Principles of Mobile Communication Third Edition
Gordon L. Stüber著，为英文原文第三版。其第二版已经翻译成中文，即为移动通信原理（第二版），由电子工业出版社出版，但翻译水平很差。Gordon L. StuberPrinciples of mobileCommunicationThird edition② SpringerGordon L. StuberGeorgia Institute of TechnologyAtlanta gaUSAstuber @ece gatech. eduISBN978-1-4614-0363-0e-ISBN978-1-46140364-7DOI10.1007/978-1-46140364-7Springer New York Dordrecht Heidelberg londonLibrary of Congress Control Number: 2011934683C Springer Science+Business Media, LLC 2002, 2011All rights reserved. This work may not be translated or copied in whole or in part without the writtenpermission of the publisher(Springer Science+ Business Media, LLC, 233 Spring Street, New York,NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use inconnection with any form of information storage and retrieval, electronic adaptation, computer softwareor by similar or dissimilar methodology now known or hereafter developed is forbiddenThe use in this publication of trade names, trademarks, service marks, and similar terms, even if they arenot identified as such, is not to be taken as an expression of opinion as to whether or not they are subjectto proprietary rightsPrinted on acid-free paperSpringerispartofSpringerScience+businessMedia(www.springer.com)To Alana, our son nickolasmy father lothar Stuiberand my late mother Beatrice stiiberPrefacePrinciples of Mobile Communication, third edition, is a major revision of thesecond edition. Like its earlier editions, this book provides a mathematicallyrigorous overview of physical layer wireless communications. The basic pedagogicmethodology is to include fully detailed derivations from first principles. The text isintended to provide enough principle material to guide the novice student, whilit the same time having plenty of detailed material to satisfy graduate studentsinclined to pursue research in the area. The book is intended to stress the principlesof wireless communications that are applicable to a wide array of wireless standardsIt is intended to serve as a textbook and reference for graduate students, and a usefulreference for practicing engineersOrganization of the bookChapter 1 begins with an overview that is intended to introduce a broad array ofissues relating to wireless communications. Included is a brief description of theevolution of various wireless standards, the basic concepts of cellular frequencyreuse, the land mobile radio propagation environment, link budgets, and coverageand capacity of cellular radio systemsChapter 2 provides an extensive treatment of radio propagation, since goodunderstanding of the physical wireless channel is essential for the developmentand deployment of wireless systems. The chapter begins with a treatment ofthe narrow-band faded envelope for conventional fixed-to-mobile channels foundin cellular radio systems, mobile-to-mobile channels found in mobile ad henetworks, and multiple-input multiple-output (MIMO) channels where multipleantennas are used at both the transmitter and receiver to achieve high capacity afterconsidering the narrow-band channel, we consider the statistical treatment of wide-band channels. The emulation of wireless channels is essential for the developmentand testing of wireless systems, and the chapter provides a