an indoor positioning system based on UILoc.The final experimental results show that the average positioning error of UILoc algorithm is only 1.11 meters,and it has better stability than SuLoc.