Research On Indoor Fingerprint Location Technology Based On Channel State Information

Nowadays,the mobile network is developing rapidly,and there are more and more applications around it,including security,shopping mall and shop navigation,industrial assembly line,and real-time positioning of children.Outdoor positioning can rely on satellite positioning systems to improve more accurate location information,but indoors,due to signal shielding,multipath effects,and personnel mobility,it is difficult for satellite positioning systems to exert their advantages.Therefore,it is imperative to study indoor positioning technology.Compared with the Received Signal Strength(RSS),Channel State Information(CSI)has richer characteristics.Therefore,the positioning method that uses CSI has a location fingerprint is attracting more and more researchers' attention.However,the indoor environment is complex and the flow of people is frequent,these factors seriously affect the positioning accuracy.Therefore,there are still many problems to be solved to improve the positioning accuracy of indoor positioning technology.This article introduces the indoor fingerprint positioning technology based on CSI in detail.Based on the previous research work,the indoor positioning technology is improved,and the corresponding improvement methods are proposed for the defects of the existing technology,and verified by experiments.The feasibility and superiority of the scheme are discussed.First,build an experimental platform.Based on the Ubuntu 12.01 operating system,using Intel 5300 wireless network card and CSI Tools software platform as tools,the sending end and the receiving end are wireless routers and laptops respectively.Second,conduct indoor experiments.Extract multiple sets of CSI signals and analyze the influence of indoor multipath effects on CSI signals.Experiments show that CSI signals have a strong correlation with the physical environment.Conduct multiple experiments indoors to compare the effects of people,objects,etc.on CSI signals.Under the same location,different environments(human flow and unmanned flow,object obstruction and no object obstruction)have obvious differences in the CSI signal.In the same environment,the CSI signals of different locations are also different,but the closer the distance between two points,the higher the similarity of the CSI signals of the two points,so that the coordinates of the point to be measured can be predicted.Third,preprocess the extracted CSI signal.First,perform Kalman filtering and confidence filtering to denoise the received CSI signal to make the collected signal more stable and credible;second,perform phase angle correction on the signal to extract the true phase angle of the CSI;again,propose The algorithm for proportionally transforming the amplitude and the phase angle,and fusing the transformed amplitude and phase angle information,forming a group of 10 adjacent data packets,calculating the variance of their corresponding carriers,and extracting improved feature information;secondly,the principal component analysis method is used to reduce the dimensionality of the data to reduce the calculation time;finally,Support Vector Machine(SVM)method is used to train and predict the reference points and the points to be measured respectively,and the predicted coordinates are averaged to obtain the coordinates of the point.By comparing the coordinates of the measuring point and comparing the BP neural network algorithm and the K-nearest neighbor algorithm,it can be concluded that in an indoor environment,the indoor positioning algorithm based on the SVM algorithm has a higher positioning accuracy than the other two algorithms.Finally,the Ant Colony Optimization(ACO)algorithm is used to improve the classic SVM algorithm.According to previous research on ant colony optimization algorithm,it is proposed to apply ACO algorithm to optimize the indoor fingerprint location algorithm based on CSI.The SVM algorithm is optimized by the ant colony optimization algorithm to find the most suitable penalty factor c and kernel function parameter g in this environment.The positioning effect of this algorithm is compared with particle swarm optimization algorithm and genetic algorithm optimization SVM algorithm.Experiments show that this algorithm is better than the other two algorithms.