Ambiguity Elimination And Real-time Generation Of The Radio Map

Real-time generation of wireless fingerprint map is the key to ensure the reliability and scalability of location services.However,the existing method of constructing wireless fingerprint map offline by manual survey has the problems of heavy labor cost and poor scalability.By using group intelligence,the cost of manual survey can be greatly reduced and the system expansibility can be improved.However,the existing group intelligence perception mode is restricted by the diversity of equipment and the accuracy of commercial equipment,and there are serious problems of trajectory ambiguity and wireless fingerprint ambiguity,which seriously restrict the extensive deployment of the system.To solve these problems,this paper studies the ambiguity elimination and real-time generation of wireless maps,and designs a low-loss trajectory and radio map generation method based on particle filter and trajectory fusion.This method does not need to set any anchor points,obtain the initial position of the user,or require specific hardware support,so it has strong real-time performance and scalability.The main research content of this paper is as follows.Firstly,an enhanced particle filter method based on pedestrian dead reckoning is proposed.The enhanced particle filter can obtain the trajectory of pedestrians by using inertial navigation data and map information.Meanwhile,it uses the uniqueness of trajectory in the map to correct the step size and direction of the pedestrian trajectory.Then this paper proposes and discusses the trajectory ambiguity caused by the difference of trajectory characteristics(i.e.,length,shape,etc.)and the similarity of spatial structures.Secondly,a method of trajectory cluster splicing based on spatial and temporal logical correlation of indoor trajectories is proposed to solve trajectory ambiguity.By using the specific constraints of each trajectory cluster,the joint constraints among trajectory clusters are established and the stronger joint constraints are used to eliminate the trajectory ambiguity.After determining the unique position of the trajectory,the trajectories are spliced to realize the integration of all the trajectories.Finally,the trajectory smoothing and fingerprint fusion algorithm based on interior plane partition are proposed.The divide and conquer algorithm are employed to smooth the trajectory and correct the fingerprint point coordinates.In order to improve the accuracy of RF-fingerprint map,the using frequency is used to filter the fingerprint data.