【作者】 周陬；

【导师】 王宏远；

【作者基本信息】 华中科技大学 ， 信息与通信工程， 2007， 博士

【摘要】 近年来,大规模集成电路和移动通信技术的进步,促进了无线局域网和移动通信网络的发展,空间中各种无线电信号的模式、频率以及功能复杂化,且形式多样化,客观上要求无线电终端能够适应多模式、多频段、多功能环境。因此,无线电用户渴望能够在使用、规划和创建无线电过程中,事先认知并描述复杂空间的无线电状态(RKRL:Radio Knowledge Representation Language)(包括波达方向(DOA:Direction Of Arrival)、频率、功率、调制模式、信道编码、传输帧结构等),进而对这个复杂的状态空间进行合理的解析,从而作出理性选择。其中方向认知(DOA估计)是最基本的认知目标。如何在复杂环境下高效、精确和实时地实现无线电信号的方向认知引起了人们越来越多的关注。在实际应用中,分辨能力、精确度、可扩展性和实时性日益成为评价无线电测向方法的主要指标。空间谱估计方法作为阵列信号处理技术的一个新的发展方向,能够较好地满足无线电测向实用化的性能需求。本文围绕“软件体系结构无线电认知研究”课题,对空间谱估计的基础理论、实现方案和应用空间谱估计进行无源测向的分辨能力、精确度、空间扩展性以及实时性能进行了广泛深入的研究。本文通过对基于空间谱估计的无源测向技术的产生背景和主要特点的分析,给出了无源测向的基本模型,在全面深入地总结空间谱估计的研究现状和已有成果的基础上,确定将MUSIC算法和循环MUSIC算法作为深入研究的基本算法,并根据本课题的应用背景给出了无源测向的技术路线和基本方案。空间谱估计本身是一种超分辨方法。本文分析了空间谱估计测向的超分辨机理,在MUSIC算法和阵列孔径扩展的基础上提出了一种提高DOA分辨率的方法,在原始阵列的基础上虚拟地扩展阵列孔径(阵元间距),提高了DOA估计的分辨能力,通过对多个具有不同阵列孔径的虚拟阵列的空间谱取平均得到方向估计,成功抑制了虚拟扩展产生的虚假谱峰。仿真实验证明该方法相对于标准的MUSIC算法,具有更高的近目标分辨性能。在复杂环境下,循环MUSIC算法是一种比较先进的DOA估计算法,具有较好的抗干扰/噪声的性能。本文在分析循环MUSIC算法估计DOA的误差的基础上,注意到导向矢量是频率f和波达方向θ的联合估计,而循环MUSIC在选择估计导向矢量的频率时有偏差,本文选择更合适的频率,提出了一种改进循环MUSIC算法以减少偏差,提高了DOA估计的精确度。二维DOA估计是无源测向应用中的重要问题。本文介绍了二维DOA估计的基本概念,随后将标准MUSIC和循环MUSIC算法推广应用于立体直角坐标形式的二维DOA估计中,给出了基本的信号模型;在该模型的基础上应用扩展循环MUSIC算法,以提高DOA估计的近目标分辨能力;并针对直角坐标二维DOA估计计算量偏大和需要较多天线阵元的缺陷,提出改进方案,引入类似于多项式求根的运算方法代替谱峰搜索过程,可以在每个坐标轴方向上用少于信源数的天线阵元有效的估计DOA,也解决了谱峰搜索计算量偏大的问题,还进一步提高了近目标分辨能力。在理论研究的基础上,本文设计和实现了一个基于高速并行优化算法的DOA估计数据处理系统。描述了高速并行优化算法的设计思想,进行了可行性分析和计算机仿真,建立了系统模型和基本框架,给出了相关模块的定义和功能及实验系统的总体仿真和性能对比测试,证明该系统具有高效、简单和实用的特点。该系统基本实现了无源测向系统的核心功能,为无源测向的实用化开发奠定了基础。本文最后对无源测向技术未来的研究方向进行了展望。更多还原

