Doherty功放数字预失真线性化技术研究

发布时间：2018-06-13 06:41

本文选题：行为建模 + 数字预失真　；参考：《电子科技大学》2017年硕士论文

【摘要】：随着无线通信系统的基带带宽需求越来越宽,信号的峰均比也急剧增加,这就要求作为无线通信系统基础的射频功率放大器有较高的回退区且保持高效率。为了解决这些问题,出现了异相技术、包络跟踪和Doherty等功率放大器结构。其中Doherty因其结构简单、成本低、易于实现和便于改进等特点,成为当前功率放大器设计的主流。而与功率放大器通信质量息息相关的数字预失真技术也愈发受到重视。本文主要内容是对Doherty功率放大器的数字预失真技术进行研究。本文选取了1.8-2.6GHz的双带Doherty功率放大器作为研究对象,首先分析其非线性特性,然后对放大器行为模型和数字预失真算法进行分析讨论。同时,引入了一种基于记忆多项式模型(Memory Polynomial,MP)的改进模型--加权记忆多项式模型(Weighted Memory Polynomial,WMP),该模型可以应用于射频功率放大器的行为建模和预失真。加权记忆多项式模型在记忆多项式模型的静态和动态项上,引入了依赖于即时功率的权重函数,本文所用的权重函数是双曲正切函数。同时本文基于WMP模型,实现了WMP模型的多步迭代预失真器,可以获得更好的线性化效果。为了更好的验证本文提出的基于WMP模型的多步迭代数字预失真结构的有效性和优越性,作者利用实验室现有仪器搭建了验证平台,并对平台可能造成的非线性失真进行了讨论分析和校准。本文以20MHz LTE(长期演进技术)信号,对放大器模型的建模和预失真进行了实验验证,并以输出信号的ACPR(Adjacent Channel Power Ratio,相邻信道功率比)作为衡量功率放大器线性度的标准。将实验结果与传统的记忆多项式模型和广义记忆多项式模型进行对比从而评估新模型。结果表明,本文使用的双带Doherty功率放大器在未进行预失真处理时,在其1.9GHz和2.6GHz的ACPR分别为-26.08dBc/-28.67dBc、-28.88dBc/-28.87dBc,而经过基于WMP模型的多步迭代数字预失真处理后,ACPR至少可以改善26.19dBc,达到预期目标-52dBc以下。
[Abstract]:As the baseband bandwidth of the wireless communication system becomes wider and wider, the peak to average ratio of the signal increases rapidly, which requires the RF power amplifier, which is the basis of the wireless communication system, to have a high fallback zone and high efficiency. In order to solve these problems, heterogeneous techniques, envelope tracking and Doherty power amplifier structures have emerged. Among them, Doherty has become the mainstream of power amplifier design because of its simple structure, low cost, easy to implement and easy to improve. Digital predistortion technology, which is closely related to the communication quality of power amplifier, has been paid more and more attention. The main content of this paper is to study the digital predistortion technology of Doherty power amplifier. In this paper, a 1.8-2.6GHz dual-band Doherty power amplifier is chosen as the research object. Firstly, the nonlinear characteristics of the amplifier are analyzed, and then the behavior model of the amplifier and the digital predistortion algorithm are analyzed and discussed. At the same time, an improved memory polynomial model, weighted memory polynomial model, is introduced, which can be applied to the behavior modeling and predistortion of RF power amplifiers. In the static and dynamic terms of the memory polynomial model, the weighted memory polynomial model introduces the weight function which depends on the instant power. The weight function used in this paper is hyperbolic tangent function. At the same time, based on the WMP model, the multi-step iterative predistorter of the WMP model is implemented, which can obtain better linearization effect. In order to better verify the validity and superiority of the multi-step iterative digital predistortion structure based on WMP model proposed in this paper, the author has built a verification platform using the existing instruments in the laboratory. The possible nonlinear distortion caused by the platform is analyzed and calibrated. In this paper, the modeling and predistortion of amplifier model are verified by using 20MHz LTE (long term evolution technique) signal, and the linearity of power amplifier is measured by the ACPRG adjacent channel power ratio of the output signal. The experimental results are compared with the traditional memory polynomial model and the generalized memory polynomial model to evaluate the new model. The results show that the dual-band Doherty power amplifier is used without predistortion. At 1.9GHz and 2.6GHz, the ACPR is -26.08dBc / -28.67dBc / -28.88dBc / -28.87dBcrespectively. After multi-step iterative digital predistortion processing based on WMP model, ACPR can improve at least 26.19dBc and reach the expected target of -52dBc.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN722.75

【参考文献】