Study On The Pedestrian Indoor Positioning Algorithms Based On WIFI And Inertial Technology

Location based service has tremendous applicable value in safety and first-aid, health care management, personnel tracking, road navigation, certain targeting and many other fields. Mature GNSS technology with high precision real-time positioning navigation in the outdoor, and as a result of building block such as indoor satellite signal attenuation sharply even does not exist, makeing indoor environment can not be real-time location via satellite system. This paper focuses on the WIFI and inertial measurement key technologies for indoor location services, mainly learning WIFI positioning technology based on a dual cluster match operation, inertial positioning technology(PDR) based on heading estimation with real-time compensation and the two technologies fusion model for indoor positioning application, the main work and contributions are summarized as follows:The fingerprint points with a triangle structure layout is designed for acquiring the WIFI signal information, which aims at long and narrow property of corridor space. The location tests show that, the design can reduce the offline sampling work of 1/2 but with a similar location accuracy. Dynamic and static positioning experiments determine a optimal clustering parameter for fngerprint clustering process with the maximum RSS. A dual cluster match method for improving the accuracy and reducing the time for cluster matching is proposed, using the addresses information between real-time and fingerprint for coarse matching to determine a scope of clusters, then using the distance difference of RSS between them to determine the only cluster.A new gait recognition method is put forward, which just need to set an amplitude threshold and a gait frequency threshold, but provide the amplitude, amplitude difference and double time differences constraints for peak-valley detection. Proved by experiments that, the method can effectively and real-time recognize the gait when people walking or running. In consideration of the deviation between real heading and the value of orientation sensor, a heading estimation method with real-time compensation is proposed, which based on a KALMAN filter with the fingerprint geometry information. The study provided that, the corrected heading can inhibit the positioning error accumulation and improve the positioning accuracy of PDR.Facing the matters of instability as rebounding or clustering that existing in WIFI positioning system, two adaptive filtering fusion schemes are realized, one based on weighted average with displacement constraints, and the other based on EKF filtering with adaptive system noise, so as to enhance reliability and improve accuracy of pedestrian positioning. Experimental data show that the second model is better than that of smooth filtering method, not only can effectively eliminate WIFI gross error and weaken the instability, but also to be a good solution to the problem of positioning error accumulation and the own correction difficulty of positioning error coused by the initial position error.