Research On Wireless Resource Allocation In Co-Time Co-Frequency Full-Duplex Networks

In recent years,there is an increasing demand for high-speed and high-capacity services.However,wireless networks nowadays have limitations such as limited spectrum resources,low spectral efficiency,and significantly increased energy consumption.Therefore,there is an urgent need to improve network capacity and energy efficiency.Co-time co-frequency full-duplex(FD),as a technology that enables signals to be transmitted on the same frequency band simultaneously,can theoretically double the spectral efficiency without introducing additional spectrum resources,thus attracting extensive interest from academia and industry.Traditional wireless networks adopt time division duplex or frequency division duplex technology,and thus it can only achieve relatively low spectral efficiency because spectrum resources cannot be fully utilized.By applying full-duplex technology to traditional wireless networks,the network spectral efficiency and energy efficiency can be effectively improved.However,the self-interference and co-channel interference caused by the intrinsic full-duplex property reduces the potential full-duplex benefit.How to further improve network spectral efficiency and quality of service(QoS)of users has become a key problem to be solved in full-duplex networks.Motivated by the above challenges,this dissertation conducts research on resource management and optimization in full-duplex networks.By solving problems such as subcarrier allocation,power control,and beamforming design,the spectral efficiency and energy efficiency of the network can be improved.The main contributions of this dissertation are as follows:Firstly,we investigate a joint subcarrier and power allocation problem for a full-duplex three-node network with the aim of maximizing spectral efficiency.Due to the coupling of power and subcarrier allocation as well as the uplink and downlink transmissions,the optimization problem is highly non-convex.Moreover,the non-convex QoS constraints and imperfect channel state information(CSI)make the problem more complex.We prove that the problem is non-deterministic polynomial time(NP)-hard.We present separate and joint optimization methods to solve the problem.Based on the preliminary results provided by the separate optimization,the joint optimization method is proved to converge to local optimal solutions.Simulation results demonstrate the impact of channel estimation error on spectral efficiency and the superiority of the proposed algorithms.When the minimum rate required by the downlink user is 1 bps/Hz,the joint optimization method achieves around 13.36%performance improvement in terms of spectral efficiency.Secondly,we investigate the joint beamforming design for access and fronthaul links in a user-centric network(UCN)with full-duplex fronthaul.In the formulated problem,clustering and beamforming are coupled due to the user-centric architecture.The access and fronthaul links are coupled due to full-duplex transmissions.Furthermore,the non-convex fronthaul rate constraint makes the problem more complex.We provide beamforming designs for perfect and imperfect CSI scenarios.Under the perfect CSI scenario,the successive convex approximation(SCA)method is applied to approximate the non-convex problem,through which the computational complexity is much lower than the traditional semidefinite relaxation method.Also,the method yields rank-1 beamforming solutions.Based on the Jensen’s inequality,a low-complexity beamforming design is proposed under imperfect CSI scenario.In this case,signaling overhead is largely reduced.Simulation results demonstrate the superiority of the proposed algorithms in terms of energy efficiency and power consumption under both CSI scenarios.When the maximum transmit power at macro base station(MBS)is 45 d Bm,the proposed FD beamforming design achieves 43.07% energy efficiency performance improvement when compared to half-duplex design.While the consumed power is only 1.84% higher than that of the half-duplex design.For our considered full-duplex fronthaul networks,moderate maximum transmit power of MBS and APs is sufficient to obtain desired energy efficiency performance.Moreover,the lower bound is verified to be tight at the low-maximum power region.Based on the previous work,multiple reconfigurable intelligent surfaces(RISs)are deployed to assist access and fronthaul transmissions to further improve energy efficiency.We investigate joint active and passive beamforming design for a reconfigurable intelligent surface assisted user-centric network with full-duplex fronthaul,aiming at maximizing proportional fair(PF)energy efficiency of the network.The problem is rather difficult to solve due to the unit-modulus constraint.The reflection model considered in the literature is oversimplified.Therefore,a simple single reflection model and a complex model that also considers inter-RIS reflection are investigated.Based on the strategy of relaxation and punishment,the unit-modulus constraint is approximated as a convex constraint.Numerical results show that the proposed schemes are superior to baseline schemes.Furthermore,the numerical results indicate the advantages of deploying RISs and optimizing phase shifts at RISs,and demonstrate that the performance gain of considering inter-RIS reflection is significant.When the maximum transmit power of AP is 19 d Bm,the performance gain of inter-RIS reflection and phase optimization are 18.7% and 65.9%,respectively.