Research On Indoor Positioning Technology Based On BLE/PDR Fusion

In recent years,the demand for Location Based Services(LBS)has grown rapidly,and positioning is the basis of LBS.At present,the Global Navigation Satellite System(GNSS)can achieve high positioning accuracy in outdoor environments.However,in the indoor environment,due to the occlusion of buildings,the satellite signal is weak after reaching the room,which cannot meet the positioning requirements in the indoor environment.Scholars at home and abroad have proposed a variety of indoor positioning technologies.However,due to the shortcomings of single positioning technology,it is difficult to achieve high-precision positioning in complex indoor environments.Therefore,exploring multi-source technology integration has become a research trend.In order to achieve high-precision and high-robust indoor positioning,this paper studies Bluetooth positioning technology and Pedestrian Dead Reckoning(PDR)technology by using Bluetooth module and inertial sensor integrated in smart phone,and uses Extended Kalman Filter(EKF)for data fusion.The positioning accuracy and stability are better than the fusion positioning system of independent technology.The main research contents of this paper are as follows:(1)By collecting Bluetooth signals,the temporal and spatial propagation characteristics of Bluetooth signals are studied,and their effects on positioning performance are analyzed.Then,the two stages of Bluetooth fingerprint positioning are optimized.In the offline fingerprint database construction stage,multiple signals are collected at the same position,and the outliers in the Bluetooth signal are eliminated by the isolated forest algorithm.Then,the mean value of the Bluetooth signal is taken as the fingerprint and inserted into the fingerprint database to construct an accurate fingerprint database.In the online matching stage,AdaBoost algorithm is used to improve the performance of KNN for the problem of weak learning ability of traditional KNN algorithm.Experiments show that the proposed algorithm reduces the average positioning error by more than 10%.(2)The key steps of PDR are studied and optimized.Aiming at the problem of poor adaptability of traditional peak detection method,the sliding window is used to extract the acceleration,and the motion state of pedestrians is judged by analyzing the acceleration characteristics in the window,and then the threshold of the peak detection algorithm is dynamically adjusted.The improved peak detection algorithm improves the accuracy of gait detection in running and mixed states by more than 4%.Then several commonly used step size models are evaluated experimentally.Finally,aiming at the problems of multi-noise and low accuracy of consumer sensors,combined with the short-term accuracy of gyroscopes and the stability of accelerometers and magnetometers,complementary filtering is used for multisensor fusion to achieve heading estimation optimization.(3)Aiming at the instability of Bluetooth positioning results and the cumulative error of PDR,EKF is used to fuse the Bluetooth positioning results with PDR,and the system equation of BLE/PDR fusion positioning is constructed.In addition,a distance constraint is added to correct the result of Bluetooth positioning for the problem of location jump and different frequencies of Bluetooth positioning and PDR positioning.In the experimental environment,compared with Bluetooth positioning,the average positioning error of the proposed fusion method is reduced from 2.09 meters to 1.28 meters,which is reduced by 39%.The maximum positioning error is reduced from 6.5 meters to 3.5 meters.The positioning error of the fusion positioning method is stable,and there is no cumulative error problem of PDR.The positioning accuracy and robustness are greatly improved compared with the single positioning technology.