Research On Multiband Digital Predistortion Technique

As to comply with the pace of increasing requirement of the linearity of power amplifier(PA),the research of digital predistortion(DPD)technique has been a significant focus in this field.In the case of the prevalence of the 4th generation mobile communication technology(4G),the distortion caused by PA can be particularly complicated in multiband communications.It shall be apparently distinguished from the communication via single channel that the solution of DPD will differ.In the context of multiband digital predistortion technique,herein,the main research is specific to the model and algorithm of DPD.In accordance with different multiband communications and PAs,it is respectively presented the detailed descriptions of a cascaded digital predistortion methodology,a 2-dimension(2-D)DPD model and a 3-dimension(3-D)DPD model.Preferably,first,a plurality of basic DPD models are described herein.Then,three specific problems of multiband communications,directed at which it is contemplated two applicable DPD structures,i.e.cascaded digital predistortion and multidimensional predistortion.Additionally,three typical DPD validation platforms are introduced as well and a simulation platform with configurable parameters is designed based on MATLAB.Second,with respect to the scenarios of three types of multiband communications,it is proposed a cascaded digital predistortion methodology with low complexity,fast training speed and dedicated functionality.Further,the principle of polynomial inversion based DPD and an adaptive delay compensation approach for handling the delay in DPDPA system loop are also explained in detail.By virtue of substantial simulation,superior benefits of the proposed DPD are exhibited compared with a single structured DPD.Moreover,in terms of the Doherty PA under the excitation of dual band signal,there are two types of the paralleled 2-D DPD compared,i.e.2-D DMP model and 2-D DVR model,wherein a dual loop learning structure provided with more precise extraction of model coefficients is schematized through the combination of direct and indirect learning mechanism.By means of actual test,it can further account for the each benefit of the aforesaid two 2-D DPD models with reference to various nonlinearity measurements.At length,as to the envelop tracking PA under the excitation of dual band signal,it is introduced a 3-D DPD which further takes into account the nonlinear term of the signal's envelop.Due to the ill-conditioned matrix resulted from conventional memory polynomial during the calculation of inverse matrix,thereby 3-D DMP model and 3-D FS model are respectively formulated from the perspectives of cutting down the number of DPD coefficients and eliminating ill condition.Each characteristic of the two models are further embodied by actual test.Preferably,on condition that the number of model coefficients is high,3-D FS model is to be utilized for avoiding the detrimental ill condition.