Research On Compensation Technology For The Radio-Frequency Non-Idealities In Terahertz Communications

With the rapid development of Virtual Reality(VR),lossless 4k video real-time transmission and other high speed transmission business,the wireless transmission rate provided by 60 GHz millimeter wave communications,the representative technology of the fifth generation mobile system(5G),can't satisfy people's needs fully.Terahertz(THz)communication has a large instantaneous bandwidth so that it can achieve the wireless transmission rate of gigabit per second or even terabit per second.These traits make THz communication highly valued by scholars around the world as an important alternative technology of the beyond 5G or even 6G.However,with the rise of communication frequency band,the system is affected by the radio-frequency non-idealities more seriously,especially the Phase Noise(PN)and IQ Imbalance(IQI),even has the trend of becoming more serious in higher frequency band.It's necessary for THz communication systems to compensate those non-linear distortions caused by radio-frequency nonidealities in digital baseband.This paper focuses on the model-building,analysis and baseband compensation of frequency independent IQI and PN for THz communications.Because THz waves are greatly attenuated by air,long-distance communications need to use a power amplifier chain to amplify the signal.Considering the signal with higher Peak to Average Power Ratio(PAPR)is more likely to be amplified nonlinearly,leading to signal distortion,and the fact that high frequency communications cause higher carrier frequency offset,so this paper chooses the Single-Carrier with Frequency Domain Equalization(SC-FDE)system which is insensitive to carrier frequency offset and has the intrinsic characteristic of low PAPR to carry out our research.The main content of this paper includes:The first chapter of this paper introduces the background of THz communications,emphasizes the necessity of baseband compensation for radio-frequency non-idealities and then does a rough summary on IQI and PN.The second chapter firstly derive the parameter model for THz frequency independent IQI and draw a conclusion that it deteriorates the system performance by introducing image interference.Then,a signal detection algorithm based on supervised learning is proposed utilizing the feature that IQI is determined by hardware and it's approximately constant for a specific system.This paper accepts raw i/q data as features to train a multi-classifier based on Support Vector Machines(SVM)utilizing the regular distribution of the constellation points after the equalization of the data affected by IQI,which achieves correct decision on the data affected by IQI.It has a significant performance improvement compared with the traditional method based on channel energy variation in Non Line of Sight(NLOS)channel.The third chapter carries out research on the time domain simulation model of THz PN.PN model of THz band can't follow that of the 60 GHz millimeter wave due to the different link structure of radio-frequency local oscillator.Firstly,This paper demonstrates the standard PN model of millimeter wave based on the link structure of radio-frequency local oscillator using the theory of transfer function.Secondly,analysing different frequency conversion structures proves that the PN of THz communication systems is determined by the THz radio-frequency local oscillator.Furthermore,the power spectrum representation of THz PN is derived through transfer function basing on the similar link structure between the radio-frequency local oscillator of THz communication systems and 60 GHz millimeter wave communication systems.Finally,the paper proposes the time domain simulation model of THz PN combining the simulation theory of power-law noise and proves that the influence of PN on SC-FDE system is embodied in Common Phase Error(CPE)and Inter-Symbol Interference(ISI).Chapter four focuses on the research of compensation algorithm for PN.Firstly,the PN compensation algorithm is summarized,and a representative algorithm architecture flow chart is given.Then,for non-iterative compensation algorithm,the paper proposes a Minimum Mean Square Error(MMSE)based interpolation method using the statistical characteristics of PN described in 802.11 ad standard,and analyze the feasibility of migrating it to THz PN compensation.Its performance improves obviously Compared with the traditional non-iterative compensation methods.Finally,for iterative compensation method,the paper proposes a decision-feedback iterative frequencydomain equalization algorithm,only using the correlation of PN points,which can be used for THz PN compensation directly.It performs better than traditional decisionfeedback Least Mean Square(LMS)algorithm in bit error performance,time complexity and hardware implementation difficulty.