Research On Beamforming Technologies Of IRS Empowered 6G Communications

Apart from the basic network performance promotion,in the future sixthgeneration mobile communications(6G),a brand new paradigm for the integration of communication,control,sensing,and learning at the physical layer is emerging globally.Driven by the rapidly growing quality of service(QoS)requirements,the new paradigm may face massive challenging communication engineering issues.Against this backdrop,industrial and academic communities are increasingly interested in proactively controlling wireless propagation environments.Recently,a new 6G technology,intelligent reflecting surface(IRS),has been considered as a spectral-efficient,cost-effective,and energy-efficient solution for future 6G wireless networks.IRS can increase the signal-to-noise ratio or eliminate signal interference by reconfiguring the wireless propagation environment,which provides a new degree of freedom for wireless system optimization.Due to low-power and decentralised communications,IRSs are promising to be widely applied in 6G networks to achieve the goals of energy saving,blind spots elimination,interference coordination,and network capacity improvement.However,the research on beamforming designs for IRS-assisted wireless systems is still in its infancy.Effective transmission schemes,wireless resource allocation and IRS configuration algorithms brought urgent challenging problems for IRS-assisted 6G wireless scenarios.In order to fully reap the immense benefits of IRS,this thesis mainly focuses on two optimization goals of energy saving and network capacity improvement in terms of the IRS design.The thesis also studies the beamforming designs for both single-IRS and multi-IRS setups,under the perfect and imperfect channel state information(CSI),respectively.The specific research contents and main contributions are briefly summarized as follows:(1)This thesis proposes a power minimization scheme in the IRS-assisted multi-user millimeter-wave(mmWave)communication by considering delaysensitive tasks.By deploying the low-cost IRS in the mm Wave network,reflection links are created to ensure the transmission delay of services.With the robust transmissions created by the IRS,the scheme optimizes the transmit power of the devices with limited battery capacity.First,the joint optimization problem of the transmit power of the users,the multi-user detector of the base station(BS)and the passive beamforming matrix of the IRS is formulated.Then,we propose a power minimization algorithm based on alternating optimization.Specifically,we utilize the Neumann series expansion to reveal the closed-form update of the power solutions,thus avoiding matrix inversions.For the passive beamforming design of the IRS,an alternating direction method of multipliers(ADMM)and a sum-of-inverse minimization-based complex circle manifold optimization algorithm are developed,respectively.Instead of the semi-definite relaxation method with high complexity,the proposed algorithms can efficiently deal with the non-convex constant-modulus constraints.The transmit power can be reduced with the stringent delay requirements by using the developed scheme.(2)On the basis of the above work,this thesis proposes a distributed IRSassisted multi-user mm Wave coverage enhancement scheme.Due to the lowrank BS-IRS channel at high frequency bands,the spatial multiplexing gains between the IRS to multiple users can be greatly suppressed.To support multiuser in the coverage-enhancing scheme for blind spots without direct links,a distributed IRS-assisted mm Wave communication model is proposed.We formulate a joint optimization problem of active beamforming at the BS and passive beamforming of multiple IRSs to maximize the weighted sum-rate of all users.For joint active and passive beamforming design,we employ the Lagrangian dual transformation of the original problem,and propose an alternate iterative optimization framework to optimize active and passive beamforming matrices in an iterative manner.To dissolve the non-convex passive beamforming of the IRS,an effective constraint relaxation method is devised.Furthermore,it is proved that the constraint relaxation-based solutions can guarantee the constant-modulus property of the IRS phase-shifts.(3)By extending to a multi-transceiver scenario,this thesis proposes an IRS-supported Device-to-Device(D2D)-underlaid cellular system.In view of multiple weak coverage spots in a cell,we propose to deploy multiple IRSs around the cell border to improve radio propagation of edge users.In the studied system,IRSs are utilized to provide the interference coordination for multiple transceivers,thus improving the system throughput.To ensure the QoS requirements of cellular users,the QoS constraints of cellular users and the D2D-cellular user pairing constraints are firstly formulated.Then,we model the joint optimization problem of D2D-cellular user pairing,power allocation of mobile devices,receive beamforming of the BS,and IRS passive beamforming.Finally,a block coordinate descent algorithm is proposed to decouple the multivariate optimization problem and solve each variable iteratively.To be specific,we model the D2D-cellular user pairing problem as a maximum weighted bipartite matching problem and solve it by using an efficient Hungarian algorithm to maximize the system sum-rate.In the considered scenario,the passive beamforming problem of the IRS is transformed into a complicated quadratic programming problem with multiple quadratic constraints.To reduce the computational complexity,an Riemannian manifold-based ADMM is proposed,which decomposes multiple quadratic constraints into independent subproblems and they can be efficiently solved in a parallel manner.(4)For the IRS design under non-ideal conditions,this thesis proposes a two-timescale transmission method based on outdated CSI for IRS-enhanced multiple-input multiple-output(MIMO)transmission systems,which can efficiently reduce the channel training overhead and signal processing complexity.Due to the fast-changing channels,the performance degradation is caused by the singular value decomposition-based precoding by using outdated CSI.To address this issue,we redesign the receiver architecture to eliminate the inter-stream interference.In particular,we propose a two-timescale optimization framework for the new transceiver structure.For the small timescale,we derive a water-filling algorithm for power allocation by using the developed transceiver architecture.For the large timescale,data-driven and model-driven particle swarm optimization(PSO)methods are developed for the IRS passive beamforming,thus maximizing the long-term average achievable rate.The proposed data-driven PSO algorithm adopts mini-batch samples to update the passive beamforming vectors,thereby further reducing the computational complexity.The model-driven PSO algorithm only requires the knowledge of statistical CSI,eliminating the need of channel samples.