Study On Digital Predistortion Inearization Technique For Radio Requency Power Amplifiers

Along with the rapid development of broadband wireless communicationtechnology, the spectrum resources are increasingly scarce. In order to improve thespectrum efficiency, high spectrum efficiency modulation and transmission technologiesare proposed, which generate high peak-to-average power ratio (PAPR) broadbandsignals. Transmitting that kind of signals through nonlinear power amplifier results inin-band and out-of-band distortions, which will increase the bit error rate and causeadjacent channel interference (ACI), so broadband wireless communication systemsrequire high linearity power amplifiers. The power amplifier linearization technique hasbecome one of the key technologies of broadband wireless communication, andcurrently the digital predistortion technique has become a research hotspot.In this dissertation, the theories of nonlinear system and adaptive signal processingare applied to deeply analyze and study some problems about the digital predistortion,and a few optimization methods are proposed to improve the linearization performanceof digital predistortion. The main innovations and contributions of this dissertation aresummarized as follows:1. A new predistortion learning structure based on direct and indirect learningstructures is proposed. In this dissertation, the commutativity of power amplifier systemand predistortion system is proved at first, which also proves the rationality of indirectlearning structure. Because indirect learning structure is susceptible to the noise of thefeedback circuit, this dissertation proposes a new predistortion learning structure basedon direct and indirect learning structures, which can reduce the effect caused by noise ofthe feedback circuit, and improve the linearization performance of digital predistortion.2. A single-feedback non-iterative predistortion method based on look-up table(LUT) is proposed. Considering the computational complexity of the previousnon-iterative method, this dissertation proposes a single-feedback non-iterativepredistortion method, which has simpler structure, less computation complexity andhigher precision than the previous method, and it can reduce the hardware costs and simplify the design of the predistortion system.3. The single-feedback predistortion method based on memory polynomial isproposed. In the digital predistortion system, the feedback and sampled high speeddigital signals need the powerful processors, which increases the costs. Thesingle-feedback predistortion method is proposed in this dissertation, which can avoidthe IQ imbalance errors caused by the down-conversion circuit, reduce the costs andimprove the linearization performance of digital predistortion.4. A novel IQ imbalance errors estimation and compensation method for bothquadrature modulator and demodulator is proposed. The errors of both quadraturemodulator and demodulator will affect the design of the predistorter, and then degradethe linearization performance of digital predistortion. The conventional estimation andcompensation methods need additional feedback circuit. This dissertation proposes anovel method to simultaneously estimate and compensate the IQ imbalance errors causeby quadrature modulator and demodulator without additional feedback circuit, whichcan reduce the complexity and costs, and the precision also can be improved.5. A high precision nonlinear model for quadrature modulator and a jointcompensation model for memory power amplifier and nonlinear quadrature modulatorare proposed. Sometimes in order to fit the gain requirement of the following poweramplifier, the quadrature modulator needs to work near the maximal output level, in thissituation both the linear and nonlinear distortions of the quadrature modulator arepresent. High precision nonlinear model and compensation model for quadraturemodulator are proposed in this dissertation. Because both the quadrature modulator andpower amplifier are in the transmission channel, so the quadrature modulator and poweramplifier can be compensated jointly. Based on the joint compensation method and thenonlinear model of quadrature modulator proposed in this dissertation, a jointcompensation model for nonlinear quadrature modulator and memory power amplifieris proposed, which can reduce the complexity of the predistortion system.Experimental platform has been designed for validation, and the correctness of themethods proposed in this dissertation is validated by using that platform.