Analysis And Optimization Of Massive MIMO Systems With IQ Imbalance

When the results of theoretical research is applied to practical systems,the hard-ware impairments brought by manufacturing techniques or system design problems must be taken into account.The existence of hardware impairments makes the sys-tem capacity,spectrum efficiency,and accuracy of signal detection have certain losses.Compared with the traditional system,the use of high-order modulation,large-scale antenna,millimeter wave and other technologies in the future communication system makes the impairment characteristics of the system and the influence of hardware im-pairments have different performance.For example,in the millimeter wave band,the impairment characteristics of the device are more prominent;the implementation of massive multi-input multi-output(M-MIMO)systems requires lowering the cost and power consumption of each RF unit of the base station,which will introduce more se-vere impairment characteristics.In the actual system implementation,the impairment characteristics can't be ignored.In the research of key technologies of communication,hardware impairments need to be added to the system model,and the influence of im-pairment characteristics should be explored,which has important research significance and practical value.This dissertation focuses on the problems of the in-phase quadrature-phase imbal-ance(IQI)in the fully-digital beamforming systems,hybrid beamforming systems and finite resolution systems with large-scale array.The channel estimation method,down-link precoding scheme and achievable channel rate in time division duplexing(TDD)transmission mode are taken into consideration.The main research results and contri-butions are as follows:First,the performance of M-MIMO systems with IQI at the user equipments(UEs)and the base station(BS)is analyzed.Based on the symmetrical IQI parameters model,the uplink effective channel is estimated using least-square(LS)and linear-minimum-mean-square-error(LMMSE)criteria,and the achievable rate in the downlink is an-alyzed with ZF precoding scheme.The closed-form expressions of the mean square error(MSE)of channel estimation and the achievable channel rate are derived,and the asymptotic behavior is given under the condition that the number of base station an-tennas or the transmit power tends to infinity.The results show that the impact of BS IQI on uplink effective channel estimation is more prominent.The impact of user IQI on channel estimation can be neglected.Meanwhile,increasing the length of training sequences can effectively reduce the impact of BS IQI.The existence of IQI limits the achievable channel rate with the increase of antenna number and transmission power,and UE RX IQI mainly determines the boundary value.The influence of the BS IQI on the achievable channel rate can be weakened by increasing the number of antennas.Next,an estimation algorithm of the IQI parameters and wireless channel is pro-posed.According to the complex model of M-MIMO systems with IQ imbalance at the UEs and the BS,the I and Q branches are regarded as two independent virtual links.Based on the augmented real-valued channel in the uplink,IQI parameters and wire-less channel are jointly estimated.The downlink precoding design is implemented with the estimated wireless channel information,and a widely-linear zero-forcing(WL-ZF)precoding scheme is used to design the precoding matrix in order to weaken the impact of the uplink IQI on the downlink achievable rate.The performances of the downlink precoding based on the uplink augmented channel information and that based on the wireless channel information are compared by simulation experiment.Results show that the latter will further improve the performance of the achievable rate,and the more serious the IQI is,the higher the benefit will be achieved.Meanwhile,comparing the performance of the algorithm under different iterations,it can achieve excellent estima-tion results within five iterations,which verifies the effectiveness of the algorithm.Additionally,the hybrid millimeter-wave M-MIMO systems with IQ imbalance are investigated.Based on the knowledge of the wireless channel,the closed expres-sion and asymptotic characteristics of the achievable rate are analyzed,considering the radio frequency(RF)beamformer designed with phase extraction and the baseband(B-B)beamformer designed with ZF precoding and WL-ZF precoding scheme.The per-formance difference between FD and HBF is provided,and the impact of phase shifter precision is given as well.The analysis and simulation results show that the impact of IQI on HBF systems is more serious than that on FD systems,and the impact of UE IQI on HBF systems is similar to that of the BS IQI.When IQI exists,WL-ZF precoding scheme can bring significant benefits to both FD and HBF systems compared with ZF precoding scheme.Meanwhile,the effect of phase shifter resolution can be neglect-ed when the quantization bit number of the phase shifter is larger than 3.The process of HBF design and channel estimation based on single frequency signal and training sequence is given,which help to complete communication link.Finally,M-MIMO systems with IQ imbalance and finite resolution are studied.Based on the additive quantization noise model,the performance of channel estima-tion in the uplink and the achievable rate in the downlink is analyzed.Considering the specific training sequences,the closed-form expressions and asymptotic characteristic-s of the complex effective channel and real-valued augmented channel estimation are given.The approximate expressions and asymptotic characteristics of achievable rate in the downlink with ZF and WL-ZF precoding schemes are presented as well.Theo-retical analysis and simulation results show that both quantization error and IQI limit channel estimation error to a non-zero error platform,and the training sequence with strong orthogonality can only eliminate the impact of IQ imbalance.Meanwhile,the quantization error and IQI limit the achievable rates to finite ceilings.The impact of the quantization error and IQI at the UEs mainly determine the final rate boundary value,and the impact of the quantization error and IQI at the BS can be weakened by increasing the number of BS antennas.