Research On Indoor Location Fingerprint Location Algorithm Based On Fuzzy C-means

With the rapid development of communication and network technologies,precise location services have been widely used in both military and civilian fields.At present,outdoor positioning technology is relatively mature,but indoor positioning technology is difficult to achieve accurate positioning due to factors such as the complex and changeable indoor environment.Therefore,accurate indoor positioning algorithms have become a hot research topic,and indoor positioning technologies still face many problems to be solved in order to adapt to diverse indoor environments.This paper focuses on the problems of data noise,large amount of calculation in the online stage,and low positioning accuracy in the process of indoor location fingerprint positioning.This paper studies the clustering area division method based on fuzzy C-means and the weighted KNN online position matching method with dynamic k An improved indoor location fingerprint positioning algorithm FDWKNN is proposed.The main work of the thesis is as follows:(1)Improve the noise reduction method,and use a combination of overall and local noise reduction methods to perform a more comprehensive noise reduction process on the location fingerprint data in the database and extract the location fingerprints with characterization capabilities.By setting the threshold of AP access point scan times,the data noise generated by outliers is eliminated as a whole;Gaussian filtering is used to filter out random noise locally,and finally the average value is used as the position fingerprint of the reference point.(2)Use the optimized fuzzy C-means algorithm to cluster the location fingerprints in the offline database and divide the fingerprint clusters,so as to realize the positioning sub-region division.The BWP index is introduced to judge the clustering result of the KNN algorithm,and the optimal classification number and the clustering center corresponding to the optimal classification number are selected as the classification number and the initial clustering center of the fuzzy C-means algorithm to reduce the iterative calculation and complete Optimization of fuzzy C-means algorithm.The optimized fuzzy C-means algorithm is used to divide the positioning sub-regions,avoiding the traditional online stage to traverse and calculate the entire database.The online positioning stage only needs to calculate the pending points to belong to a certain positioning sub-region to achieve "coarse positioning" and reduce the amount of calculation in the online positioning stage.(3)Design the dynamic KNN algorithm,set the fixed k value in the traditional KNN algorithm as the dynamic value,and set the dynamic weight.The k value is dynamically determined according to the distribution of the reference points of the positioning sub-region to which the to-be-determined point belongs,and different weights are assigned according to the different contributions of the reference points to the calculation of the coordinates of the to-be-fixed position,and the calculation of the coordinates of the to-be-fixed position is optimized.The thesis expounds and analyzes from three aspects of data filtering,positioning sub-region division and online phase position matching.Experimental tests show that the FDWKNN position fingerprint positioning algorithm has improved in accuracy and stability.