| In recent years,indoor positioning has become an essential service in life and work with the rapid development of computers and smart devices.Although the development of the Global Positioning System(GPS)is getting better and better,the signal is attenuated seriously for the satellite signal is blocked,which lead to low positioning accuracy.Indoor visible light positioning system based on Light Emitting Diode(LED)has gradually become the focus of research because of its wide coverage,long transmission distance,fast information transmission speed and simple implementation.In this paper,the basic principles and channel characteristics of indoor visible light positioning system are studied.Main factors affecting positioning accuracy is discussed.Some improvements have been made to the fingerprint algorithm which used in indoor visible light positioning technology regularly.For example,the received signal intensity fluctuates in a range when it is collected and it is considered as a normal distribution model.So the received signal strength indoors is preprocessed;Large amounts of data that is recorded and stored in the database has to be calculated with the RSSI at receiver in the positioning stage,which makes the calculation more complicated;The estimated position using the Bayes's theorem is quite different from the real position of the receiver.The specific work of the paper is as follows:Firstly,the model of indoor visible light positioning system is established.The transmitted signal strength,received signal strength as well as Signal to Noise Ratio(SNR)are simulated and analyzed.Secondly,the K-nearest neighbor algorithm and Bayes's' theorem,which are used in the fingerprint positioning algorithm more often,are simulated and analyzed.According to the positioning results and the advantages of the two algorithms,an improved algorithm is proposed to reduce the positioning error and the power consumption of the indoor visible light positioning system.Then the two filtering algorithms,mean filtering and Gaussian filtering,are used to preprocess the received signal strength.Besides,the filtering effects of the two algorithms are compared and analyzed.After establishing the fingerprint database,higher probability interval in the Gaussian filter process is selected to filter out some data in the fingerprint database,which improves the positioning accuracy and stability of the system.Finally,the fingerprint database is divided into several clusters.The clustering analysis algorithm is used to calculate the cluster with the largest posterior probability in the fingerprint database.The Bayes's theorem is used to estimate the position of the receiver in this cluster,which makes the positioning result of the Bayes's theorem much better. |
PDF Full Text Request