Research And Application Of Indoor Location Based On Adaptive Fingerprint Database

With the continuous development of mobile devices such as cell phones,a variety of wireless technologies can provide accurate positioning and navigation services in the field of indoor positioning.The Bluetooth Received Signal Strength(RSS)based fingerprint library positioning technology is widely used in indoor positioning systems due to its low power consumption,low cost,high ease of use and high positioning accuracy.However,in indoor scenarios,various mobile devices are used by each user,so the signal intensity from bluetooth beacon is different,resulting in different experiences for each user of the same positioning system,which does not allow everyone to obtain accurate positioning results.At the same time,the indoor environment is not immutable.The location change of the Access Point(AP)or signal fading can cause changes in the fingerprint database and affect the positioning results.For some large scenarios,a complete database cannot be obtained by a single or several acquisitions,and the fingerprint database needs to be supplemented later.When new data is added,the traditional method needs to retrain the model,which is less adaptive.To solve the above problems,the localization method using fingerprint database is improved.The main content of the paper consists of the following parts:(1)For the problem of fingerprint library changes due to heterogeneous devices and AP changes in Bluetooth fingerprint-based indoor localization,a fingerprint library adaptive algorithm based on Procrustes Analysis and Gaussian Process Regression(GPR)was proposed.In the offline phase,the target area is divided using a K-Means clustering algorithm optimized by Pearson coefficients in order to reduce the computational effort of the localization data by narrowing the area to which the point to be measured belongs;subsequently,the received signal strength is converted into a standardized location fingerprint based on Procrustes Analysis to overcome the influence caused by heterogeneous devices;when the AP signal changes,gaussian process regression is used to recalibrate the AP signal and update the fingerprint database.Test analysis shows that the algorithm can effectively overcome the impact of heterogeneous devices and better adapt to environmental changes.(2)To address the problems of low localization accuracy of traditional fingerprint database localization methods and the need to retrain the localization model when new data are added,a fingerprint adaptive indoor location algorithm based on SSAOS-KELM is proposed.In the offline phase,the Slap Swarm Algorithm(SSA)is used to optimize the parameters of the kernel extreme learning machine to construct the initial localization model for each area;when additional data are added,the parameters of the initial localization model are adjusted using online sequential learning,and the updated model is used for real-time localization to adapt to environmental changes.The experimental results show that the algorithm can achieve higher positioning accuracy and adjust to environmental changes,and has obvious advantages in accuracy and self-adaptability compared with other traditional algorithms.(3)The paper designs and implements a mobile-based indoor positioning system based on BLE RSS fingerprint library technology.The software and hardware of Bluetooth beacon and Bluetooth gateway are designed to realize the functions of signal broadcasting and fingerprint library calibration respectively.A mobile APP was designed and written based on the We Chat applet platform for collecting offline fingerprints,uploading them to the server and completing the construction of the positioning model.In the online phase,the mobile APP is used to collect data in real time and communicate with the server to realize the online positioning and navigation functions.