Research On WLAN Based Indoor Location Estimation Technology

With the popularization of mobile Internet, people upgrade their demands and expectations in location-based services. They hope the location-based services are confined to outdoor navigation, but also can be used in indoor places such as shopping malls, supermarkets, airports and other complex scenes. At the same time, with mobile terminals and wireless LAN rapid development, for the penetration of smartphones and the coverage of WiFi to meet the requirements for the indoor positioning, the indoor positioning based on wireless LAN technology has become a hot point in the field of location-based services.Aiming at solving the problems of lack of accuracy and stability in current indoor positioning based on wireless LAN technology,the thesis intensively studied the indoor positioning based on wireless LAN at first, then transferred the way of positioning to continuous positioning rather than discrete positioning, and added historical information to the positioning algorithm. Finally some experiments were made, and the experimental analysis proved that the algorithm can work feasibility and effectively as we supposed. The specific topics are follows:First, researching the key factors in indoor positioning provides us with a theoretical basis for the indoor positioning algorithm. The first factor is RSS features including the relationship between RSS and the distance, the interference of human or other factors, and the probability distribution of RSS. The second kind of factors are the factors of the error influences.These factors should be studied because the number of AP and the separation distances of finger points are very important to the positioning algorithm, but there is no standard number of APs, nor standard distance between the finger points. So this thesis will find these standards through experiments.Second, to address the problems of lack of accuracy and stability, two modules of the positioning algorithm have been changed. The first module is data preprocessing.A kind of smoothing filter has been used in this module to take advantage of RSS signals collected at off-line phase, and this change could improve the positioning accuracy. The second module is the precise positioning. Considering the position of human can't change suddenly,the user's historical locations can be used at the real-time positioning phrase, and a weighting algorithm has been designed.Thisalgorithm will weight every potential position with user's historical location information.At the same time, if the probability of last location is lower than a certain value, the algorithm will discard the historical location information to prevent the accumulation of errors. Experiments show that, compared to the traditional fingerprint positioning method, this method we proposed has significantly improved the accuracy and stability. Compared to another method that used historical location information,this method we proposed effectively avoid the accumulation of positioning errors.Third, an indoor positioning system based on Android platform has been designed and implemented. This system adopts the server-based structure, and uses the algorithm that we just proposed. By testing in the actual environment, the positioning system works well, and has excellent accuracy and stability. This system will work as a good beginning for further research in the future of indoor positioning based on wireless LAN.