【Abstract】 In recent years, the progresses of large scale integrated circuit and mobile communications have promoted the development of wireless local area network and mobile communication network, the various signal pattern , frequency and function of radio are made complicated in space, and the form diversifies, objectively, require that the radio terminal is able to adapt to many patterns, multifrequency, multifunctional environment. Therefore, in the process of using, planning and creating, the radio consumer longs to be able to predict and describ the states of complicated space radio signals using RKRL(Radio Knowledge Representation Language), include DOA(Direction Of Arrival), frequency, power, modulation modes, signal channel encode, frame structure, etc.. Then analyse on this complicated state space rationally, thereby make reasonal choices. Among them, direction cognition is the most fundamental target. How to realize the direction cognition of radio signals effectively, exactly and quickly, has attracted more and more attention. Researches on space array signal processing based radio direction finding methods have come to be hot spots all over the world. In the actual application, resolution response, precision, expansibility and real-time response have become the most important indexes to appraise the radio direction finding methods. Spatial spectral estimation method is a new development direction of array signal process technology which satisfies perfectly the applicative requirements of radio direction finding. Supported by the projects of“Software System Structure based Radio Cognition Studies”and“Location Technology Studies in Wireless Local Area Network and Mobile Communication”, this thesis has engaged in extensive research on some relevant basic theories, implementation methods of spatial spectral estimation and the resolution response, precision, expansibility and real-time response of radio direction finding method based on spatial spectral estimation.In this thesis, we analyze the background and features and give the models of radio direction finding technology based on spatial spectral estimation. Then select MUSIC and Cyclic-MUSIC algorithm as basic algorithm to study roundly based on the analysis and remark of status and problems remaining unsolved. In the meantime, give the technology and implementation method for radio direction finding based on the application background of the research. Spatial spectral estimation is a super-resolution technique. We analyze the super-resolution principle of spatial spectral estimation, present an improved super resolution method based on MUSIC algorithm for estimating the DOA of signals. The proposed method expands a real sensor array into several virtual arrays to enhance the resolution, each virtual array has a different aperture size which is bigger than the original array’s, and aperture expanding results in spurious peaks of spatial spectrum, then the proposed method averages the spatial spectrum of the virtual arrays to restrain the spurious peaks. Simulation results show the effectiveness of the method, and prove the method has a higher close-target resolution than standard MUSIC algorithm for incoherent signals.Cyclic-MUSIC algorithm as an advanced method for DOA estimation under complicated radio environment, it has much better anti-jamming and anti-noise performance than MUSIC algorithm. We analyze the bias of Cyclic MUSIC algorithm, explore the effects of carrier frequency, signal bandwidth, cyclic frequency, sensors distance, and DOA angle on the performance of DOA estimation. Then propose an improved cyclic MUSIC algorithm based on the result of the analysis, and reduces the bias substantially by properly choosing the frequency for evaluating the steering vector. Simulation results show the exactitude of the error analysis and the effectiveness of the improved algorithms.2D-DOA estimation is an important problem of radio direction finding application. We introduce the concept of 2D-DOA estimation, and then standard MUSIC and Cyclic-MUSIC algorithm are applied to 3D rectangular formed 2D-DOA estimation. Then we give the basic signal models; and apply Eextended-Cyclic-MUSIC algorithm for enhancing close-target resolution; to avoid spectral peak searching in angle domain, the approach is finding roots of complex function in z-domain, which is similar to root-MUSIC algorithm. As a result, the new algorithm improves the performance of estimating DOA greatly with less sensor requirement and less calculation. In the meantime, the method has a higher close-target resolution than Eextended-Cyclic- MUSIC algorithm.Based on the basic theory of direction finding techniques, a data process system based on high-speed parallel optimization algorithm is designed and developed. The thesis describes the design idea of high-speed parallel optimization algorithm, gives the feasibility analysis, and proves the feasibility by computer simulation. We establishe the system model and overall framework, gives the definition and functions of related modules, and give the integrated simulation and contrast performance test of the system, the system is proved to be efficient through the performance test. This prototype system realizes the core functions of direction finding system and establishes a solid foundation for putting direction finding products into practice.In conclusion, we propose the further development directions of direction finding system.更多还原