Feedback Consolidation Algorithms for ABR Point-to-Multipoint Connections in ATM Networks
关于物流的英语作文
关于物流的英语作文Title: The Evolution and Impact of Logistics in the Modern World。
Introduction。
Logistics plays a crucial role in the global economy, facilitating the movement of goods and services from producers to consumers efficiently and effectively. Over the years, logistics has evolved significantly, driven by technological advancements, globalization, and changing consumer demands. This essay explores the evolution and impact of logistics in the modern world.Historical Evolution。
The concept of logistics traces back to ancient civilizations, where armies organized supply chains to support military campaigns. However, modern logistics emerged during the Industrial Revolution when the need forefficient transportation and distribution systems became apparent. The invention of the steam engine revolutionized transportation, enabling goods to be transported over longer distances at a fraction of previous times.With the advent of the 20th century, logistics underwent further transformation with the introduction of assembly lines and mass production techniques. This period saw the rise of companies like Ford, which revolutionized manufacturing processes and supply chain management. The development of standardized containers in the mid-20th century revolutionized maritime transportation, further enhancing the efficiency of global trade.Technological Advancements。
DoS_攻击下具备隐私保护的多智能体系统均值趋同控制
DoS 攻击下具备隐私保护的多智能体系统均值趋同控制胡沁伶 1郑 宁 1徐 明 1伍益明 1何熊熊2摘 要 均值趋同是一种广泛应用于分布式计算和控制的算法, 旨在系统通过相邻节点间信息交互、更新, 最终促使系统中所有节点以它们初始值的均值达成一致. 研究拒绝服务(Denial-of-service, DoS)攻击下的分布式离散时间多智能体系统均值趋同问题. 首先, 给出一种基于状态分解思想的分布式网络节点状态信息处理机制, 可保证系统中所有节点输出值的隐私. 然后, 利用分解后的节点状态值及分析给出的网络通信拓扑条件, 提出一种适用于无向通信拓扑的多智能体系统均值趋同控制方法. 理论分析表明, 该方法能够有效抵御DoS 攻击的影响, 且实现系统输出值均值趋同. 最后, 通过仿真实例验证了该方法的有效性.关键词 多智能体系统, 均值趋同, 拒绝服务攻击, 隐私保护, 网络安全引用格式 胡沁伶, 郑宁, 徐明, 伍益明, 何熊熊. DoS 攻击下具备隐私保护的多智能体系统均值趋同控制. 自动化学报, 2022,48(8): 1961−1971DOI 10.16383/j.aas.c201019Privacy-preserving Average Consensus Control forMulti-agent Systems Under DoS AttacksHU Qin-Ling 1 ZHENG Ning 1 XU Ming 1 WU Yi-Ming 1 HE Xiong-Xiong 2Abstract Average consensus is a widely used algorithm for distributed computing and control, where all the nodes in the network constantly communicate and update their states in order to achieve an agreement. In this paper, we study the average consensus problem for discrete-time multi-agent systems under DoS attacks. First, a distributed network node state value processing mechanism based on state decomposition is given, which can ensure the pri-vacy of the output values of all nodes in the system. Then, through using the decomposed node state values and the network topology conditions given by the analysis, an average output consensus control law for distributed discrete-time multi-agent systems is proposed. Theoretical analysis shows that the proposed method can effectively resist the influence of DoS attacks on the system, and achieve the convergence of the average value of system initial outputs.Finally, numerical examples are presented to show the validity of the proposed method.Key words Multi-agent systems, average consensus, denial-of-service attack, privacy-preserving, cyber securityCitation Hu Qin-Ling, Zheng Ning, Xu Ming, Wu Yi-Ming, He Xiong-Xiong. Privacy-preserving average con-sensus control for multi-agent systems under DoS attacks. Acta Automatica Sinica , 2022, 48(8): 1961−1971多智能体系统是由多个具有一定传感、计算、执行和通信能力的智能个体组成的网络系统, 作为分布式人工智能的重要分支, 已成为解决大型、复杂、分布式及难预测问题的重要手段[1−2]. 趋同问题作为多智能体系统分布式协调控制领域中一个最基本的研究课题, 是指在没有协调中心的情况下, 系统中每个节点仅根据相互间传递的信息, 将智能体动力学与网络通信拓扑耦合成复杂网络, 并设计合适的分布式控制方法, 从而在有限时间内实现所有节点状态值的一致或同步.然而具备分布式网络特点的多智能体系统由于普遍规模庞大, 单个节点结构简单且节点地理位置分散等原因, 使得系统中易产生脆弱点, 这就使其在推广应用中面临两项基本挑战: 1)节点状态信息的隐私泄露问题; 2)节点或节点间的通信链路可能会遭受网络攻击的问题, 如欺骗攻击、拒绝服务(Denial-of-service, DoS)攻击等.针对节点状态信息的隐私泄露问题, 即在考虑多智能体网络趋同的同时, 保证系统中节点的初始状态值不被泄露, 已有较多研究人员开展相关的工作. 其中, 有学者借助于传统的安全多方计算方法,收稿日期 2020-12-09 录用日期 2021-03-02Manuscript received December 9, 2020; accepted March 2, 2021国家自然科学基金(61803135, 61873239, 62073109)和浙江省公益技术应用研究项目(LGF21F020011)资助Supported by National Natural Science Foundation of China (61803135, 61873239, 62073109) and Zhejiang Provincial Public Welfare Research Project of China (LGF21F020011)本文责任编委 鲁仁全Recommended by Associate Editor LU Ren-Quan1. 杭州电子科技大学网络空间安全学院 杭州 3100182. 浙江工业大学信息工程学院 杭州 3100231. School of Cyberspace, Hangzhou Dianzi University, Hang-zhou 3100182. College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023第 48 卷 第 8 期自 动 化 学 报Vol. 48, No. 82022 年 8 月ACTA AUTOMATICA SINICAAugust, 2022例如Yao等[3]提出混淆电路算法, Shamir等[4]提出秘钥共享算法等. 然而这类通用的隐私保护方法因计算和通信消耗较大, 不适用于单个智能体节点结构较为简单的分布式系统, 尤其是受到硬实时约束的一类多智能体系统应用. 如上述的混淆电路的计算延迟为秒级[5], 而对于多智能体系统一些典型应用如多无人飞行器编队的实时控制, 其容许的计算延迟仅为毫秒级[6]. 针对多智能体系统均值趋同过程中节点信息泄露问题, 有研究人员提出了一系列专门的隐私保护策略[7−10]. 这些方法大多基于模糊处理的思想, 即通过加入噪声来掩盖真实的状态值.其中一种常用的手段是差分隐私方法[11], 然而这种差分隐私下的模糊处理方法会影响最终趋同值的精度, 即使系统无法收敛到精确的节点初始状态的平均值. 最近文献[12]提出的一种基于相关噪声混淆技术的改进方法, 克服了传统差分隐私方法中精度下降的问题, 但却需要较多的算力. 最近的文献[13]采用一种基于状态分解的方法, 将每个节点的初始状态分解为两个随机的子状态, 只令其中一个子状态参与相邻节点间的信息交互, 而另一子状态保留在本节点内部, 不参与邻居间信息传递. 只要两个随机子状态的和满足特定条件, 在作者所设计的趋同算法下, 系统能够达成均值趋同, 且保护每个节点的状态信息不被泄露.此外, 有学者研究基于可观测性的方法用来保护多智能体系统中节点的隐私[14−16]. 