Study Of Indoor Localization Algorithm Based On Multi-source Information Fusion

In the field of location-based services,the coverage rate of global navigation satellite system(GNSS)has exceeded 98%.In outdoor conditions,meter-level positioning accuracy can satisfy the needs of people's positioning applications.However,in indoor conditions,the satellite signal is weak,and the environment is complex,unpredictable,and noisy,which makes it difficult for satellite-based positioning system to maintain good positioning performance.Although there are many research results about indoor positioning,the disadvantage of GNSS in indoor positioning has not yet been completely resolved.Therefore,it is necessary to develop a universal,cost-effective,and high-precision indoor positioning technology.In view of the deficiency of relying on a single information source for indoor positioning,this paper realizes an indoor positioning algorithm through multi-source information fusion,which fuses Wi-Fi signal,heterogeneous sensors,and map information,under the support of Wi-Fi fingerprint positioning,pedestrian dead reckoning(PDR),and map mapping.The main contents of this paper are as follows.(1)Wi-Fi fingerprint-based localization: a coarse query algorithm based on received signal strength indicator(RSSI)is proposed to reduce the time complexity of Wi-Fi fingerprint matching.By using Bayesian algorithm for online fingerprint matching,the positioning accuracy is up to 81% within 3 m of level walking.For convenience,a software based on Android smart phone is designed and implemented for Wi-Fi fingerprint-based localization.(2)Pedestrian dead reckoning: an improved six-position correction algorithm of accelerometer is proposed to ensure that the six-position algorithm is still reliable without professional calibration instruments.The problems existing of heading estimation when fusing magnetometer are analyzed,such as the inaccurate initial heading,the difficulty in determining the magnetic declination,the regional disturbance of indoor magnetic field,and the deviation of heading estimation caused by the overall indoor magnetic field deviation.An adaptive step detection algorithm is proposed based on the three-axis acceleration characteristics of pedestrian.Besides,on the basis of peak detection,the accuracy of step detection is further improved by dynamic time warping.(3)Multi-source data fusion-based localization: Based on particle filter,the fusion algorithm of Wi-Fi fingerprint location,PDR and map information is realized,which solves the problem of insufficient positioning accuracy of Wi-Fi fingerprint and cumulative error of PDR.With the support of map information,through the wall-crossing detection and heading recorrection,the algorithm still achieves good positioning results under the conditions of inaccurate initial position and heading angle,unknown magnetic declination,and difficult to guarantee the accuracy of heading angle,which achieves a more universal,cheap and reliable indoor positioning algorithm.