【作者】 陈晓毅；

【导师】 姚庆栋；

【作者基本信息】 浙江大学 ， 通信与信息系统， 2005， 博士

【摘要】 软件无线电是1992年美国首次提出的一种实现无线通信的新的体系结构。它是一种用软件来实现物理层连接的无线通信设计,其基本概念是把硬件作为无线通信的基本平台,把尽可能多的无线通信及个人通信功能用软件实现。这样,无线通信新系统、新产品的开发将逐步转到软件的开发上来,而无线通信产品的价值也将越来越多地体现在软件上,这是无线通信领域继固定到移动、模拟到数字之后的第三次革命。软件无线电的理想是把天线上的信号或中频信号直接由高性能模数转换器数字化,再传送给终端,这样软件无线电电台就可在同一平台上,安装不同的软件来实现灵活的通信功能。但对于大多数民用系统来说,按目前的芯片技术,这还只是一个遥远的梦想。因此,如何的进行软件无线电系统的设计,软硬件的划分,数字处理器的设计成了软件无线电中关键的技术。根据软件无线电的应用环境及目前硬件水平,本文研究了数字调幅广播技术,它是为了解决模拟调幅广播中频谱利用率低、信号接收质量差等问题提出的,全球统一的技术标准采用了OFDM并行传输体制。本文在参加了浙江大学信息与电子工程学系SOC研究小组承担的国家863超大规模集成电路设计重大专项项目,参与开发了具有自主知识产权的DSP处理器-MD32。在此基础上研究了基于数字广播接收机的软件化设计。本文主要研究数字广播数据传播技术,比较了现行数字广播各种方案。整合各种现有OFDM同步技术,根据数字广播的特殊帧、导频结构作了相应改进,提出了频率、符号的粗、细、跟踪三步骤一种完整数字广播基带接收方案。 对数字广播系统接收机进行了软件化研究,提出了一种多模式,多频段数字广播的软件化接收机结构,并分析了算法特点和复杂度,根据算法/结构协同设计的方法提出了应用于无线接收处理的坐标转换、加法比较选择等专用指令用来提高处理器的性能。 研究了应用于软件无线电技术的数字信号处理器中三个关键部件的优化。通过逻辑和电路上的并行执行思想对译码级数据旁路进行电路优化。利用分散集中控制,减少驱动器件的方法对原来的中央控制器进行了优化。在算法上采用冗余二进制并行加法算法对四级流水的乘累加(MAC)进行了电路上的优化,使得在满足设计要求的前提下减少为三级流水。更多还原

【Abstract】 Software radio is first introduced in 1992 by USA and it is a new architecture for wireless communication. It is a wireless technology that uses software to realize the physical link. The basic idea is let the hardware as common platform, put more and more wireless arithmetic and personal function to software platform. So the developing of new wireless system and new wireless production will transform to software design, and the value of wireless communication will appearance in software development. This is the third revolution after fixed to mobile and analogy to digital transform.The ideal case of software radio is transform the signal from antenna or IF signal to digital using high performance A/D apparatus, and then send to personal terminal. The software terminal can receive different mode signal by using different program in the same platform. But most of system for common people is more complex than the ability of current integer ciruit(IC) for software realization, so this is just a good dream for us.In this way, how to design a good software system, how to depart the software from hardware and how to design a high performance DSP become the key technology in software radio. According to the ability of current DSP This thesis study a case for digital Radio mondial(DRM) system. DRM is developed to solve the analogy radio’s shortcoming. OFDM is adopted in global standard-DRM. This thesis is based on the work on China 863 program -32 bits digital signal processor design for SOC and discussed the software design in DRM system. This paper first research the electric wave circumstance in radio frequency band. Comparing different existing system for digital radio. Combining the several OFDM synchronized schemes and taking the special frame architecture into accord, the paper gives a complete frequency, time synchornize method that using 3 steps- coarse, fine estimation and track.Study the software architecture for DRM system, show a software receiver that combing multi-mode(analyog,digit) and mult-band(AM,FM).Analyse the main receiving algorithm-FFT, channel estimation and so on and the characteristic of those algorithm.The complexity of those is also discussed. In order to improve the DSP performance, some special instruc-tions(CORDIC,ACS) fit wireless communication process is adopted through algorithm/architecture Co-design.Optimize three key units in DSP that used in software radio, parallel execution of bypassing unit(BPU) with instruction decode(ID) in architecture level and transform serial select with priority to parallel execution are applied to reduce time delay. Separate the center control to some small disperse control units and decrease the main signal’s drive by sorting all interface registers into small ones to optimize the original pipeline control unit. Adopting redundant binary representation algorithm to transform the serial adder to parllel one that used in multiply-add instruction(MAC) to reduce delay. Reduce the original four stages pipline exection to three stages.更多还原