基本思想是设计网络的交互拓扑结构以最小化某个节点的观测性, 本质上相当于最小化该节点推断网络中其他节点初始状态的能力. 然而, 这类基于可观测性的方法仍然存在隐私泄露的风险. 为了提高对隐私攻击的抵御能力, 另一种常见的方法是使用加密技术.然而, 虽然基于密码学的方法可以很容易地在聚合器或第三方[17]的帮助下实现隐私保护, 例如基于云的控制或运算[18−20], 但是由于分散密钥管理的困难,在没有聚合器或第三方的情况下, 将基于密码学的方法应用到完全分散的均值趋同问题是很困难的.同时, 基于密码学的方法也将显著增加通信和计算开销[21], 往往不适用于资源有限或受硬实时约束的分布式网络控制系统.以上的工作均是在安全的通信环境下完成的,然而在实际应用场景中, 由于物理设备和通信拓扑结构都有可能遭受网络攻击, 导致以往有关多智能体系统趋同研究的失效, 这使得针对多智能体系统在网络攻击下的趋同研究发展迅速, 并取得了一些显著成果[22−26]. 目前多智能体系统中常见的网络攻击主要有两种形式: 欺骗攻击[22, 25, 27−28]和DoS攻击[29−33].r其中DoS攻击是多智能体系统中最常见也是最容易实现的攻击形式, 只要攻击者掌握系统元器件之间的通信协议, 即可利用攻击设备开展干扰、阻塞通信信道、用数据淹没网络等方式启动DoS攻击.在DoS攻击影响下, 智能体间交互的状态信息因传递受阻而致使系统无法达成一致. 近年来, 研究者们从控制理论的角度对DoS攻击下的系统趋同问题进行了研究. 其中, 有研究人员通过构建依赖于参数的通用Lyapunov函数设计一种趋同方法[31],使其能够适用于因通信链路存在随机攻击导致通信拓扑随机切换的情况. 此外, 有研究者通过设计一个独立于全局信息的可靠分布式事件触发器[32], 很好地解决了大规模DoS攻击下的一致性问题. 更有研究者开始研究异构多智能体系统在通信链路遭受攻击时的趋同问题[33], 通过设计基于观测器的控制器, 实现在通信链路存在DoS攻击时两层节点间的趋同问题. 而在本文中, 考虑多智能体之间通信链路遭受DoS攻击的情况, 通过攻击开始时刻与攻击链路矩阵刻画DoS攻击模型, 通过增强网络拓扑以满足所谓的-鲁棒图来刻画信息流的局部冗余量[34],从而抵御DoS攻击的影响.然而, 针对趋同问题, 将网络攻击和隐私保护两者结合起来考虑的研究还鲜有见文献报道. 2019年Fiore等[24]率先开展了同时考虑隐私保护和网络攻击的研究工作, 但所得成果仍存在一定的局限性: 1)所提方法虽能保护节点隐私且最终达成状态值趋同, 却无法确保系统达成均值趋同; 2)作者仅考虑了欺骗攻击下的控制器设计问题, 因此所得结论并不适用于网络中存有DoS攻击的系统.y基于上述观察与分析, 本文主要致力于研究DoS 攻击下具备节点信息隐私保护的多智能体系统均值趋同问题, 从而补充现有趋同算法的相关结果. 同时, 考虑实际环境对测量条件等的限制, 不易直接获取节点的真实状态值[35], 为此本文围绕节点的输出值, 即通过观测矩阵获取的系统输出, 进行趋同控制器的设计工作. 本文的主要贡献包括:1)针对DoS攻击在多智能体系统分布式协同控制中的攻击特性和发生范围, 及对网络拓扑连通性的影响, 建立相应数学模型;2)针对一类DoS攻击下的无向通信网络多智能体系统, 提出一种基于状态分解的节点信息隐私保护策略. 当满足特定条件时, 所提策略可确保系统输出状态不被窃听者准确推断出来;y3)针对DoS攻击的影响, 分析给出了系统中节点通信拓扑的鲁棒性条件, 并据此设计一种基于输出量测值的分布式控制方法, 理论分析并证明1962自 动 化 学 报48 卷系统可容忍特定数目的链路遭受DoS 破坏, 并实现输出均值趋同.本文内容结构为: 第1节介绍本文所需要用到的图论知识, 网络拓扑图的相关性质以及均值趋同算法; 第2节主要对DoS 攻击模型和拟解决问题进行描述; 第3节提出系统在DoS 攻击下的隐私保护均值趋同控制方法, 并分别对在攻击下的网络拓扑鲁棒性、系统收敛性以及隐私保护能力进行分析;第4节通过一组仿真实例验证算法的有效性; 第5节是总结与展望.1 预备知识1.1 图论知识M G =(V ,E ,A )V ={v 1,v 2,···,v M }E ⊂V ×V A =[a ij ]∈R M ×M (v j ,v i )∈E a ij >0a ij =0(v j ,v i )∈E (v i ,v j )∈E a ij =a ji a ii =0v i N i ={v j ∈V|(v j ,v i )∈E}G L =D −A 考虑由 个智能体组成的多智能体系统, 节点之间为双向传递信息, 其通信网络可抽象地用一个无向加权图 表示. 其中 表示节点集合, 表示边集. 两个节点之间的连接关系用邻接矩阵(权重矩阵) 表示, 如果 , 则 ; 否则 . 在无向图中, 邻接矩阵是对称的, 即如果, 则同时有 , 且 . 本文不考虑节点自环情况, 即令 . 节点 的邻居集合表示为 . 无向图 对应的Laplacian 矩阵为 , 其中D 为度矩阵, 定义为:除了上述无向图的基本知识, 本文的研究工作还用到了r -可达集合和r -鲁棒图的概念. 这两个概念最早由文献[36]提出, 随后被文献[22, 27]等利用并扩展, 主要用于分析节点间拓扑抵御网络攻击的鲁棒性. 经笔者少许修改, 具体定义如下:G =(V ,E )S ⊂V S v i N i \S r S 定义1[36]. r -可达集合: 对于图 及其中一非空子集 , 如果 中至少有一个节点 在集合 中有不少于 个节点, 则称 为r -可达集合.G =(V ,E )V S 1,S 2⊂V S 1∩S 2=∅G 定义 2[36]. r -鲁棒图: 对于图 , 如果对 中任意一对非空子集 , , 保证至少有一个子集为r -可达集合, 则称 为r -鲁棒图.以下是一些关于r -鲁棒图的基本性质.G =(V ,E )ˆGG s (s <r )ˆG(r −s )引理1[22]. 考虑一个r -鲁棒图 , 令 表示 中每个节点至多移除 条边后的图,则 是一个 -鲁棒图.G G 引理2[22]. 对于一个无向图 , 如果 满足1-鲁G 棒图, 则有 为连通图.1.2 均值趋同算法M x i [k ]∑Mi =1x i [0]/M 考虑有 个节点组成的无向加权多智能体系统. 为了让系统实现均值趋同, 也就是所有节点的状态 最终收敛到它们初始状态的平均值, 根据文献[13, 37], 其节点动态更新方程可设计为:x i [k ]v i k ε∈(0,1/∆)∆式中, 为节点 在 时刻的状态值, 为系统增益系数, 通常定义为:η>0η≤a ij <1文献[38]表明, 当系统拓扑满足连通图, 且存在 使得 时,系统可在更新规则(1)下实现均值趋同, 即:2 问题描述M 本文研究对象为如下 个智能个体组成的一阶离散时间多智能体系统, 其动力学模型为:x i [k ]∈R N u i y i [k ]∈R Q y i [k ]nC i ∈R Q ×N n n ∈R +式中, 为系统的状态值, 为控制输入, 为系统经通信链路传输得到的量测信号, 需要注意的是, 由于通信链路中存在DoS 攻击, 可能遭受影响而无法被邻居节点接收到. 为观测矩阵, 其中 为从观测矩阵中抽取出的系数, 为大于0的正实数.2.1 攻击模型本文所讨论的DoS 攻击表现为某种传输尝试失败的情况[39], 其存在于多智能体系统中各智能体之间的通信链路中, 即当通信图中两个节点间的链路发生DoS 攻击时, 其通信链路将会被切断, 此时两个节点无法通过该链路进行信息交互, 进而达到攻击多智能体系统的目的. 在多智能体系统分布式协同控制中, 运载节点输出量测值的通信链路遭遇DoS 攻击的示意图如图1所示.(P,k 0)P =[p ij [k ]]∈R M ×M v i v j k 本文以Adeversory 刻画系统遭遇DoS攻击的情况. 其中 表示攻击状态矩阵, 当节点 和节点 之间在 时刻发生DoS8 期胡沁伶等: DoS 攻击下具备隐私保护的多智能体系统均值趋同控制1963p ij [k ]=0p ij [k ]=1k 0攻击时, ; 否则 . 为系统遭遇DoS 攻击的开始时刻.考虑攻击者资源的有限, 本文假设攻击发生范围满足f -本地有界[22]的定义, 该假设在文献[22−23,25]中被广泛采用. 结合DoS 攻击, 具体定义如下:f 定义3 (f -本地有界DoS 攻击) . 对于系统中的任一节点, 如果与其相邻节点的通信链路中, 任意时刻遭遇DoS 攻击的链路条数至多不超过 条, 则称此类攻击模型为f -本地有界DoS 攻击.2.2 系统假设(P,k 0)结合上述给出的Adeversory 和攻击发生范围模型, 本文对所研究的系统作出如下假设:f 3v i ∈V k 假设1. 系统中任意一个节点的通信链路中在任意时刻至多有 条链路同时遭受DoS 攻击, 即满足定义 攻击模型. 具体地, 则对于任意 , 在任意时刻 , 都有下式成立:G [k ]=(V ,E [k ],A [k ])虽然本文考虑的是固定无向拓扑, 但在DoS攻击影响下, 可以看到系统的通信图却会与之发生变化. 因此, 本文接下去用时变图符号 表示系统在DoS 攻击影响下的真实通信情况.η0<η<1i,j ∈{1,···,M }a ij [k ]>0η≤a ij [k ]<1假设2. 存在一个标量 满足 , 对于所有的 , 如果 , 那么 .x i ∈R N X i ∈R N X =∩M i =1X i X =∅假设3. 系统任意节点状态值 受限于一个非空闭凸集, 表示为 , 令 ,则 .根据上述假设, 可以得出系统具备如下属性:引理3[38]. 当系统的网络通信图为有向连通图v i ∈V (1)且邻接矩阵为双随机矩阵时, 并且满足假设2和3时, 那么对于系统中任意节点 在动态更新式 下, 有:{h [k ]}式中, 为一个定义的辅助序列, 对于每个时根据文献[38], 因邻接矩阵为双随机矩阵, 由式(7) ~ (8)可得:v i ∈V 引理4. 当系统的网络通信图为无向连通图, 并且满足假设2和3时, 那么由引理3可知, 对于系统中任意节点 在动态更新式(1)下, 式(10)仍然成立.证明. 根据引理3可知, 在网络通信图为有向图情况下, 邻接矩阵为双随机矩阵表明在该网络通信图中, 所有节点通信链路满足出度等于入度的条件, 而在无向图中, 该条件同样成立, 因此在无向图中, 式(10)仍然成立. □针对上述建立的网络攻击模型和相关的系统假设, 本文的研究目标是, 设计一种控制策略, 使得:1)系统的输出达到趋同并且趋同值是等于所有智能体初始输出状态的平均值; 2)在整个趋同过程中保护每个节点的信息值隐私.3 控制器设计3.1 DoS 攻击下网络拓扑鲁棒性条件首先对网络通信链路图的鲁棒性条件进行讨论, 以便于开展后续控制器的设计工作.引理5. 考虑多智能体系统(4), 如果其网络拓图 1 DoS 攻击下的多智能系统框图Fig. 1 The diagram of the multi-agent systemunder DoS attacks1964自 动 化 学 报48 卷(f +1)扑结构满足 -鲁棒的无向图, 那么系统在遭受f -本地有界DoS 攻击下, 即满足假设1, 其通信图仍可保持连通性.f 证明. 根据假设1可知, 网络中每个节点任意时刻至多有 条通信链路遭受DoS 攻击破坏. 再由引理1可知, 此时网络拓扑结构至少是1-鲁棒图.最后由引理2可知, 系统网络拓扑仍然能够保持连通性. □3.2 DoS 攻击下隐私保护控制上述小节给出了系统遭受DoS 攻击下通信网络仍旧保持连通的条件, 接下去本小节给出本文核心的控制器设计方法.x i x αi x βi x αi [0]x βi [0]x αi [0]+x βi [0]=2x i [0]受文献[13]启发, 此处引入状态分解方法: 将每个节点的状态值 分解成两个子状态, 用 和 表示. 值得注意的是, 初始状态的子状态值 和 可在所有实数中任取, 但需满足条件: .x αi x i v i x βi x αi x βi v i v 1x α1x 1x β1v 1x α1x αi x βi a i,αβ[k ]a i,αβ[k ]η≤a i,αβ[k ]<1为便于理解, 本文以5个节点的无向连通图为例, 通信拓扑如图2所示. 