LED Selection Techniques In Indoor Multi-user Visible Light Communications Systems

As the demand for high-speed wireless broadband networks continues to grow,traditional radio frequency(RF)communications are no longer able to meet people’s needs due to the exhaustion of the frequency spectrum.In recent years,visible light communications(VLC)has been researched as an effective complement to RF technology due to its high bandwidth,immunity to electromagnetic interference,environmental friendliness and security advantages.However,indoor VLC systems require a high number of light emitting diodes(LEDs)to cover the entire service area as evenly as possible while considering the indoor lighting function.In addition,as the light intensity from each LED to the user’s receiving end is equal,the simultaneous participation in the communication will lead to low energy efficiency,adjacent channel interference and high system complexity.In order to overcome the problems of high energy efficiency and complexity in multi-antenna services,spatial modulation(SM)and related antenna selection techniques have been widely used and researched,but the time complexity of traditional LED selection methods increases dramatically as the number of LEDs and users increases.In order to solve this problem,this thesis investigates the LED selection problem for multi-user scenarios in indoor VLC systems based on machine learning theory and methods.The main work in this thesis is as follows:● The LED selection model in an indoor multi-input single-output(MISO)VLC system is introduced.The shortcomings of traditional LED selection techniques with high time complexity are analysed,and an LED selection algorithm based on naive Bayes(NB)and support vector machine(SVM)is proposed,experimental simulations demonstrate that the NB algorithm proposed in this chapter significantly reduces the time complexity compared to other algorithms and achieves a good balance between time complexity and user transmission sum-rate.● In order to further enhance the spectral efficiency and improve the system performance of indoor VLC systems,this chapter discusses the GSSK VLC system based on non-orthogonal multiple access(NOMA)technology,Firstly,NOMA technology and indoor multi-user NOMA GSSK VLC system model are introduced;Then,the LED selection technique based on traditional methods in indoor multi-user NOMA GSSK VLC system still has the problem of high computational complexity,and the LED selection algorithm of NOMA GSSK VLC system based on NB and SVM is proposed;Finally,it is shown through simulation experiments that the NB-based LED selection algorithm proposed in this chapter can significantly reduce the time complexity and is more practical in practice than other algorithms in indoor multi-user NOMA GSSK VLC systems.