Indoor Visible Light Fingerprint Localization Algorithm

In the process of social information construction,location-based service has become an indispensable part of people’s life.At this stage,navigation and positioning systems such as GPS and Beidou are limited and cannot provide high-precision positioning in indoor environment.At the same time,traditional indoor positioning systems such as Wi-Fi and UWB are limited by factors such as facility deployment,positioning cost and electromagnetic sensitivity.Visible light positioning(VLP)based on LED has the advantages of low installation cost,no electromagnetic interference and high security,which has attracted extensive attention of scholars.This thesis studies the indoor visible fingerprint location technology.Firstly,it summarizes the research background,application prospect and research status of indoor visible light positioning technology,and defines the research objectives and contents of this paper.Secondly,the channel characteristics of visible light communication(VLC)based on LED lighting are modeled and analyzed.Then the typical methods for indoor visible light location are introduced.On this basis,the fingerprint location method based on neural network is studied.The main work includes:In order to solve the problem of low accuracy and poor stability,a novel fingerprint localization algorithm based on one-dimensional convolution neural network is proposed.The algorithm uses receive signal strength(RSS)as fingerprint information and the same fingerprint library to analyze the two network models experimentally.In the indoor positioning scene of 5m×5m×3m,the proposed positioning scheme can obtain high positioning accuracy with the average positioning error of 4.44 cm.Through simulation experiments,the performance of several different indoor visible light positioning algorithms is compared and analyzed,and also the technical advantages of the proposed algorithm is verified.Aiming at the improvement of generalization of indoor fingerprint location algorithm based on neural network,the time difference of arrival(TDOA)fingerprint information of VLC information transmission is collected by cross-correlation detection method,and it is combined with angle of arrival(AOA)to construct fingerprint database.Using the characteristics of joint fingerprint database,a fingerprint location algorithm based on long short term memory network(LSTM)is proposed.The proposed algorithm is evaluated by simulation experiments in the positioning area of 10m×10m×3m.After paralleling with the localization algorithms of different network structures,it highlights the generalization of the proposed algorithm,that is,it has excellent positioning accuracy under different signal-to-noise ratio.In order to locate and track indoor mobile personnel,a fingerprint location algorithm based on Kalman filter is proposed.Because the traditional fingerprint matching algorithm KNN has poor positioning effect,the random forest algorithm is introduced into fingerprint positioning.Firstly,the random forest algorithm is used to make the initial position estimation of the moving target to be measured,and then the result is optimized by Kalman filter.Simulation results show that the filtered personnel movement trajectory comes closer to the real trajectory,which shows that Kalman filter can ameliorate the positioning performance.Further experiments suggest that the the fingerprint location method with Kalman filter can ameliorate the location effect of indoor edge area,and significantly improve the location performance in the case of indoor occlusion.