从示例图中可以看出: 子状态 充当原 的作用, 即与邻居节点进行信息交互, 并且实际上是节点 的邻居节点唯一可以获知的状态信息. 而另一个子状态 同样存在于该分布式信息交互中, 但是其仅与 进行信息交互. 也就是说子状态 的存在, 对于节点 的邻居节点是不可见的. 例如, 在图2(b)中, 节点 中的 相当于图2(a)中 的角色和邻居节点进行信息交互,而 仅对节点 自身可见, 而对其他节点不可见.但是它又可以影响 的变化. 两个子状态 和 之间的耦合权重是对称的, 表示为 , 并且所有的 满足 .基于上述方法, 本文给出具体的具备隐私保护的输出均值趋同控制协议:并且I L ′[k ]式中, 为单位矩阵, 为DoS 攻击下的Lapla-cian 矩阵,其满足:A ′[k ]=[′]式中, DoS 攻击下系统对应的邻接矩阵为D ′[k ]A ′[k ] 为对应于邻接矩阵 的度矩阵.y [k ]=nCx α[k ]C 另外, 在协议(11)中, 为系统的状态输出方程, 为输出方程的观测矩阵, 定义为:e i R M i i 式中, 表示 中第 个规范基向量, 该向量中第个位置数为1, 其他位置数为0.n ∈(0,1)n =1n ∈(1,∞)注1. 考虑实际环境中不同情况, 当 时, 系统输出方程将会缩小状态值进行信息交互,适用于节点状态值过大的情况; 当 时, 系统状态输出方程将会输出原本节点需要进行信息交互的状态值; 当 时, 系统状态输出方程将会放大状态值进行信息交互, 适用于节点状态值过小的情况.x α[k ]值得注意的是, 对于系统中的节点, 用于和邻居节点进行信息交互的状态值 是无法被邻居节点获取的, 需通过系统状态输出方程传递给邻居图 2 5个节点组成的示例图Fig. 2 Example of network with 5 nodes8 期胡沁伶等: DoS 攻击下具备隐私保护的多智能体系统均值趋同控制1965x α[k ]y [k ]节点. 简言之每个节点经过信息交互接收到的邻居节点的值并不是 , 而是经过输出方程输出的 .A αβ[k ]v i ,i =1,2,···,M x αi [k ]x βi [k ]a i,αβ[k ]令 为每个节点 的和两个子状态之间的耦合权重N =1,Q =1为便于叙述, 本文考虑节点的状态值及输出值为一维的情况, 即令 . 从而, 基于输出状态值的控制协议可表示为:事实上, 只要向量状态中的每个标量状态元素都有独立的耦合权重, 本节所提出的控制方法所有分析及结果同样适用于向量状态的情况.()ε1/∆1/(∆+1)注2. 与文献[13, 37]的更新式(1)相比, 本文给出的协议(19)中, 由于每个可见子状态的邻居数增加了一个 不可见子状态 , 因此 的上限从 降低为 .注3. 相比于文献[13, 37]设计的更新式(1),本文在协议(19)的设计过程中考虑了系统通信链路中存在DoS 攻击的情况, 可确保在存在一定能力DoS 攻击时, 系统在协议(19)的约束下实现均值趋同.3.3 输出均值趋同分析在给出本文主要结论前, 需要下述引理知识.引理6. 考虑多智能体系统(4), 如果其网络通信图是一个无向连通图, 则对于状态分解后的网络,所有节点子状态总和是固定不变的.y i [k ]=nx αi [k ]证明. 由输出方程 , 推导可得:再将式(20)代入式(19), 可得:进一步, 由式(21), 可得:因此有:∑M i =1{∑Mj =1a ′ij[k ](y j [k ]−y i [k ])}而在式(23)中的部分, 可进一步分解为下式:a ′ij [k ]=a ′ji [k ]v i ,v j ∈V 根据无向图属性: , 对于任意 , 有:将式(25)代入式(24), 可得:1966自 动 化 学 报48 卷(26)(23)将式 代入式 , 可得:由式(27)容易看出, 对于进行状态分解后的网络, 系统节点子状态的和是固定不变的. □下面给出本文的主要结论.(f +1)定理1. 考虑DoS 攻击下多智能体系统(4), 在满足假设1、2和3条件下, 若其通信拓扑满足 -鲁棒图, 且系统节点在所给的分布式协议(19)下进行状态更新, 则系统可实现输出值均值趋同.(f +1)证明. 由于系统的通信图是一个 -鲁棒图, 根据引理5可知, 系统在满足假设1的DoS 攻击下, 其网络图仍能够保持连通. 显然, 经过状态分解之后的系统同样能够保证网络图的连通性. 根据x αi [k ]x βi [k ]随后, 根据引理4和式(28)可知, 系统可以实现均值趋同, 即任意节点的子状态 和都x αi [0]+β再根据式(28)和状态分解约束条件y i [k ]=nx αi [k ]最后, 根据式(29)和输出方程 , 可得: □y 注4. 相比于文献[13]设计的隐私保护状态更新协议, 本文在协议(19)的设计过程中进一步考虑了在实际环境对测量条件等的限制导致难以获得系统中节点的真实状态值的情况, 引入了节点输出值的概念, 通过观测矩阵获取的系统输出 进行协议(19)的设计, 可确保系统在该协议下实现输出值均值趋同.3.4 隐私保护分析本节对趋同控制过程中单个节点信息的隐私保护进行分析. 本文考虑两种隐私窃听者: 好奇窃听者和外部窃听者. 好奇窃听者是指一类能够正确遵循所有控制协议步骤但具有好奇性的节点, 这类节点会收集接收到的数据并试图猜测其他节点的状态信息. 而外部窃听者是指一类了解整个网络拓扑结构的外部节点, 并能够窃听某些内部节点的通信链路从而获得在该通信链路交互的信息.一般来说, 这里的外部窃听者比好奇窃听者更具有破坏力, 因为外部窃听者会窃听多个节点通信链路上交互的信息, 而好奇窃听者只能窃听该节点通信链路交互的信息, 但好奇窃听者有一个外部窃听者无法得知的信息, 即该好奇窃听者的初始状态值.v i ∈V k I i [k ]={a ′ip [k ]|v p ∈N i ,y p [k ]|v p ∈N i ,x i [k ],x αi [k ],x βi [k ],a i,αβ[k ]}v i I i =∪∞k =0I i [k ]定义好奇窃听者 在第 次迭代时所获得的信息为: . 随着状态值迭代更新, 窃听者 收集获得的信息表示为 .x i [0]v i 定义4. 如果窃听者无法以任何精度保证估计节点状态信息 的值, 则称节点 得到了隐私保护.在给出结论前, 需要用到下述引理.v j v i v m v j x j [0]=x j [0]v i I i =I i 引理7[13]. 在采用状态分解方法的信息交互通信中, 如果正常节点 具有至少一个不与好奇窃听节点 直接相连的正常邻居节点 , 则对于节点 的任意初始状态 , 窃听节点 获得的信息始终满足 .v j v m a jm [0]v j a jm [0]a j,αβ[0]a m,αβ[0]v j x j [0]引理8[13]. 在采用状态分解方法的信息交互通信中, 如果正常节点 存在至少一个正常邻居节点, 其 的值对于外部窃听者不可见, 则节点 的任意初始状态的任何变化都可以完全通过对外部窃听者不可见的 , 和 的变化来补偿, 因此外部窃听者无法以任何精度保证估计正常节点 的初始状态值 .v j ∈V v j x j [0]定理2. 考虑DoS 攻击下多智能体系统(4), 对于系统中任意正常节点 , 如果 在所给的分布式协议(19)下进行状态更新, 则在整个信息交互过程中, 其状态信息值 具备隐私保护.v i v j x j [0]=x j [0]I i =I i v j v j x j [0]证明. 首先, 分析系统存在好奇窃听者 的情况. 对于任意正常节点 , 在所给的分布式协议(19)下, 其初始状态显然满足 , .再由引理6可知, 该条件下好奇窃听者无法准确估计节点 的初始值, 因此节点 的状态值 得到了隐私保护.v j ∈V v j v j 随后, 分析系统存在外部窃听者的情况. 在本文所提的分布式算法(19)下, 外部窃听者对于系统中任意正常节点 的其中之一子状态不可见. 根据引理7, 初始状态值的变化则对于外部窃听者不可见, 故外部窃听者无法准确估计正常节点 的8 期胡沁伶等: DoS 攻击下具备隐私保护的多智能体系统均值趋同控制1967。
个人简介(Biography)的编写及翻译
在投寄论文时，有些刊物要求作者在论文后面附上“个人简介”(biography)，内容包含何时何校何专业毕业，获得何学位，以及主要工作经历（学术活动）等。
下面是7个“个人简介”(biography)范例，撰写个人简历时常用单词(词组)以粗体标记，供读者参考。
范例1：Maryam Kolahdoozan received her MSc of applied science, specializing in concrete materials, from Ryerson University, Toronto, ON, Canada. Her areas of research include sustainable alternatives for producing unshrinkable fill and deterioration mechanism of concrete exposed to internal sulfate attack and method for mitigation.范例2：Douglas F. CRICKNER was born at McAlpin, a southern West Virginia coal camp, and had his first experience in coal mining at age 18 at the McAplin mine where for three successive summers he was employed on conveyor sections in 34-in. coal to drag pans and supplies, hang canvas, etc. He graduated from Virginia Polytechnic Institute with a degree in mining engineering in 1941. Following this, he served with Koppers Coal Co. and The New River Co. as mining engineer and assistant mine foreman. In 1946, he was employed by his present company, Pocahontas Land Corp., which he has served as mine inspector, chief engineer, general manager, and currently as vice president. During one four-year period he was transferred to an associated company, Norfolk & Western Railway Co., as superintendent of mines at its Pond Greek Colliers in Pike County, KY. He has taken an active part in a number of engineering and coal institutes. He is a past chairman of The Society of Mining Engineers of AIME Coal Division and is currently serving as a director ofSME and vice president-AIME Easter Region.