Research On Indoor Visible Light Positioning Optimization Algorithm Based On Received Signal Strength And Image Sensor

Indoor positioning technology has great needs and wide applications.At present,many indoor positioning technologies are mainly based on radio frequency(RF)and visible light communication(VLC).The positioning accuracy that can be achieved by RF-based positioning technology ranges from a dozen centimeters to a few meters,which is a large error for indoor positioning and has electromagnetic interference problems.With the development of visible light communication,the visible light positioning technology(VLP)is considered to be an attractive indoor positioning technology.However,there are still many important issues to be solved in designing VLC-based positioning systems.First,for the positioning system using the image sensor at the receiving end,since the image sensor reads the pixel coordinates of the image,there is a quantization error when estimating the image coordinates of the light source image,which may result in a large positioning error.Secondly,the positioning system using the photodiode at the receiving end is sensitive to the received signal strength.Therefore,the influence of the reflected light on the ranging cannot be ignored,and there is also a problem of generating a large positioning error.In the face of these problems,the main research work of this paper is as follows:(1)An indoor positioning system based on visible light communication is introduced.The channel modeling of visible light communication,including the line-of-sight(LOS)link and the non-line-of-sight(NLOS)link model,is established.The main ranging methods and positioning algorithms for visible light positioning and the evaluation indicators of the positioning system are introduced.(2)For the image sensor(IS)based positioning system,a single IS based localization algorithm is proposed.The imaging area of the LED light source is considered in the algorithm,and an improved bilinear interpolation algorithm is used to calculate the pixel coordinates of the image.The research shows that the proposed positioning error is better than the nearest neighbor(NN)algorithm.The simulation results show that the proposed scheme can achieve centimeter-level accuracy,the average positioning error on the two-dimensional plane is 5.83cm,and the 90%error is within 9.75cm.The average positioning error is better than the nearest neighbor(NN)algorithm.At the same time,the parameters affecting the positioning system are simulated.The results show that the larger the resolution or the larger the focal length,the higher the positioning accuracy.When the number of pixels per line is greater than 2500,the positioning accuracy tends to be stable.On different height planes,the proposed positioning scheme can maintain centimeter-level positioning accuracy.(3)For the photodiode(PD)based positioning system,a positioning scheme that reduces the influence of reflection is proposed.This scheme combines an efficient signal selection algorithm with a weighted mean filtering algorithm in image processing.Firstly,using a k-dimensional tree(kd tree)and a k-Nearest Neighbors(kNN)to filter out four source signals participating in the positioning.The normalized power distribution is then used as a weighting factor to calculate the coordinates of the location.Simulation results show that compared with the algorithm of selecting the signal by sorting,the proposed signal selection algorithm can save ten times of running time.The average positioning error of the proposed scheme is 8.06cm.Compared with the traditional received signal strength ratio(RSSR)based algorithm,the positioning accuracy improves by approximately 3 8 cm.