Research On Fingerprint Indoor Localization Algorithm Based On Multidimensional Information Fusion

With the rapid development of Internet technology,the continuous improvement of the performance of intelligent mobile terminals and the continuous advancement of China's urbanization process,indoor location information has gradually become the most important information in various tasks,prompting the Location Based Service(LBS)research and development.Among them,the Wi-Fi-based indoor positioning system can use the existing Wi-Fi network infrastructure in the building,and has become the most popular and practical indoor localization system.And the Wi-Fi fingerprint location method based on Received Signal Strength(RSS)does not require the physical location information of the deployed Access Point(AP),and the use of fingerprint technology can alleviate the widespread wireless signals in indoor environments the existence of multipath effects.The performance of the fingerprint location method is severely affected by the heterogeneity of the devices.This heterogeneity exists between different devices due to hardware differences in RSS benchmarks.In order to improve the universality of Wi-Fi localization for heterogeneous hardware environments,this paper proposes a new fusion positioning algorithm—a hybrid fingerprint-based hybrid classification localization algorithm.The algorithm converts RSS to construct two kinds of fingerprints that can overcome heterogeneity,such as Signal Strength Difference(SSD)and Hyperbolic Location Fingerprint(HLF),uses the complementarity between the three fingerprints to construct a composite fingerprint set and uses linear discriminant analysis(LDA)to perform dimensionality reduction analysis.In the online prediction stage,a multi-classifier fusion strategy based on entropy selection is proposed.Experimental results show that the positioning algorithm proposed in this paper can overcome the heterogeneity of equipment and effectively improve the positioning accuracy.The accuracy of the fingerprint location method depends on the degree of matching between the fingerprint library and the current signal environment,and the Wi-Fi signal is easily affected by changes in the external environment and has poor stability.In order to further effectively alleviate the impact of AP's large changes and environmental dynamics,this paper proposes a fingerprint update and localization algorithm based on heading features.This algorithm can automatically complete fingerprint update when the AP is adjusted without additional offline survey.This algorithm first uses the inertial navigation sensor to calibrate the special points on the map,and proposes a basis for judging whether the AP has changed.When AP changes are detected,Gaussian Process Regression(GPR)is used to calibrate and update the fingerprint database.Experimental results show that the algorithm can adapt to changes in environmental signals and achieve high-precision localization,which is in line with the current development trend of indoor localization.