范例3：M. E. HOPKINS received his education in geology from the University of Arkansas, then earned his Ph.D. at the University of Illinois in 1957. He served in the Army Air Corps in 1946-47. In 1951, he joined the Coal Section of Illinois State Geological Survey as a part-time research assistant. In 1955, he joined the faculty of the Department of Geology at the University of Tulsa but in 1963 returned to the Illinois Survey, becoming head of the Coal Section in 1968 where he served until 1975. He was vice president of Harry Williamson, Inc., Benton, IL, and is now director of geology, Peabody Coal Co., St. Louis, MO.范例4：H. L. WASHBURN received a BS and MS in mining engineering from the University of Kentucky. His first job was with the Research and Development Division, Consolidation Coal Co. in Pittsburgh, PA, where he worked for three years on a variety of projects including the development of pipeline transportation of coal. He was transferred to Fairmont, WV, as manager of preparation, and subsequently became chief engineer of the Mountainer Division of Consolidation Coal Co. His next assignment was a three-year period with Clinchfield Coal Co. as director of preparation, which involved the actual management of all the plants of Clinchfield. He went to North American Coal Corp. in 1966 as assistant vice president and was later promoted to vice president-engineering and senior vice president-operations, his present position.范例5：Jack A. SIMON first worked on coal geology for the Illinois State Geological Survey as a student in 1937. After serving in the Army Air Corps during World War II, he returned to the Survey on a full-time basis. He was head of the Coal Section, 1953-1967, after which he became head of the Geological Group and principal geologist. Since 1974 he has been chief of the Illinois State Geological Survey. Professional activities have included not only coal geology but, in recent years, environmental problems associated with mining transportation and use of fossil fuels.范例6：Lu Wang has been a Preceptor in the Department of Chemistry and Chemical Biology since 2013. She is involved in teaching several courses at Harvard including Physical Sciences 10 and 11, and Chemistry 301hf -- a teaching practicum course for chemistry graduate students. Before becoming a Preceptor, she held a two-year postdoctoral appointment at Massachusetts Institute of Technology. In addition to her research on two-dimensional nanobiosensors, she also participated in developing and teaching a college chemistry curriculum in the founding year of Singapore University of Technology and Design, which is established in collaboration with MIT. Dr. Wang received her Ph.D. in Chemistry from Harvard University in 2011, where she worked in Professor Charles Lieber‘s group to develop label-free silicon nanowire field-effect biosensors. She received her B.S. in Chemistry from Peking University in 2006.范例7（注重学术方面）：Navin Khaneja(Gordon McKay Professor of Electrical Engineering) is broadly interested in the area of mathematical control theory, signals and systems. His current research lies at the interface ofcontrol and information theory and physics. He is working on developing geometric techniques for optimal control of quantum mechanical phenomenon.The work has proven promising for optimal pulse sequences in high-resolution nuclear magnetic resonance spectroscopy. These optimal pulse sequences minimize the effects of decoherence and maximize the sensitivity of NMR experiments. Optimal pulse design can lead to significant reduction of time required for structural analysis of proteins. The work also has immediate applications to the areas of quantum information and computing.Professor Khaneja is also very interested in the areas of robotics, computer vision, statistical inference and image understanding. His current work in the area of robotics involves design of feedback controllers for stabilization of nonholonomic control systems with applications to locomotion systems.His recent work in Medical Imaging and Human Brain Mapping develops rich class of probabilistic models to capture inherent variability present in the anatomies and involves design of computationally efficient Bayesian inference algorithms for extracting anatomical information from noisy data collected through various imaging modalities.(Education: 1.B.Tech., 1994, Electrical Engineering, IIT, Kanpur; 2.M.A./M.S., 1996, Mathematics/Electrical Engineering, Washington University; 3.S.M., 1999, Applied Mathematics, Harvard University; 4.Ph.D., 2000, Applied Mathematics, Harvard University)BiographyA biography narrates the life story of a person, as written by another person or writer. It is further divided into five categories:Popular biographyHistorical biographyLiterary biographyReference biographyFictional biographyMemoirThis is a more focused writing than an autobiography or a biography. In a memoir, a writer narrates the details of a particular event or situation that occurred in his or her lifetime.Examples of Biography in LiteratureExample #1: Shakespeare: A Life (By Park Honan)This biography is the most accurate, up-to-date, and complete narrative ever written about the life of William Shakespeare. Park Honan has used rich and fresh information about Shakespeare