detailed discussion ofchannel simulation techniques. Finally, the chapter concludes with a discussion ofshadowing and path loss models for land mobile radio environmentsPreraceChapter 3 provides a detailed treatment of co-channel interference which is therimary impairment in spectrally efficient cellular frequency reuse systems. Veryoften the receivers in such systems are affected by multiple co-channel interferersand the probability distribution of the total interfering power is considered. Thchapter also considers the link outage performance due to co-channel interferencein a variety of wireless environmentsChapter 4 covers the various types of modulation schemes that are used in mobilecommunication systems along with their spectral characteristics. The chapter beginswith the mathematical representation of bandpass modulated signals, along withNyquist pulse shaping. Later, a large variety of modulation schemes used in wirelesssystems are considered, including both single-carrier and multi-carrier modulation,and both linear and nonlinear modulation techniques This is followed by a treatmentof the power density spectrum of modulated signals. Although quite mathematicalin nature, power spectrum is an important topic, since wireless systems are requiredto operate within a specified out-of-band emission maskChapter 5 discusses the error probability performance of various digital modulation schemes on narrow-band fat fading channels. The performance is evaluatedwith a variety of receiver structures, including coherent detectors, differentialcoherent detectors and noncoherent detectorsChapter 6 includes a treatment of multi-antenna techniques for combatingenvelope fading. The chapter includes a discussion of various diversity combiningtechniques for coherent, differentially coherent, and noncoherent receiver detectionof signals on fading channels with additive white gaussian noise. also consideredis optimal combining which is effective when the primary additive impairment isco-channel interference rather than noise. Finally, the chapter considers the use ofmultiple antennas at the transmitter in the context of classical beam-forming andtransmit diversityChapter 7 provides an extensive treatment of digital signaling on intersymbolinterference(ISI) channels that are typical of broadband land mobile radio systemsThe chapter begins with the characterization of Isi channels and goes on todiscuss techniques for combating ISI based on symbol-by-symbol equalization andsequence estimation. Later, error probability for maximum likelihood sequenceestimation is considered. The chapter concludes with a discussion of co-channeldemodulation for the purpose of mitigating co-channel interference on Isi channelsChapter 8 covers error control coding techniques for wireless systems. Thechapter begins with a discussion of basic block coding including space-time blockcodes Convolutional coding is considered next along with the Viterbi and BCJRalgorithms for decoding convolutional codes, followed by trellis coded modulationThe chapter then provides a detailed discussion on the design and performanceanalysis of convolutional and trellis codes for awgn channels, and interleaved fatfading channels, and fading isi channels. Later, space-time trellis