Research On Indoor Lighting Source Layout Based On Neural Network

At present,the indoor visible light communication technology that combines lighting and communication has attracted the attention and research of many scientists and scholars.It has many advantages such as high electro-optical conversion efficiency,rich spectrum resources and simple modulation.How to improve the flatness of the communication plane receiving power distribution is essential for improving the stability of the entire communication system.This article has carried out in-depth research on this issue,mainly completed the following work:Firstly,theoretical analysis and channel modeling of indoor communication system are carried out.The luminescence principle and basic optical characteristics of white LED,such as luminance,flux and luminescence intensity,are mainly analyzed.Then the system channel is modeled and analyzed,and the concept of channel matrix is proposed for the first time,which lays a theoretical foundation for the following simulation work.Secondly,the problems and shortcomings of the existing lamp source layout scheme are studied.The existing four-source,grid,round,elliptical,convex optimization and evolutionary based lighting source layout schemes are simulated and analyzed by MATLAB.And the following conclusions are drawn: the existing various light source layout schemes are not particularly perfect,and all have various defects and deficiencies,which cannot better ensure the stable communication and office lighting requirements in the room.Finally,in order to further improve the flatness of the power distribution of the communication plane we present a light source layout scheme based on neural network.The ceiling is evenly divided,an optical transmitter is placed in the center of each divided block,and its transmit power is set between 5 and 10 watts.The trained neural network containing with four hidden layers is used to search the power matrix and obtain the required lamp source layout scheme.The simulation analysis of the scheme is carried out by MATLAB.The results are as follows: The power distribution scheme of the communication plane based on the neural network has a power distribution range of-3.3dBm~-3.0dBm,and the fluctuation of the received power is 0.2dBm.The illumination range of the communication plane is 770lx~830lx.The experimental results show that the proposed scheme can further improve the flatness of the planar power distribution compared with the existing schemes...