in order to change the perceptions of readers for the playwright, and his role as a poet and actor.This book completely differs from other biographies that imagine different roles for him, commenting on his sexual relationships and colorful intrigues. Though detailed psychological theories and imaginative reforms about the famous playwright could be amusing, in fact, they damage the credibility of the sources. Therefore, many attempts have been made to know about Shakespeare, but this one is a unique example.Example #2: Arthur Miller: Attention Must Be Paid (By James Campbell)This biography is written in the form of a drama, presented in just two acts. In the first act, the author shows the famous dramatist, Arthur Miller, in his early success, having the love of the most beloved woman in the world, and resisting tyranny. However, in the second act of thisbiography, the author shows that the hero was badly assaulted and ridiculed by a rowdy mob called critics, who are expelled from the conventional theater. He ends his book with rhetorical details related to a revitalization in the fortunes of the playwright.Example #3: The Life of Samuel Johnson (By James Boswell)This biography is frequently hyped as a perfect example of modern biography, and all-time best example in the English language. This masterpiece of James Boswell has covered the whole life of the ubiquitous literary writer Samuel Johnson, with whom Boswell was well-acquainted. The unique quality of this book is that it shows Johnson as a walking intellectual amongst us.Example #4: The Bronte Myth (ByLucasta Miller)Emily, Anne, and Charlotte Bronte were very famous and eminent writers in the history of English literature. Many rumors and gossips were associated with them when they reached the peaks of their careers and received great approval for writing the most admired novels of the nineteenth century. In their biography, Lucasta Miller chunks the myths related to these young enigmatic women. This is a fine example of a biography.Example #5: Why this World: A Biography of Clarice Lispector (ByBenjamin Moser)After perusing his own private manuscripts and writings, this modernist writer, Benjamin Moser, has explored the mystique surrounding Brazilian writer Clarice Lispector. This is one of Moser’s biographies, which comes a little closer to finding her true nuances. All those readers who are going to read her myriad of works for the first time would find this biography interesting, and her life as beautiful and tragic, yet riveting.Function of BiographyThe function of writing biographies is to provide details regarding the life of a person or a thing in an entertaining but informative manner. By the end of a biography, readers feel like they are well-acquainted with the subject. Biographies are often non-fictional, but many biographers also use novel-like format, because a story line would be more entertaining with the inclusion ofstrong exposition, rising conflict, and then climax. Besides, the most inspirational life stories could motivate and put confidence into the readers.。
异常反馈流程学习评估
异常反馈流程学习评估Learning to evaluate the process of providing feedback on exceptions can be a challenging yet rewarding experience. Understanding the importance of properly addressing and resolving issues that arise is crucial in any team or organization. By honing the skills necessary to effectively assess feedback, one can contribute to a more efficient and harmonious work environment.学习评估异常反馈流程可能是一个具有挑战性但有益的经验。
理解妥善处理和解决问题的重要性对任何团队或组织都至关重要。
通过磨练必要的技能来有效评估反馈，可以为更高效和和谐的工作环境做出贡献。
When looking at the process of feedback evaluation, it is important to consider the various stakeholders involved. This includes not only the individuals providing the feedback, but also those responsible for receiving and addressing it. Understanding the perspectives and expectations of each party is essential in order to accurately assess the effectiveness of the feedback process.在审视反馈评估过程时，考虑到参与其中的各方利益攸关者是很重要的。
课堂教学反思模板全英
---Title: Classroom Teaching Reflection on [Subject]Date: [Date of Reflection]Teacher's Name: [Your Name]Subject: [Subject Taught]Grade Level: [Grade Level]---Introduction:The purpose of this reflection is to critically analyze and evaluate my recent classroom teaching experience. By examining my teaching methods, student engagement, and the overall learning outcomes, I aim to identify areas of strength and areas for improvement. This reflection will help me refine my teaching strategies and enhance the learning experience for my students.---Teaching Methods and Techniques:1. Lesson Planning:- What were the main objectives of the lesson?- How did the lesson plan align with the curriculum standards?- Did the lesson plan allow for flexibility to cater to diverse learning needs?2. Instructional Strategies:- What teaching methods did I use (e.g., lectures, discussions, group work)?- Were the methods appropriate for the subject and the students' learning styles?- Did the strategies promote active learning and critical thinking?3. Assessment:- How did I assess student understanding and progress?- Were the assessment methods effective in measuring learning outcomes?