Research On The Key Technology Of Sub-sampling Digital Predistortion

In high-speed,high-bandwidth communication systems,the performance of the power amplifier(PA)is increasingly important to the system.However,the nonlinear distortion problem of PA can lead to a deterioration in the quality of the output signal,which in turn limits the performance of the communication system.In order to compensate for the nonlinear distortion of PA and thus improve the performance of the system,the researchers conducted in-depth research on PA linearization technology.As one of the linearization technologies,digital pre-distortion(DPD)technology has been widely concerned and applied by academia and industry because of its excellent linearization ability,moderate cost and strong adaptability.In conventional DPD systems,in order to obtain sufficient distortion information of the PA output signal for the calculation of pre-distortion coefficients,the bandwidth of the feedback channel often needs to be 3 to 5 times the bandwidth of the original signal.As the transmission signal transitions from narrowband to wideband,the bandwidth requirements of the feedback channel increase accordingly,which places higher demands on the feedback branch ADC,resulting in increased power consumption and hardware cost of the system.In addition,compared with narrowband signals,the nonlinear distortion effect caused by wideband signals is more serious,and conventional DPD systems are difficult to meet the needs of linearization.In this paper,the related techniques of DPD are studied in depth,focusing on the method of using low sampling rate feedback signals to achieve pre-distortion coefficients estimation,thereby reducing the system demand for the ADCs in the feedback branch,which can meet the challenges of wideband communication ADCs and reduce hardware costs in a certain extent.Firstly,in order to cope with the problem that conventional DPD systems need higher sampling rate ADCs when processing wideband signals,a DPD method of signal reconstruction by modeling PA is proposed.This method reconstructs the feedback signal sampled by the low-rate ADCs to approximate the real PA output signal at the original rate to improve the accuracy of pre-distortion coefficients estimation,thereby improving the linearization ability of the DPD system.Simulation results show that the proposed DPD method can reduce the EVM from 7.654% of the DPD-free system to 0.213% and 0.363%when the ADC sampling rate is 1/2 or 1/4 of the DAC sampling rate.Then,the influence of IQ imbalance on the linearization ability of DPD system is discussed through theoretical analysis and simulation experiments.In order to cope with the IQ imbalance problem introduced by the quadrature demodulator,a DPD structure that uses the in-phase signal of the feedback signal to realize signal reconstruction and then estimate the pre-distortion coefficients is proposed,which can avoid the influence of IQ imbalance in a certain extent.Subsequently,the DPD structure is generalized to the sub-sampling system for application and testing,and the simulation results show that the sampling rate of the feedback branch ADC has little effect on the linearization ability of the structure,and the EVM of the system is about 1.195%.Finally,the problem of delay of signal transmission in actual circuits is discussed,and the influence of delay on DPD system is studied.Aiming at the delay estimation problem in sub-sampling DPD system,two delay estimation structures suitable for sub-sampling system are proposed.By introducing a cascaded integrator-comb filter to upsample the feedback signal,the low-rate feedback signal is matched to the baseband signal rate,and the loop delay is estimated by combining the Euclidean distance,and the feasibility and effectiveness of the method are verified by simulation experiments and FPGA board experiments.