The Analysis And Optimization Of The Efficiency In Full-Duplex Two-Way Relay Networks

As the potential technology of next generation communication system,massive MIMO and full-duplex two-way relaying both have the potential to improve the spectral efficiency and energy efficiency.In this paper,we mainly analyze and optimize the spectral efficiency and energy efficiency in the full-duplex two-way relay networks.Massive MIMO full-duplex two-way relay network is firstly investigated,the pilot contamination and RLI which can not be replaced by an additional Gaussian niose are both considered.When the number of relay antennas tends to be infinite,based on MRC/MRT beamforming and ZFR/ZFT beamforming at the relay,we analyze the asymptotic system performance.The optimization algorithm of power allocation and antenna allocation at the relay is proposed for minimizing the whole system power consumption and maximizing the energy efficiency,respectively.These two optimization problems are both transformed to iterative geometric programming by approximation.Simulations show that both of the proposed algorithms can effectively optimize the system performance.Then,multi-pair MIMO full-duplex two-way relay network is investigated,channel estimation error at the relay and the RLI,which has been regarded as an additional Gaussian niose,are considered.An iterative algorithm is proposed to iteratively optimize the beamforming matrix at the relay and the power allocation for jointly maximizing spectral efficiency and energy efficiency.Semi-definite positive relaxation technology and the approximation method based on Taylor Expansion are utilized to obtain the beamforming matrix.And the optimization of transmission power is transformed to geometric programming.Simulations have proven that the iterative algorithm can effectively improve the spectral efficiency and the energy efficiency of full-duplex two-way relay network.At last,RLI is considered at single-pair full-duplex two-way relaying with channel estimation error,and the RLI can't be replaced by a Gaussian noise.Three different power allocation schemes for maximizing the individual end-to-end SINR,the minimum end-to-end SINR and the spetrcal efficiency,are proposed.Compared to the full power scheme,simulations show that all of the proposed schemes can improve the corresponding system performance while reducing the power consumption.