- Did the assessments provide constructive feedback for students?---Student Engagement and Participation:1. Student Interaction:- How did students respond to the lesson activities?- Were there any students who were particularly engaged or disengaged?- How did I encourage participation from all students?2. Group Dynamics:- How did group work contribute to the learning process?- Were there any challenges in managing group dynamics?- How did I address conflicts or issues within groups?3. Individual Student Feedback:- Did I provide timely and specific feedback to students?- How did students react to the feedback?- Did the feedback help students to improve their understanding or performance?---Learning Outcomes:1. Achievement of Objectives:- Did the majority of students achieve the learning objectives?- Were there any gaps in understanding that need to be addressed?- How did the lesson contribute to the overall curriculum?2. Student Progress:- How did individual students progress throughout the lesson?- Were there any students who needed additional support or challenge?- How did I adapt my teaching to cater to different learning levels?3. Overall Satisfaction:- How satisfied were the students with the lesson?- Did the lesson meet their expectations?- How could the lesson be improved to enhance student satisfaction?---Areas for Improvement:1. Teaching Methods:- What specific teaching strategies could be more effective?- How can I incorporate more interactive or technology-based learning activities?- What additional resources or materials could enhance the learning experience?2. Student Engagement:- How can I better cater to diverse learning styles and needs?- What strategies can I use to increase student participation and engagement?- How can I foster a more inclusive and supportive classroom environment?3. Assessment and Feedback:- What types of assessments would be more beneficial for this lesson?- How can I provide more meaningful and actionable feedback?- How can I use formative assessments to inform my teaching and adapt to student needs?---Conclusion:This reflection has provided me with valuable insights into my recent classroom teaching experience. By acknowledging both my strengths and areas for improvement, I am better equipped to refine my teaching practices and create a more effective and engaging learning environment for my students. I look forward to applying these learnings to future lessons and continuing to grow as an educator.---References:- [Any relevant educational theories or resources used in the reflection]---[Your。
(2021年整理)ISPE指南目录
（完整）ISPE指南目录编辑整理：尊敬的读者朋友们：这里是精品文档编辑中心，本文档内容是由我和我的同事精心编辑整理后发布的，发布之前我们对文中内容进行仔细校对，但是难免会有疏漏的地方，但是任然希望（（完整）ISPE指南目录）的内容能够给您的工作和学习带来便利。
同时也真诚的希望收到您的建议和反馈，这将是我们进步的源泉，前进的动力。
本文可编辑可修改，如果觉得对您有帮助请收藏以便随时查阅，最后祝您生活愉快业绩进步，以下为（完整）ISPE指南目录的全部内容。
ISPE指南按系列分类的目录清单:GAMP5GAMP 5: A Risk-Based Approach to Compliant GxP Computerized SystemsGAMP 5: 保证GXP计算机系统符合性的基于风险的方法GAMP Good Practice GuidesA Risk-Based Approach to Calibration Management （Second Edition）基于风险的校正管理方法（第二版）A Risk-Based Approach to Electronic Records and Signatures基于风险的电子记录和签名方法A Risk—Based Approach to GxP Compliant Laboratory ComputerizedSystems (Second Edition）基于风险的GXP符合性实验室计算机化系统方法(第二版)A Risk—Based Approach to GxP Process Control Systems （SecondEdition)基于风险的GXP工艺控制体系方法（第二版）A Risk-Based Approach to Operation of GxP Computerized Systems -A Companion Volume to GAMP 5基于风险的GXP计算机系统操作方法——-GAMP 5姊妹篇A Risk-Based Approach to Regulated Mobile Applications基于风险的移动APP管理方法A Risk-Based Approach to Testing of GxP Systems (Second Edition）基于风险的GXP系统检测方法（第二版）Electronic Data Archiving电子数据归档Global Information Systems Control and Compliance全球信息系统控制和符合性IT Infrastructure Control and ComplianceIT基础设施控制和符合性Legacy Systems遗留系统Manufacturing Execution Systems – A Strategic and Program Management Approach生产执行系统—策略和编程管理方法GAMP Good Practice Guides Under Development制订中的GAMP GPGISPE Baseline Pharmaceutical Engineering Guides for New and Renovated FacilitiesISPE基准：新设施和创新型设施药品工程指南Volume 1: Active Pharmaceutical Ingredients （Second Edition） - Revision to Bulk Pharmaceutical Chemicals卷1：活性药物成分（第二版）---对散装药用化学品的修订Volume 2: Oral Solid Dosage Forms （Second Edition)卷2:口服固体制剂（第二版）Volume 3: Sterile Product Manufacturing Facilities (Second Edition）卷3：无菌药品生产设施（第二版)Volume 4： Water and Steam Systems (Second Edition)卷4：水和蒸汽系统（第二版）Volume 5: Commissioning and Qualification卷5：调试和确认Volume 6： Biopharmaceutical Manufacturing Facilities （Second Edition）卷6：生物药品生产设施（第二版）Volume 7： Risk—Based Manufacture of Pharmaceutical Products (Risk—MaPP）卷7:基于风险的药品生产（风险MAPP)Baseline Guides Under Development制订中的基准指南ISPE GuidesISPE Guide: Science and Risk-Based Approach for the Delivery of Facilities, Systems， and EquipmentISPE指南:基于风险的设施、系统和设备传送科学方法ISPE Guide： Biopharmaceutical Process Development and ManufacturingISPE指南：生物药品工艺开发和生产(新出版）ISPE Guides Under Development在制订中的ISPE指南ISPE Good Practice Guides 优良规范指南ISPE Good Practice Guide： Applied Risk Management for Commissioning and QualificationISPE GPG：在调试和确认中应用风险管理ISPE Good Practice Guide： Approaches to Commissioning and Qualification of Pharmaceutical Water and Steam Systems （Second Edition）ISPE GPG:药用水和蒸汽系统调试和确认方法（第二版）（新出）ISPE Good Practice Guide： Assessing the Particulate