Performance Optimization Of Massive MIMO Systems Over Ricean Fading Channel

The unprecedented demand in services and applications requires the new generation cellular network to make a leap in the area throughput.For this reason,the 5th generation(5G)cellular network introduces new radio access that relies on the emerging advanced technology.That is,the massive Multiple-Input and Multiple-out(MIMO)systems,which can operate in the small cell networks(SCN)and heterogeneous networks(Het Net)to achieve multiple performance objectives.While the massive MIMO system uses massive antenna arrays at the base station(BS)to provide a spatially multiplexing of the users,the SCN and Het Net designs provide substantial cell densification and coverage solutions.The closeness of the BSs and users in the SCN and Het Net,coupled with the massive number of antenna array introduce a change in the ideal channel characterization.Compared to the Rayleigh fading channel,the Ricean fading channel provides a practically viable massive MIMO performance since the line of sight(Lo S)channel can be exploited in the SCN and Het Net,likewise,millimeter wave(mm Wave)band and massive MIMO antenna array harness the Lo S propagation.However,the asymptotic performance analysis of the Ricean massive MIMO channel is constrained by the Lo S propagation conditions.Therefore,this thesis investigates the performance optimization of the massive MIMO system in the Ricean fading channel,with the following summarized contributions.Firstly,the thesis analyzes the performance of the Ricean massive MIMO with a finite number of antenna arrays.By considering zero-forcing(ZF)beamforming(BF),the Lo S channel and finite antenna reduce the multiplexing and diversity gains.A novel channel covariance approximation that exploits the arbitrary rank channel mean is proposed and the eigenvalue moments are calculated.With the aid of series expansion,approximated and closed-form expressions of the ZF achievable channel rates are determined.It is shown that the proposed ZF achievable rates converge to the ideal channel rate in the respective number of finite BS antenna and Ricean component.The higher expansion order offers the required multiplexing and diversity gains to achieve superior performance.Secondly,the analysis is extended to an infinite BS antenna array system.By using Block diagonalization(BD),the optimal ZF precoder rotation becomes restricted,whereby the nullspace is larger than required due to the non-square Ricean channel matrix.So,the best optimal solution cannot be achieved with the per-antenna power allocation.A joint precoding and BF approach is designed to resize and focalize the Ricean massive MIMO channel,besides,the lattice basis is transformed to be near orthogonal.The proposed optimal ZF precoder outperformed other precoding schemes,where received bits were decoded in less error,reduced PAR,and considerable complexity.However,the Ricean-ness and correlated BS antenna reduce the effective beamforming channel dimension,which degrades the overall optimal ZF precoder performance.Substantial performance is,thus,achieved with both a small number of BS antenna and Ricean component.Thirdly,the correlated BS antenna and low-rank Lo S channel that reduce the beamforming channel dimension is addressed in the SCN.By estimating the Ricean channel in the uplink,a novel ZF beamforming that considers inter-cell channel gain is proposed in the downlink.Then,a tight per cell area rate and cell transmit power approximations are obtained.Simulation results depict the accuracy of the approximations in which the proposed ZF beamforming outperforms the conventional ZF beamforming,this reduces the impact of pilot and inter-cell interference contamination.Ultimately,with an increase in the BS antenna and Ricean component,the channel estimation error decreases significantly whereas the per cell area rate improves substantially,in spite of the cell size.Finally,due to the complexity of the channel estimation and severe inter-cell interference in the ultra dense SCN,the massive MIMO in the Het Net is studied.By assuming that locations of the BS and users as points of a Poisson point process,different distributions of Ricean(for the macro BS)and Rayleigh(for the femto BS)fading conditions are convoluted to obtain the signal to interference ratio(SIR).Then,approximated and closed-form expressions of the coverage and area rate are determined.And,a resource allocation algorithm is proposed to efficiently utilize the macro BS antenna,power and bandwidth,while alleviating inter-cell interference and guaranteeing the quality of service fairness for the users.The coverage performance is shown to be accurate and monotonically increases with the Ricean factor and BS antenna,especially,for the cell-edge users.