System Design And Theoretical Analysis Of The Spectral And Energy Efficiency For Massive MIMO Systems

With gradually verifying the technical proposal of the fifth-generation(5G)mobile communication sys-tem,the world embraces a new generation of the communication system.As a key technique,massive multi-input multi-output(MIMO)has received an increasing amount of academic attention and industrial interest.By deploying a very large antenna array on the base station or the user terminal,massive MIMO can make full use of the space resources and significantly boost the space resolution.Hence,massive MIMO has the ability to improve the spectral efficiency and energy efficiency.However,some challenges associated with implementing massive MIMO should be considered,e.g.the pilot contamination caused by deficient pilot resources,insufficient accuracy of the channel estimation,and the increasing overhead of synchronization signal in time division duplex(TDD)systems.Focusing on spectral efficiency and energy efficiency of mas-sive MIMO,we determine the optimal system parameters according to the system performance analysis,and therefore provide a valuable guideline for developing an efficient system.Besides,we propose several novel techniques to further improve the system performance for massive MIMO,including joint processing of pilot and data signals,network full duplex(NFD),and antenna clustering of distributed antennas.Firstly,considering a multi-cell multi-user massive MIMO system with pilot contamination,we inves-tigate the system performance in terms of spectral efficiency and energy efficiency,and optimize the system parameters.Utilizing the joint multi-cell minimum mean squared error(MMSE)channel estimation,we de-fine an equivalent channel,and then derive the expression for the lower bound of the sum-rate in uplink transmission.In addition,we derive a closed-form expression of the sum-rate lower bound by using the large dimensional matrix theory,the properties of the convex function,and the order statistics theory.The proposed closed-form expression is independent with the small scale fast fading and the large scale fading related to the random locations of users.Specifically,the closed-form expression of the sum-rate only depends on the known system parameters,including the radius of the cell,the number of the cells,the distance between the adjacent cells and so on.Thus we obtain the optimal number of users and the power ratio between pilot and data.Then we focus on the power consumption model,and study the performance trade-off between spectral efficiency and energy efficiency.Besides,we derive the optimal power ratio between pilot and data of the maximum energy efficiency.The derived closed-form expression of the spectral efficiency clearly reveals the effects of the system parameters on the system performance.Furthermore,we study the spectral efficiency of the massive MIMO with the joint processing of the pilot and data in Ricean fading channel.Making full use of the channel state information(CSI)involved in the data transmission,joint processing can improve the system performance.Based on the relation of the mutual information to the information entropy,we divide the achievable rate into three parts with clear meanings:the capacity with the perfect CSI,the uncertainty of the channel estimation by solely pilot estimation,and the uncertainty of the channel estimation by joint processing.Assuming the line-of-sight(LOS)part of the Ricean fading channel is already known,we obtain the asymptotic expression of the achievable rate.Simulation results show that,comparing with the solely pilot estimation,joint processing applied in massive MIMO can achieve better spectral efficiency.When joint processing the pilot and data signals,increasing the length of the pilot signal always reduces the spectral efficiency.On the other hand,the spectral efficiency can be benefited from lengthening the channel blocks.Additionally,we present the concept of the NFD system and analyze the spectral efficiency and energy efficiency.From the point of view of the whole system,we design a novel full duplex strategy which does not need any strict synchronization signals,and it facilitates simultaneous uplink and downlink communications between users and different base stations.Besides,the NFD system can dynamically adjust the numbers of uplink and downlink base stations to cope with the growing trend of the asymmetric data traffic.We derive the asymptotic approximations of the achievable uplink and downlink rates with maximum ratio transmission(MRT)precoder and maximum ratio combination(MRC)receiver,as the antenna number approaches infinity.Considering an optical fiber connected backhaul network,a practical power consumption model is presented to study the system energy efficiency.Furthermore,we prove that a part of the downlink-to-uplink interference vanishes asymptotically,and this provides an opportunity of implementing NFD massive MIMO system in practice.Numerical results verify that when the number of antennas is large,the NFD system outperforms the TDD system in both the spectral efficiency and energy efficiency.Finally,aiming at maximizing the spectral efficiency of the entire system,we propose a large-scale dis-tributed antenna clustering strategy in NFD.We present an improved could radio access network(C-RAN)used in NFD,which introduces two switches to flexibly determine the working modes of antennas.In order to reduce the complexity,we derive the initial distributed antenna candidate sets by setting the thresholds of re-ceived signal-to-noise ratios(SNRs)for the uplink and downlink transmissions.Then,using reasonable power weight coefficients,we design a novel user ordering metric with considering the overlapped antenna number between different kinds of antennas.Furthermore,we use another metric called benefit-to-interference ratio to specify the antenna order for each user.In addition,we investigate a complementary antenna algorithm to reduce the system performance loss.Numerical results indicate that our proposed distributed antenna cluster-ing strategy in NFD with massive MIMO achieves a satisfying spectral efficiency and shows the application feasibility in reality.