Joint Optimization Of Precoding And Decoding Matrices For MIMO VLC Systems

Visible light communication(VLC)is a communication technology that uses visible light as a transmission medium for information transmission.Compared with traditional radio frequency(RF)communication,VLC has the advantages like abundant spectrum resources,high transmission rate,good confidentiality and low cost.VLC is deemed to be an emerging wireless communication technology with wide application prospects.Since the multiple-input multiple-output(MIMO)transmission method can greatly expand the channel capacity,the issue of how to apply MIMO transmission method in VLC system has received a lot of attention from academics.This paper focuses on the joint optimization of the precoding matrix at the transmitter side and the decoding matrix at the receiver side of MIMO VLC systems,and the main work is as follows.Firstly,the problem of robust precoding design for single-user MIMO VLC systems in the case of imperfect channel state information(CSI)is investigated.To enhance the robustness of the transceiver scheme,two commonly used channel estimation error models-stochastic error model and bounded error model-are introduced.Optical power constraints with no amplitude and mean restrictions on the modulation method are established based on the characteristics of light-emitting diode(LED)and indoor dimming control.And the maximization of system signal-to-noise ratio(SINR)is taken as the optimization objective,and a corresponding mathematical optimization problem is formulated for the design of the precoding matrix at the transmitter side.Under both two error models,the nonconvex optimization problem are transformed into approximately convex problems using an iterative method named successive convex approximation(SCA),and then the suboptimal solutions of the original problems are obtained by a robust convex optimization method.Simulation results show that the proposed robust designs can obtain significant performance gains compared to the non-robust design.Secondly,the problem of precoding design based on reinforcement learning(RL)method for maximizing the lower bound of achievable rate in single-user MIMO VLC systems is studied.The precoding scheme selection in the MIMO VLC system is modeled as a reinforcement learning problem,and a precoding design method based on the Q-Learning algorithm of RL is proposed.The method takes the precoding scheme that satisfies the optical power constraint as the selectable actions,while the two performance metrics,system achievable rate and user bit error rate(BER),are used as the immediate reward function.Simulation results show that the RL-based precoding design method can approach the optimal precoding strategy after a sufficient number of time slot learning iterations to accomplish the design goals of improving the system achievable rate and reducing BER of user.Thirdly,the problem of robust transceiver design for multi-user MIMO VLC systems in the case of imperfect CSI is discussed.Under the bounded error model,the single-user optical power constraint is extended to the multi-user scenario,and the joint optimization problem of the transmit precoding and receive matrices is formulated with the maximized minimum fairness(MMF)function,the harmonic mean(HM)function,the proportional fairness(PF)function,and the weighted fairness(WF)function,respectively,as the objective functions.The solution methods are given for each of these four objective functions.For MMF problem,the bisection method and robust convex optimization method are used to transform the problem into an equivalent feasibility problem.For HM,PF and WF problems,the successive parametric convex approximation(SPCA)method is used to solve the approximate convex problem to obtain the local optimal solution.The simulation results show that in terms of the total system SINR performance,the WF scheme is the best,followed by PF and HM schemes,while MMF scheme is worst.However,in terms of user fairness metrics,the situation is reversed.Finally,the robust transceiver design problem based on the block diagonalization(BD)criterion for multi-user MIMO VLC systems in the case of imperfect CSI is investigated.Under the multi-user optical power constraint and stochastic error model,the BD algorithm is introduced,and the joint optimization problem of multi-user MIMO VLC transceiver is formulated with the optimization objective of maximizing sum rate of the system.The optimization problem is solved by incorporating the weighted mean square error minimization(WMMSE)problem into the iterative solution of the sum-rate maximization problem.Simulation results show that the proposed robust design has good system sum rate performance and MSE performance,and is robust to channel uncertainties compared to the conventional BD-based transceiver design.