codes are treatedand the chapter concludes with Turbo codinChapter 9 is devoted to spread spectrum techniques The chapter begins with anintroduction to direct sequence and frequency hop spread spectrum. This is followedy a detailed treatment of spreading sequences. Also included is a discussionPrefaceof the effects of tone interference on direct sequence spread spectrum, and theRAKE receiver performance on wide-band channels. The chapter wraps up witha discussion of cdma multiuser detectionChapter 10 is devoted to multi-carrier techniques. It considers the performanceof ofdm on frequency-selective channels and considers the effect of residualISI and problem of residual ISI cancellation. Later, the chapter examines singlecarrier frequency-domain equalization(SC-FDE)techniques. This is followed by atreatment of orthogonal frequency division multiple access (OFDMA)on both theforward and reverse links. The chapter concludes with a discussion of single-carrierfrequency division multiple acceSs (SC-FDMA)Chapter 11 considers frequency planning techniques for cellular systems. Thechapter begins with a discussion of cell sectoring, cell splitting, and reuse partition-ing. Later, the chapter considers radio planning for OFDMa cellular systems. Thisis followed by hierarchical overlay/underlay architectures based on cluster planningFinally, the chapter wraps up with macrodiversity TDMA cellular architecturesChapter 12 considers CDMa considers CDMA cellular systems, consideringtopics such as capacity and power control This is followed by a discussion ofhierarchical macrodiversity CDMa architectures and their performanceChapter 13 is devoted to cellular radio resource management. The chapter beginswith an introduction to basic hard and soft handoff. Later, the chapter considers theimportant problem of link quality evaluation, including signal strength averaging,velocity estimation, and velocity adaptive hard handoff algorithms later, a detailedanalysis of hard and soft handoff is provided. Finally, the chapter wraps up withmethods for estimating received carrier-to-interference plus noise ratio(CINR)The Appendix includes a brief and focused tutorial discussion of probabilityand random processes. A good understanding of the material in the Appendix isessential, since the concepts are widely used throughout the textUSing This Book for InstructionThe book has been developed from a graduate-level course on physical wirelesscommunications that I have taught at Georgia Tech since 1993. Normally, I prefera graduate-level course in digital communications as a prerequisite for this courseHowever, such a prerequisite may be waived to the extent that there is extensivebackground material in each chapter. A course may cover the introductory materialin each chapter and skip the more specialized material. In my own classes, I alwaystry to judge the mathematical level of the students early and adapt accordinglyThe book obviously contains far too much material to be taught in a onesemester course. However, i believe that it can serve as a suitable text in mostsituations through the appropriate instructor selection of background sections. Myown preference for a one semester course is to include the following material inorder: Chap. 1, Chap 2(skipping the more advanced material), and the first twosections of Chap 3. In moving to modulation waveforms in Chap. 4, an instructormay have to treat/assume basic signal-space representation. However, most students
