The Research Of Indoor Positioning Based On WLAN Signal Strength Fingerprint

With the rapid increasing of indoor activities, the demand for indoor position technology and service are also busting day by day. It is one of the most convenient and practical technology solutions to apply indoor position technology based on widely deployed WLAN systems. The indoor wireless signal propagation environment is complex, which are influenced by the multipath effect,the near frequency and same frequency interference,occlusions and body absorption that have bad effects on the accuracy of positioning.Considering the advantage of the high accuracy, excellent adaptability, low cost and complexity, signal fingerprint position is the ideal method and has been widely recognized and applied in the indoor real-time location based on WLAN signal strength.There are two stages that are offline signal training and online position calculation respectively in positioning process based on signal strength fingerprint. We presented the extraction of data characteristics based on Gaussian cloud model for fingerprints signal in offline phase in order to improve the accuracy of positioning. And we improved the k-nearest neighbor matching method in online position calculation phase. We proposed a parameter estimation method based on small sample data which significantly reduces data collection using the online positioning signal acquisition system. The simulation result showed the improvement in real time positioning data acquisition compared to some other methods.First, the WLAN signal strength fingerprint positioning method was chosen after careful analysis of those advantages and disadvantages of the traditional indoor positioning technology. In order to verify the feasibility and effectiveness of the proposed experimental program by experiment methods, we set up an experimental environment.Second,it is analyzed the uncertainty of experimental data, the source of measurement errors and error handling methods. After building the Gaussian cloud model and portraying the uncertainty measurement data, we proposed a signal feature extraction method on the Gaussian cloud,fully excavates and made use of the fingerprint data of the amount of information, and added corresponding coordinate information for the fingerprint signal characteristic value. It is also provided the updation in the offline phase fingerprint data updates and the offline matches fingerprint database.Third,in online position processing stage,it’s proposed a parameter estimation method to real-time data process based on small sample data which is named Bootstrap method. It’s demonstrated that similar estimation parameters are obtained irrespective of sample size,which significantly decreases information collection in the online stage. To improve the positioning accuracy,it’s proposed the improved method of matching neighbor and it’s analyzed the similarity of eigenvalues in the two stages of localization process. Compared with the other methods, the method proposed is effective to improve positioning accuracy.Finally, by simulation, measurement and comparison under different methods, it is proved the positioning accuracy and effectiveness of indoor positioning which based on technology of WLAN signal strength fingerprint.