Performance Analysis And Optimization Algorithm Of Fingerprint Localization Technology

In the era of Internet of things,location-based service is one of the basic functions with the highest utilization rate and the widest coverage in the Internet.Limited by the propagation conditions of satellite signals,satellite positioning system cannot meet the basic requirements of indoor positioning.Therefore,the research on high-precision indoor positioning system has become a research focus in the industry.Among them,due to the increasing popularity of Wi-Fi hotspots and devices equipped with Wi-Fi chips,the indoor positioning system based on wireless local area networks(WLAN)avoids the extra equipment cost and has high practical value,which has attracted extensive research.Fingerprint positioning technology is the most commonly used positioning method in WLAN positioning system,which is mainly divided into two stages:offline fingerprint collection and online positioning.This paper studies and opti-mizes the establishment of the radio map in the offline phase and the fusion posi-tioning in the online phase.The main work is as follows:In the offline stage,in order to reduce the workload of manual fingerprint collection and the storage space of the radio map,the positioning error caused by the initial positioning is introduced,and the high noise grids aggregation into quadrilateral algorithm is proposed to realize the spatial aggregation of high noise areas.On this basis,an uneven grid fingerprint localization model based on posi-tioning error is further proposed to construct radio map reasonably.The experi-mental results in simulation environment and real environment show that the pro-posed uneven grid fingerprint localization model reduces the number of reference points stored in the radio map,and effectively improves the positioning accuracy with fewer offline fingerprints.In the online phase,in order to combat the degradation of the accuracy of a single algorithm in the complex electromagnetic environment,the idea of the fu-sion-based indoor positioning is used to combine the complementarity of different positioning algorithms.Based on the Bayesian criterion and the idea of K-nearest neighbors,this paper constructs the confidence of neighborhood label and the evaluation matrix of classifier to quantify the reliability of the classifier.On this basis,a fusion localization algorithm based on the confidence of neighborhood label is proposed,which realizes dual fusion.Firstly,the evaluation matrices of all classifiers are fused to construct the confidence candidate set,and the weight proportion of elements in the confidence candidate set is calculated.Secondly,the position estimation of the online fingerprint is obtained by fusion-based indoor positioning algorithm based on the weight proportion of elements in the confi-dence candidate set.The experimental results of the public data show that the algorithm proposed in this paper improves the positioning effect,and also shows strong stability in the changes of the neighborhood number and the environment.The confidence of neighborhood label and evaluation matrix of classifier play a great guiding role in the construction of confidence candidate set.