- 2020-12-09下载
- 积分：1
晶圆缺陷检测与分类的卷积神经网络
晶圆缺陷检测与分类的卷积神经网络；针对晶圆检验时扫描电镜图像的缺陷检测和缺陷分类两问题，采用了“ ZFNet”的卷积神经网络来分类晶圆缺陷，并基于该分类器实现了一种“基于块的卷积神经网络”缺陷检测算法。为了提高准确率和加快速度，又改动“更快的区域卷积神经网络”实现了另一种检测算法。第卷第期邡鑫,史峥:晶圆缺陌检测与分类的卷积神经网络ZENet classifierDarker ImIn.ril” HumpBitel检测算法示意图在训练检测器时,数据集是检测器原始尺寸的图像,且包含标记好的缺陷区域和类型。我们结构通过·系列数据扩张操作,得到组数据,随机选取相比于检测算法主作为训练集,作为测试集。要从以下三方面进行了针对性的改进算法中需要优化的参数有滑动窗口尺寸滑()针对重复计算卷积的缺点,采用先动步幅、概率阙值、面积阙值,由于无法求出统一计算特征图,再按)进行映射各参数与检测结果的明确关系式,所以采用遍历法优化参截取的办法。如图,先通过卷积网络(数。因为检测到的缺陷尽量正确和尽量检测到所有缺陷是)对输入图像计算得到其特征图,因为在输入图像矛盾的,故以精确率和召回率的调和平均值作为优上的都能映射到特征图上,所以从输入图像上按化目标,也可根据实际需要调整两者权重满足不同侧重。割取图像进行卷积运算可以替代为直接从特征图上按测试结果映射后的范围割取,从而避免多次重复计算卷积。由于用训练好的检测模型对测试集检的大小形状不·,而全连接层的神经元连接数是固定的,测,计算模式下每张图大概耗时如果检测到的缺所以对割取得到的子特征图,通过层次采样到统陷与标准答案的且类型相同,则判为正确,否尺寸以连接到全连接层。则判为错误。得到结果如表,计算得:laut Image精确率Feature Map召回率ROI其屮正确缺陷的平均表检测器测试结果数量正确错误network有缺陷(正类)图映射示意图从检测结果来看该算法基本实现∫对图像上晶圆()针对滑动窗口尺寸单·的缺点,增缺陷的检测和分类,但是值较低,缺陷检测位置不加了滑动窗口的尺寸类型,并且增加由一个全卷积网络准确,检测耗时较长,分析其原囚如下)组成的()检测出错的数据中,缺陷较大的类型易判断错,)来预判断是否有缺陷。本文采用面积缺陷较小的容易被漏掉,说明只使用一种尺寸的滑动框很分别为,长宽比分别为、共难适应尺寸变化范围较大的缺陷种尺寸的滑动窗口,依次计算其中有缺陷的概率,再从中)滑动框步幅减小则算法耗时平方倍增加,而步幅筛选出一定数量最有可能有缺陷的区域,进行非极大值抑过长造成缺陷概率分布图分辨率较差,从而检测到缺陷位制(),最后得到一定数置准确度较差量的候选区域。()相邻滑动框都有大量重叠,所以每个区域都被多()针对缺陷检测位置准确度差的缺点,次重复送入计算卷积,导致算法耗吋较长。在全连接层后连接一个边界回归层在与上述检测算法相似的图像目标检测领域,近来出用来修正缺陷位置,该回归层与分类层并列。现的很好的克服了以上缺点并取得了很好的针对本文的缺陷检测问题,直接套用标准效果,所以下面介绍如何通过改动实现品圆并不能解决问题。因为判断晶圆的缺陷类型通常需缺陷的检测与分类。要结合缺陷区域周围的图形信息,而在预判断是否有C1994-2017ChinaAcademicJournalElcctronicPublishinghOusc.Allrightsrescrved.http://www.cnki.nct计算机工程年月日缺陷吋还进行了边界回归。虽然更加准确的给出缺陷的位()将原尺寸为的图像调整为置,但送入检测网络的特征儿乎不包含缺陷周围图肜信息,使得滑动窗口尺寸能够适应缺陷大小的变化范围,也可以导致缺陷分类不准。故木文对标准徹了一些根据实际情况来具体调整。改动:得到缺陷检测算法如图,卷积网络(()将改为只判断滑动窗口内是否有缺陷,而,)将输入图僚转换成多种特征图;根据不进行边界回归,也就是只计算所有滑动窗口有缺陷的概特征图从滑动窗口中选出最有可能存在缺陷的率,选取可能性最大的个,做非极大值抑制,再选出层根据特征图中抽取出对应特征组成特可能性最大的个进行检测。征向量;检测网络()根据特征向()将卷积层的尺寸加大为,加大感受野量判断缺陷类型,并进行边界回归;最后通过和概率),从而在判断滑动框內是否有缺陷吋能参阈值对候选缺陷进行过滤即可得到最终缺陷。考更多的周围信息。Detection NetworkonFolutionnl actorSoftmaxRuI Puling liver,e Prop卟 edMS+PrubilitessionInput Image 1024*1024Fully 10 dyercrectCcrvchrionalLaver size 747图检测算法示意图模型训练和平均值作为优化目标,并且使用相同的训练集和图中的检测算法也是基于架构实现,因为卷测试集积网络提取的特征类型对相似普遍有效,故其卷积网络的测试结果参数是直接迁移第章分类器的卷积层参数。但是用训练好的检测模型对测试集检测,和的参数则需要通过方法进行训练,标准计算模式下每张图大概耗时,采用相同判定标准,提供了分开和联合两种训练方式。为了节约得到检测结果如表(其中负类总数与表中总数不同是因时间,本文采用联合训练方式,并结合缺陷检测问题的实为同一张图屮可能检测到多个缺陷),计算得际情况调整超参数精确率在训练时,对每张输入图像,要计算的滑动窗口召回率数量庞大(种尺寸的滑动窗口,滑动步幅)。所以从中随机抽取个作为训练集,其中正例其中正确缺陷的平均负例,且正例占比不超过。分类器采用表检测器测试结果损失函数数量正确错误在训练时,设置提供个,从中随有缺陷(正类)机选取个作为训练集,其屮正例无缺陷(负类)负例,且正例占比不超过。另外设置学从结果来看该算法各方面都优于检测算习率分类器采用损失函数,而边界回法和值更高说明检测检测缺陷类型正确归采用函数。且位置准确,而且速度也大大提高(检测一张图像耗时从为了与检测算法对比,在最后通过遍历法缩小到)。如图为检测缺陷示例,共中标注了缺陷优化和概率阈值时,同样以精确率和召回率的调位置、类型和对应概率C1994-2017ChinaAcademicJournalElcctronicPublishinghOusc.Allrightsrescrved.http://www.cnki.nct邡鑫,史峥:晶圆缺陷检测与分类的卷积神经网络I I图检测结果示例图结束语而对图像上的缺陷检测和缺陷分类这两个问题,本文提出的改动后的检测算法能够精准、快速地从图像中检测出缺陷并同吋进行分类。得益于卷积神经网络良好的特征学习能力,该检测算法能够根据标记好缺陷位置和类型的数据自动学习特征,从而尽量避免人工千预,使算法具有较强的适应能力。参考文献徐姗姗刘应安徐昇基于卷积神经网络的木材缺陷识别山东大学学报工学版刘云杨建滨王传旭基于卷积神经网络的苹果缺陷检测算法电子测量技术江帆刘辉王彬等基于模型的图像识别计算机工程C1994-2017ChinaAcademicJournalElcctronicPublishinghOusc.Allrightsrescrved.http://www.cnki.nct
- 2021-05-06下载
- 积分：1
