RIS-assisted Wireless Communication System Design Utilizing Statistical CSI

As one of the promising enabling technologies for future wireless communication systems,reconfigurable intelligent surface(RIS)has received wide attentions in both academia and industry.The RIS is a kind of artificial electromagnetic surface consisted of a large number of passive and phase-adjustable reflection units.It has the advantages of low cost,low power consumption,simple structure and high reliability.It can effectively improve the spectral efficiency,energy efficiency and coverage of the system.However,the above advantages depend on the accurate acquisition of channel state information(CSI).Due to its characteristic of passively reflecting electromagnetic waves,the acquisition of the instantaneous CSI through RIS is a challenging work.Compared with instantaneous CSI,the statistical CSI is approximately unchanged for a relatively long period of time.The transmission scheme using the statistical CSI has the advantage of strong robustness,and can be applied to both time division and frequency division duplex systems.It is important in both theoretical and practical aspects to investigate the design of RIS-assisted communication systems based on the statistical CSI.Therefore,this thesis studied the transmission design based on statistical CSI for the RIS-assisted multi-cell downlink system,so as to improve the service quality of users in the coverage hole at the cell-edge.Firstly,this thesis studied a multi-cell cooperative downlink transmission system in which each cell shares their statistical CSI.With only statistical CSI,one beamforming vector design algorithm and two reflection phase shift matrix design algorithms were proposed to maximize the ergodic weighted sum rate.Firstly,a closed-form approximated expression of the ergodic weighted sum rate was derived.Then,based on this closed-form approximated expression,the ergodic weighted sum rate maximization problem was modeled.To obtain the beamforming vector,the fractional programming(FP)and quadratic transformation were adopted to transform the original problem,and the Lagrange multiplier method was applied to solve the transformed problem.To obtain the reflection phase shift matrix,the successive convex approximation(SCA)-based algorithm was proposed.The original problem was relaxed into a convex problem,and the reflection phase shift matrix was ?obtained by solving the convex problem.Furthermore,in order to reduce the complexity of the SCA-based algorithm,the complex circle manifold(CCM)-based algorithm was proposed,which transforms the original problem into an unconstrained optimization problem in the complex circle manifold space.The simulation results verified the accuracy of the obtained ergodic sum rate approximation expression,as well as the effectiveness of the proposed SCA-based algorithm and CCM-based algorithm.Secondly,to reduce the information exchange overhead of each cell as well as the computation algorithm complexity,the RIS-assisted multi-cell non-cooperative downlink transmission system in which each cell only has the statistical CSI of its own users was studied.To maximize the ergodic weighted sum rate,a closedform beamforming vector expression was derived,and two RIS reflection phase shift matrix design algorithms were proposed.The closed-form asymptotic expression of signal-to-interference-plus-noise ratio(SINR)was derived.Based on the SINR asymptotic expression,the ergodic weighted sum rate maximization problem was modeled.To obtain the beamforming vector of each base station(BS),the original ergodic weighted sum rate maximization problem was decomposed into several independent problems,and the closed-form optimal beamforming vector of each BS was proved to be the statistical maximum ratio transmission(MRT).To obtain the reflection phase shift matrix,applying FP,the original problem was transformed into an unconstrained convex optimization problem in complex circular manifold space,and then solved by a proposed joint FPCCM-based algorithm.To further reduce the complexity of the joint FP-CCM-based algorithm,the CCMbased algorithm was proposed to directly solve the original problem to obtain the reflection phase shift matrix.The simulation results verified the accuracy of the approximate expression of the ergodic sum rate and the effectiveness of the two proposed design algorithms.Finally,for the above-mentioned multi-cell non-cooperative downlink transmission system,this thesis further considered the fairness among users.To maximize the minimum ergodic rate of cell-edge users,two reflection phase shift matrix design algorithms were proposed.With the derived SINR asymptotic expression,the ergodic minimum user rate maximization problem was modeled,and it was concluded that the optimal beamforming vector of each base station was still the statistical MRT.To obtain the reflection phase shift matrix,the SCA-based algorithm was firstly proposed,which used the first-order Taylor approximation to approximate the original non-convex objective function.By relaxing the non-convex unit modulus constraint of the RIS element’s reflection coefficient in the original problem,it was transformed into a convex problem,and the reflection phase shift matrix was ?obtained by solving the convex problem with the CVX toolbox.Furthermore,to reduce the computational complexity of the SCA-based algorithm,the projected subgradient(PSG)-based algorithm was proposed,and the double subgradient projection search was applied to solve the problem that the objective function can not be guaranteed to decrease in each search step.The simulation results verified the effectiveness of the proposed SCA-based algorithm and PSG-based algorithm.