Wireless Transmission Methods Of Asymmetrical Massive MIMO Systems

The 5th generation mobile communication system(5G)has been commercially deployed.Supported by the massive multiple-input multiple-output(MIMO)technology,5G provides higher rate,lower latency,and accesses for more devices.Meanwhile,the overwhelming hardware cost and power consumption caused by the large-scale array has hindered the further development of massive MIMO.To reduce deploying cost while keeping the advantages of massive MIMO as much as possible,a novel asymmetrical full-digital architecture is recently proposed.By decoupling the transmit and receive radio frequency(RF)chains,the asymmetrical architecture allows different number of RF chains for downlink and uplink according to the practical system demand.The uplink array can be equipped with fewer RF chains than the downlink array to reduce the hardware cost,which reaches a compromise between the cost and performance.However,the downlink and uplink channel are not reciprocal due to the different channel dimensions,and the receive antenna selection also significantly impacts the energy efficiency.To solve the above problems,we investigate the wireless transmission methods in asymmetrical systems,which are listed as follows.Firstly,to cope with the non-reciprocity between the downlink and uplink channel,two efficient algorithms are proposed to reconstruct the complete downlink channel by utilizing the coprime array and channel sparsity.Considering one base station(BS)serves one single-antenna user with an asymmetrical massive array,the BS utilizes only partial antennas to receive user pilots,where the uplink antennas are connected to RF chains by the topology of coprime array.According to the received pilot signal,the observed signal on an virtual array which has the same dimension to the downlink array can be formed.Utilizing the observed signal and channel sparsity,two efficient algorithms are proposed to reconstruct the downlink channel based on the block-sparsity orthogonal matching pursuit and the alternating direction method of multipliers,respectively.The numerical simulation validates that the proposed scheme can reconstruct the downlink channel with high accuracy.Secondly,to maximize the energy efficiency of uplink asymmetrical systems,we exploit statistical channel state information to jointly select receive antennas and design the transmit covariance.Considering one BS serves one multi-antenna user with an asymmetrical massive array under Rician fading channels,the BS utilizes only partial antennas to receive user signal.For uncorrelated channels,the optimal number of selected antennas(denoted by K)and user transmit power are derived.For correlated channels,we propose the Frank Wolfe-Water Filling(FW-WF)algorithm to optimize the design of receive antenna selection and user transmit covariance,which can identify the optimal K swiftly.To further improve the energy efficiency,with a given K,we further propose the norm-minimization selection-WF(NMS-WF)algorithm.The numerical simulation validates that the proposed algorithms can achieve performance which is very close to the ones by exhaustive search with much lower complexity.