Research On Indoor Localization Technology Based On Lightweight Radio Map Construction

The rapid popularization of smartphone and the broad-spectrum demand of indoor location service lead to the fine-grained location information are in urgent need,especially in a complex indoor environment.In general,the fingerprint-based positioning system needs to construct radio map offline in different indoor scenes,and the corresponding matching algorithm is used to complete the real-time localization.To heighten the localization accuracy,increasing the density of reference points(RPs)and the number of access points(APs)is the main prescription.However,there are some side-effects in cost,scalability,the real-time performance and localization accuracy.Therefore,achieving low-cost and scalable indoor fine-grained localization is a challenging problem.In this paper,the fingerprint database is constructed by lightweight pattern.Furthermore,the SVM classification algorithm and the contour clustering algorithm are used to obtain the fine-grained localization.In summary,the major research contents and contributions are as follows:Firstly,the circular-based radio map construction method is proposed for combining with the distance measurement in indoor environments.This paper compares the merits and drawbacks of spatial partitioning between the grid-based and the circular-based.Moreover,the partitioning diversity is analyzed.Secondly,the indoor subarea localization algorithm is proposed.Due to multipath effect,the RSSI collected by different devices at the same RP might be changed dynamically over time.Employing the cosine similarity to calibrate the equipment for different devices,the scalability of the actual deployment is improved.Taking advantage of relieving the cost of large-scale location search,the support vector machine(SVM)is used to classify the indoor subareas.Thirdly,a fine-grained indoor localization algorithm is proposed.In order to further improve the positioning accuracy,the contour clustering algorithm is exploited to increase localization accuracy and the real-time performance.Finaly,the real RF fingerprint data is used for experiment evaluation and analysis,the results show that the proposed method has better performance than the traditional method.