【作者】 詹鹏；

【导师】 秦开宇；

【作者基本信息】 电子科技大学 ， 测试计量技术及仪器， 2012， 博士

【摘要】 随着宽带无线通信技术的迅速发展，频带资源变得越来越紧张。为了提高频谱利用率，人们提出了多种高频谱利用率的调制和传输技术，采用这类技术所传输的信号具有宽频带、高峰均比等特点。受功放非线性特性的影响，宽带信号经功放放大后将产生严重的带内和带外失真，增大通信系统误码率，并干扰邻近信道，所以，宽带无线通信系统对功率放大器线性度的要求较高。功率放大器线性化技术已成为宽带无线通信中的一个关键技术，其中，数字预失真线性化技术是当前的一个研究热点。本文应用非线性系统理论、自适应信号处理等方面的知识，对数字预失真系统中的若干问题进行了深入分析和研究，提出了针对数字预失真的优化方法，可提高数字预失真的线性化性能。本文的主要贡献和创新包括：1.提出一种基于直接型和间接型学习结构的预失真学习结构。本文首先证明了功放系统与预失真系统的可交换性，从而证明了间接学习结构的合理性。针对间接学习结构易受系统噪声影响的缺点，本文提出一种基于直接型和间接型学习结构的预失真学习结构，该结构可降低反馈电路噪声所带来的影响，提高预失真的线性化性能。2.提出一种基于查找表的单反馈非迭代预失真方法。针对现有非迭代预失真方法计算量较大的缺点，提出一种结构更简单、计算量更低、精度更高的单反馈非迭代预失真方法，可降低预失真系统的硬件成本并简化系统设计。3.提出一种基于记忆多项式的单路反馈预失真线性化方法。传统预失真系统中反馈并采集回的IQ两路信号对处理器速度的要求较高，硬件开销大。本文提出一种单路反馈预失真方法，不但可以降低系统硬件成本，还能消除下变频通道的增益和相位不平衡失真，从而提高预失真的线性化性能。4.提出一种针对正交调制解调器的IQ不平衡误差估计和补偿方法。正交调制解调器的IQ不平衡误差会影响预失真器的设计，从而降低预失真的线性化性能。传统的IQ不平衡误差估计与补偿方法需要附加反馈电路，本文提出一种可以同时估计并补偿正交调制解调器IQ不平衡误差的方法，该方法不需要附加反馈电路，在降低系统硬件复杂度和成本的同时还提高了误差估计的精度。5.提出一种高精度的正交调制器非线性失真模型，以及一种可以同时补偿有记忆功放和非线性正交调制器的联合补偿模型。为满足后级功放的增益要求，正交调制器有时需要工作在接近最大输出的状态，此时正交调制器存在着IQ不平衡失真以及非线性失真。本文提出一种高精度的正交调制器非线性失真模型及补偿器模型；由于正交调制器和功放都在发射通道上，故可以将二者看作一个整体进行联合补偿，基于联合补偿方法和本文提出的正交调制器非线性失真模型，本文提出一种针对非线性正交调制器和有记忆功放的联合补偿模型，可降低预失真系统的复杂度。为验证上述方法，设计了一个数字预失真物理实验平台，利用该平台验证了上述预失真线性化方法的正确性。更多还原

【Abstract】 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.更多还原