Research On Indoor Positioning Technology Based On The Fusion Of Particle Filter And Electronic Map

WiFi signal fingerprint is often used to infer the location coordinates of mobile devices in the room. However, the complexity of the indoor environment leads to great difficulty of this speculation. Firstly, it is difficult to establish a precise relationship between the WiFi signal fingerprint and the location of the mobile device. Secondly, continuous navigation and positioning service, without considering the relationship of the two adjacent timing positioning results, often appear distant mutations phenomenon. Again, a lot of positioning systems, which were considered the timing factor, still have a not correct situation of two continuous positioning results through walls.First, we propose the use of spatial clustering algorithm divides the large indoor space area into a smaller and more obvious characteristics of the sub-region, combined with a two-stage algorithm modeling the relationship between WiFi fingerprint and position coordinates used SVM classification and regression algorithm. Thereby more accurately estimate the specific position coordinates of the mobile device.Then, we propose using the particle filtering fusion the above-mentioned WiFi model positioning method of stable positioning effect and the timing related PDR navigation method. The PDR navigation and positioning result will be as the transfer amount of particle filter algorithm, a result of the WiFi model positioning will be as an observer status of particle filter. Thereby combining the advantages of both localization algorithms, weakened weakness of positioning of each individual.Finally, this paper uses indoor map-matching techniques to constrain and correct positioning results on the problem of crossing the wall at fusion location algorithm. First we use points, lines and shapes to establish an indoor vector map model for the indoor environment. On this basis, from two aspects to constraint and correct the positioning results of particle filter fusion indoor positioning system, one is constraint the particles passing through the wall in particle filter algorithm, second is to adjust the through wall localization result to an adjacent door.The results show, in solitary point positioning, the proposed two-stage WiFi indoor positioning of space-based division, its classification accuracy is up to 94.58%, the average localization error distance is 2.58 meters. By comparison with various positioning methods, the paper has obvious advantages in terms of classification accuracy, positioning accuracy and response time. In terms of navigation and positioning, the proposed indoor navigation and positioning system model based on particle filter fusion WiFi positioning technology, PDR technology and map-matching technology, has a very distinct advantage in the complex polyline navigation walking track test, compared to a separate location subsystem. The average positioning error distance is up to 1.81 m, it is very close with real walking track. In a simple linear navigation walking track test, the average positioning error distance even becomes as low as less than 1 meter. This proposed navigation and positioning system greatly improves the indoor positioning accuracy, and enhance the usefulness of the positioning system.