Research On Improved WLAN Indoor Fingerprint Location Based On Database Partition

Location Based Service is the cornerstone of realizing personalized smart services in the Internet of Things.Indoor positioning is easily affected by human activities,and improving indoor positioning accuracy is a hot spot in the field of Io T research.The positioning method based on Wireless Local Area Networks(WLAN)has the characteristics of wide coverage,low cost,high transmission rate,and scalability,and is widely used.The indoor fingerprint positioning technology based on WLAN is systematically researched and analyzed in this thesis.The main work is as follows:Firstly,the database establishment process of WLAN-based indoor location fingerprinting technology is studied.Aiming at the problem that the dimension of the collected original RSSI vector is too high,which leads to a large amount of calculation and the Wi Fi signal is interfered by environmental noise,kernel principal component analysis is used to extract fingerprint features and reduce the influence of noise on the RSSI value.In view of the traditional clustering algorithm only uses the Euclidean distance between RSSI vectors to represent the similarity between fingerprint data,which leads to some reference points being physically far away from other reference points of the same cluster,the position of AP corresponding to the strongest RSSI value combining with Euclidean distance is used to describe the similarity of fingerprints in this thesis.A database partition method based on SAP-FCM clustering algorithm is proposed.The simulation results show that the positioning accuracy of the proposed algorithm is better than the traditional database partitioning algorithm.Secondly,the location matching process of WLAN-based indoor location fingerprinting technology is studied.In order to improve the positioning accuracy of Wi Fi location fingerprints and reduce the computational complexity,a positioning algorithm combining improved particle swarm optimization and Least Squares Support Vector Regression(LSSVR)is proposed.The Particle Swarm Optimization(PSO)algorithm is optimized by simulated annealing to overcome the problem that the traditional particle swarm optimization algorithm is easy to fall into local optimum.The SAPSO algorithm is used to optimize the penalty factor and kernel function parameters of the LSSVR model to avoid the problem of low positioning accuracy caused by improper parameter selection.The simulation results show that,compared with the traditional LSSVR,GA-LSSVR and PSO-LSSVR algorithms,the proposed SAPSO-LSSVR algorithm is superior to the other three algorithms in terms of positioning accuracy and positioning time.