A Study On Indoor Localization Of Passive Tags Based On Multiple Antennas Fingerprinting

In recent years,intelligent equipment is gradually diffusing across all the scenarios.Correspondingly,the precise location in the indoor environment is essential and has become a rising concern in science and engineering.Radio Frequency Identification(RFID)is a key technology for supporting the Internet of Things,which is capable to acquire the identity and location information of objects.RFID-based indoor localization has become a research hotspot for scholars.Fingerprint-based method is one of the indoor localization techniques,which has the advantages of not being affected by multipath effect and no additional hardware cost.However,the existing RFID localization technique based on fingerprint relies on the single fingerprint libraries and the global reference points.It is hard to collect and maintain a huge and redundant dataset,though which is essential to ensure location accuracy.Besides,the environmental influences affect the localization accuracy.In this dissertation,an RFID localization method based on orientation priority multi-information fingerprints was proposed.First,orientation priority reference tags were selected by beam scanning,which reduce the collection of fingerprint database without reducing the localization accuracy.Then,a multiple fingerprint database is built which is the fusion of signal covariance matrix,signal subspace,and received signal strength of the above preferred tags in the multifrequency multipath environment.Obviously,the fusion dataset can describe the given scenarios from a higher dimension and provide higher stability under the influences of environment changes.Finally,the proposed method transforms the location problem into a pattern recognition problem.It means that multiple strong classifiers can be designed to train different types of fingerprints on the basis of the integrated learning algorithm.Subsequently,a posterior weight estimation localization algorithm was proposed to combine the predictions of different classifiers and samples.Both simulation and real data verify that the localization performance of OPMIF is superior to the traditional single fingerprint-based methods.The simulation experiment data proves that the localization error of the proposed method is less than 1.2 m with the probability of 80%.