Research On Channel Capacity Of MIMO Visible Light Communication System

With the rapid development of wireless network and its applications,the existing spectrum already cannot satisfy the increasing communication demand.Visible Light Communication has become the research hot spot of wireless communication because of energy-saving,the abundant frequency spectrum resources,safety.Multiple-input multiple-output technology has been applied in the VLC for the demand of high-speed communication.Moreover,with IM/DD,the transmitted signal of VLC should be real and non-negative.Besides,peak power of VLC should be limited for human eye safety.Thus,distribution of the optimal input channel capacity is different with the traditional RF signal.The channel capacity and the optimal power allocation mechanism of VLC MIMO are studied in this thesis.The main work and innovation points are as follows:1.To solve the problem of unknown channel capacity of VLC MIMO system,the approximate gradient descent method is proposed to obtain the exact channel capacity and optimal discrete input distribution.Under the peak,average optical the electrical power constraints,an optimization problem of exact channel capacity is established with discrete input distribution.The numerical integration method is proposed to approximate the objective function and its gradient,and the inexact gradient descent method is taken to solve the non-convex problem to obtain the exact channel capacity and optimal discrete inputs.Simulation results show that only two discrete points are needed to achieve the exact channel capacity in low signal-to-noise ratio region.2.To obtain the closed-form expression of channel capacity of VLC MIMO system,Taylor expansion and Frank-Wolfe algorithm are constructed,and the expression of upper and lower bounds of channel capacity are obtained.Specifically,problems of finding upper and lower bounds of channel capacity are modeled into optimization problems which are non-convex.With first order Taylor expansion,the problems can be turned into convex,and Frank-Wolfe algorithm can be using to solve them.In addition,in the case of continuous distributed inputs,a closed-form expression is proposed.Simulation results verify that discrete and lower bounds are compact in the low SNR region,the channel capacity of discrete lower bound is higher than that of continuous lower bound.When SNR is 5d B,the channel capacity of discrete lower bound is 0.526bit/s/Hz higher than that of continuous lower bound.3.Aiming at the problem of maximizing the achievable rate of VLC MIMO system,this thesis designs optimal power allocation scheme based on arbitrary discrete input distribution.To be specific,the relationship between the minimum mean square error and mutual information is used to establish the power allocation optimization problem.Then,Lagrange function and Karush-Kuhn-Tucker(KKT)conditions are constructed to derived the expression of the optimal power allocation parameters,and bisection method is proposed to find the optimal power allocation scheme.Then,the optimal power allocation scheme based on Jensen's inequality is designed.Finally,this thesis analyzes the total bit error rate of On-Off Keying(OOK)Modulation and Pulse Amplitude Modulation(PAM)for VLC MIMO system while constellation position is given.Simulation results show that optimal discrete inputs have better performance on achievable rate and bit rate error than equal probability discrete inputs.This thesis contains 27 figures,4 tables and 86 references in total.