Containment Performance of Pharmaceutical Equipment (Second Edition）ISPE GPG:制药设备颗粒密闭性能的评估（第二版）ISPE Good Practice Guide： Booklet LabelsISPE GPG:书册标签ISPE Good Practice Guide: Clinical Supply SystemsISPE GPG：临床补给系统（新出）ISPE Good Practice Guide： Cold Chain ManagementISPE GPG：冷链管理ISPE Good Practice Guide: Comparator ManagementISPE GPG：对照组管理ISPE Good Practice Guide： Development of Investigational Therapeutic Biological ProductsISPE GPG：临床前治疗用生物产品开发ISPE Good Practice Guide: Good Engineering PracticeISPE GPG：优良工程规范ISPE Good Practice Guide： Harmonizing the Definition and Use of Non—Investigational Medicinal Products (NIMPs）ISPE GPG:协调非临床前药品的定义和使用ISPE Good Practice Guide: Heating, Ventilation, and Air Conditioning （HVAC）ISPE GPG：HVACISPE Good Practice Guide： Interactive Response TechnologyISPE GPG：互动反馈技术ISPE Good Practice Guide： MaintenanceISPE GPG:维护ISPE Good Practice Guide: Ozone Sanitization of Pharmaceutical Water SystemISPE GPG：制药用水系统的臭氧消毒ISPE Good Practice Guide: Packaging， Labeling, and Warehousing FacilitiesISPE GPG：包装、贴标和仓储设计ISPE Good Practice Guide： Process GasesISPE GPG:工艺用气ISPE Good Practice Guide： Project Management for the Pharmaceutical IndustryISPE GPG：制药行业的项目管理ISPE Good Practice Guide: Quality Laboratory FacilitiesISPE GPG：质量化验室设施ISPE Good Practice Guide： Technology Transfer (Second Edition）ISPE GPG:技术转移(第二版）(新出）ISPE Good Practice Guides Under Development制订中的ISPE GPGPQLI Guides 药品质量生命周期实施指南PQLI Overview Good Practice GuidePQLI概览GPGProduct Quality Lifecycle Implementation （PQLI) from Concept to Continual ImprovementPart 1： Product Realization using QbD， Concepts and Principles从概念到持续改进的药品质量生命周期实施(PQLI）第一部分：利用质量源于设计（QbD)实现实现,概念和原则Product Quality Lifecycle Implementation (PQLI） from Concept to Continual ImprovementPart 2： Product Realization using QbD， Illustrative Example从概念到持续改进的药品质量生命周期实施(PQLI)第二部分:利用质量源于设计（QbD）实现实现，实例解说Product Quality Lifecycle Implementation （PQLI) from Concept to Continual ImprovementPart 3: Change Management System as a Key Element of a Pharmaceutical Quality System 从概念到持续改进的药品质量生命周期实施（PQLI）第三部分:药品质量体系关键要素变更管理Product Quality Lifecycle Implementation (PQLI) from Concept to Continual ImprovementPart 4： Process Performance and Product Quality Monitoring System (PP＆PQMS）从概念到持续改进的药品质量生命周期实施（PQLI）第四部分：工艺性能和药品质量监测体系（PP＆PQMS）ISPE PQLI Guides Under Development制订中的ISPE PQLI指南。
Positive and negative feedback striking a balance between necessary antagonists
1
1
Introduction
The quantity of biological information available for analysis is expanding at a tremendous rate, especially information obtained from molecular genetics techniques, but the methods of analysis are lagging behind. One of the most interesting challenges from the theoretical and practical points of view is to unravel the networks of regulation, involving for example metabolism or cell differentiation. These networks are expected to exhibit stability (since loss or change of cell differentiation are usually observed under unfrequent conditions, respectively anaplasia and metaplasia) and multistationarity (as there are many different cell types in metazoans), and we thus investigate both aspects. Our approach is to find conditions that are fulfilled by biological systems; our goal is to derive both new theoretical concepts, and practical constraints on interactions networks, to make more tractable the task of reverseengineering such networks from partial, sometimes unreliable, experimental data. To become viable, regulation models are in need of quantification, as pointed out by Koshland (1998). Boolean networks do not meet this need entirely, and biological systems have not been proven to exhibit behaviours easily modelled in a boolean fashion (Kringstein et al., 1998, have even shown in detail that eukaryotic transcription can give a graded response to concentrations of molecules); we thus consider continuous systems, which are much more difficult to deal with, but which provide much more accurate modelling of real systems. There have been attempts to study properties of generic regulation systems of dimension 1 or 2; for example, a beautiful characterisation of multistability has been achieved by Cherry & Adler (2000). But higher-dimensional systems display much greater complexity, and probably are the rule rather than the exception in real-world systems (it is becoming clearer and clearer that signalling pathways are extremely intricate, see for example Jordan et al., 2000). It seems difficult to scale up the results in 2 dimensions to higher dimensions, which is why we try not to make assumptions about the dimensions of the systems we study, even if it is at the cost of sometimes less powerful results. In the first part of this article, we study steady states of a generic system, and derive qualitative and quantitative properties of the system according to its behaviour at these steady states. In the second part, we build on these results to tackle ”Delbr¨ uck’s conjecture”; Max Delbr¨ uck proposed in 1948 that cell differentiation could be established by a unique regulating system having distinct attractors (Delbr¨ uck, 1949). In 1980, R. Thomas made the conjecture that ”the presence of a positive circuit in the logical structure 2
