Research On Indoor Positioning Method Based On AP Point Selection And Signal Compensation

In the era of 5G big data,the development of various network applications has greatly facilitated people's lives,and the improvement of user experience is inseparable from the accurate acquisition of location information.In the outdoor environment,positioning technology based on satellite signals can achieve very high positioning accuracy,but in the indoor environment,satellite signals cannot be transmitted in line of sight and fail.Indoor positioning technology based on location fingerprints has become a widely used indoor positioning method due to its low deployment cost and positioning accuracy that can meet user needs.With the continuous development of wireless technology,the number of AP points that can be collected in indoor environments has increased dramatically.If all the AP points searched are recorded in the location fingerprint database,the dimension of the fingerprint database will increase,resulting in excessive calculations in online location estimation.Great,affect the real-time of positioning.If the unstable AP points are recorded in the location fingerprint database,it will also cause the fingerprint database to expire and reduce the accuracy of positioning.Aiming at the above problems,this thesis proposes an improved AP point selection and signal compensation method.The Fisher criterion and Relief F algorithm are introduced into the field of indoor positioning.Before establishing the location fingerprint database,select the AP points twice to determine the subset of AP points with high stability and strong location discrimination ability to establish the location fingerprint database to enhance the robustness of the fingerprint database and improve the positioning accuracy.After completing the selection of stable AP points,for AP points that still have some missing RSS values,multiple imputation methods are used to compensate for the missing signal values.The improved Relief F algorithm can effectively measure the position discrimination ability of each AP point.By selecting the AP point set with strong position discrimination ability,deleting the AP points that are useless for positioning can reduce the calculation of the positioning method on the basis of ensuring the positioning accuracy.Complexity,reduce the energy loss of positioning.This article uses Java to compile a mobile app for RSS value collection,and uses Python to preprocess the collected raw data,and then establish a location fingerprint database after selecting AP points using the method in this article.After analyzing the results of the positioning experiment in a complex multi-wall indoor environment,under the same location estimation method,using the method in this thesis to establish a location fingerprint database positioning system,the probability of an error distance within 3 meters reaches 98%.The method in this paper not only improves the positioning accuracy but also reduces the energy loss of the positioning,which has high practical value.