Research And Implementation Of Low-power Bluetooth Indoor Positioning Based On Location Fingerprinting Technology

With the richness of social production and life,people have diversified their demand for location-based services.Traditional global satellite navigation systems are difficult to meet the high accuracy requirements of indoor positioning due to the complexity of the indoor environment,and along with the continuous maturation of wireless communication technology,indoor positioning technology has entered a period of rapid development.Among them,lowpower Bluetooth indoor positioning technology has been widely researched and applied by researchers and scholars for its advantages of low power consumption,low cost and easy deployment.In this paper,we study the basic principles of indoor positioning and related positioning techniques,focusing on the relevant positioning algorithms of location fingerprinting technology,and carry out optimisation and improvement in the offline library building stage and online positioning stage of location fingerprinting.In the offline database building phase,the signal outlier rejection,multi-directional mean filtering and Kalman filtering are adopted to establish the location fingerprint database in order to address the problems of complex and variable indoor environment and the influence of human obstruction on the collected signals,which lead to the low quality of database building.In the online positioning stage,in order to improve the positioning accuracy and ensure the real-time positioning,the selection of the initial classification centre points and the calculation of the classification edge points of the K-means clustering algorithm are optimised and improved,and the K points with the least similar characteristics are selected as the initial class cluster centres by setting a threshold value to avoid the local optimal solution caused by the random initial class cluster centres.The algorithm is improved based on the discrepancy between the Euclidean distance and the actual physical distance,and the Adaptive Distance Weighted K-Nearest Neighbour(ADWKNN)localisation algorithm is proposed.The algorithm introduces both the Euclidean distance and the Manhattan distance,and each point to be located adaptively selects the distance estimation method by comparing the standard deviation of the two distances,and sets a threshold to achieve dynamic changes in the K value.To verify the performance of the improved algorithm,an iBeacon-based indoor positioning system is built to implement the basic positioning function,and the four algorithms in the paper as well as the ADWKNN algorithm are compared and experimented by MATLAB simulation.The simulation experimental results show that the average positioning accuracy of the weighted K-nearest neighbour improvement algorithm with adaptive distance proposed in this paper improves by 16.2%-30.7%compared with the remaining four positioning algorithms in the paper,and meets the real-time requirements of positioning,verifying the effectiveness and feasibility of the overall improvement algorithm in this paper.