北京市部分区2021-2022学年高一下学期期末考试英语试题汇编：阅读表达
阅读下面短文，根据题目要求用英文回答问题。
50. How did Einstein get himself on the path towards the Theory of Relativity?
51. According to this passage, what does “the relativity of time” mean?
52. What did Einstein believe in as being important for thinking?
53. What do you think of “creative thinking”? Explain why. (In about 40 words)
北京市房山区2021-2022学年4小题；第40、41题各2分，第42题3分，第43题5分，共12分）
It would appear that the secret to improving your memory is really simple. It’s to take breaks. As a matter of fact, the brain has more limited energy resources than you probably think. Therefore, it tires quickly and its levels of attention begin to drop after about 30 to 45 minutes.
我的数学不好英语作文
I have always found mathematics to be an enigmatic realm, one that seems to defy my comprehension despite my best efforts. This admission is not made lightly; it carries with it a weight of frustration, self-doubt, and a persistent longing for understanding. In this essay, I delve into the multifaceted reasons behind my struggle with mathematics, examining the cognitive, emotional, pedagogical, and environmental factors that have contributed to this challenge.Firstly, from a cognitive perspective, I acknowledge that my difficulties in math might stem from inherent learning preferences and strengths. As an individual with a strong inclination towards linguistic and creative thinking, I often find myself thriving in subjects like literature, history, or art where concepts are conveyed through narratives, metaphors, and visual representations. Mathematics, on the other hand, with its abstract symbols, rigid structures, and emphasis on logical reasoning, presents a stark contrast to these preferred modes of learning. The language of math, characterized by formulas, equations, and algorithms, feels foreign and inaccessible to me, akin to trying to decipher an ancient script without a decoder ring. Moreover, my brain seems wired to excel in tasks that require synthesis, interpretation, and subjective evaluation, rather than those demanding precision, sequential thinking, and algorithmic problem-solving, which are paramount in mathematics.Secondly, emotions play a significant role in shaping my relationship with math. Fear, anxiety, and a sense of inadequacy have become inextricably intertwined with my mathematical experiences. The fear of failure, exacerbated by the cumulative nature of math where each concept builds upon the previous ones, creates a paralyzing effect. Even a minor misunderstanding at one stage can snowball into a major obstacle in subsequent lessons. This fear is further fueled by the societal perception that math proficiency is a marker of intelligence, leading to a deep-seated feeling of inadequacy when I struggle to grasp mathematical concepts. The resulting anxiety hampers my ability to concentrate, retain information, and think logically, creating a vicious cycle where poor performance reinforces negative emotions, which in turn hinder further learning.Thirdly, the pedagogical approach to teaching mathematics has also played a part in exacerbating my difficulties. Traditional math instruction often relies heavily on rote memorization, formulaic problem-solving, and a 'one-size-fits-all' curriculum that fails to cater to diverse learning styles. For someone like me, who thrives on context, meaning, and personal connections, this methodical, decontextualized presentation of math can be incredibly disengaging. Furthermore, the speed at which new concepts are introduced and the lack of adequate time for consolidation and practice often leave me struggling to keep up. The absence of real-world applications or interdisciplinary connections further distances me from appreciating the relevance and beauty of mathematics, leaving me feeling like an outsider looking in on a secret club.Lastly, environmental factors, both at home and in school, have contributedto my struggle with math. Growing up, math was not a subject actively discussed or valued in my family. The absence of positive mathematical role models or a culture that encourages mathematical exploration meant that I did not develop a mindset that embraces challenges and sees mistakes as opportunities for growth. Similarly, in school, a competitive atmosphere that prioritizes speed and accuracy over deep understanding and collaborative learning has stifled my confidence and willingness to engage with math. The pressure to perform well on standardized tests, coupled with a lack of individualized support, has left me feeling overwhelmed and unsupported in my mathematical journey.In conclusion, my struggle with mathematics is a complex interplay of cognitive preferences, emotional barriers, pedagogical limitations, and environmental influences. Recognizing these multifaceted factors is crucial not only for my own self-awareness but also for informing potential strategies to overcome my challenges. To improve my relationship with math, I need to find ways to bridge the gap between my linguistic inclinations and the abstract nature of mathematics, manage and mitigate math-related anxiety, advocate for a more inclusive and adaptive teaching approach, and cultivate a supportive, growth-oriented environment that values the process of learning over mere outcomes. By addressing these issues systematically, I hope to unlock the mysteries of mathematics and perhaps even discover the joy and beauty that lie beneath its seemingly impenetrable surface.。
RCS-Based RoboCrane Integration Abstract
AbstractThe Intelligent Systems Division (ISD) of the National Institute of Standards and Technology (NIST) has been researching new concepts in robotic cranes for several years. These concepts use the basic idea of the Stewart platform parallel link manipulator. The unique feature of the NIST approach is to use cables as the parallel links and to use winches as the actuators. Based on this idea, a revo-lutionary new type of robot crane has been developed and aptly named the RoboCrane. The RoboCrane provides six degree-of-freedom load stabilization and maneuverability. This is accomplished through the following controlmodes: master/slave; joystick input; operator panel input; preprogrammed trajectory following (teach programing, graphical off-line programing, or part programing); and sensor based motion compensation. The current control system includes both the controller and user interface within the same control level, which makes controller enhancements and modifications difficult and error prone. A Real-time Control System (RCS) [1] is currently being developed to include the above control modes, while pro-viding an open systems architecture. This supports hierar-chical control modules, along with a separate user interface. A remote telepresence system is also being implemented to provide foveal/peripheral stereo images and other necessary data to enable a remote operator to perform a variety of tasks with the RoboCrane.The objective of this paper is to describe past and future efforts toward integration of an RCS open system architec-ture controller into the RoboCrane Integration Testbed (RIT). An RCS-based Robo-Crane controller will allow for continuing research into parallel-link manipulator con-trollers and application oriented controller capabilities. Specifically, this paper introduces the RoboCrane concept, describes the current control system, the envisioned RCS-based RoboCrane control system, and the intended inte-gration procedure.Keywords: Real-Time Control System (RCS), open sys-tem architecture, parallel link manipulator, Stewart plat-form devices, cable-based robots, cranes.1.0 IntroductionThe RoboCrane prototype (FIGURE 1) was first developed by NIST in the late 1980's [2]. A NIST program on robot crane technology, sponsored by the Defense Advanced Research Projects Agency (DARPA), developed and tested several potential robot crane designs to determine the desired performance characteristics of a proposed robot crane. Initial testing of these prototypes showed that a six cable design results in a remarkably stable platform capa-ble of performing accurate six degree-of-freedom (DOF) manipulations. This stabilized platform can be used to improve typical crane operations or as a maneuverable robot/tool base.The RoboCrane is based on the Stewart platform parallel-link manipulator [3], but uses cables as the parallel links and winches as the actuators. By attaching the cables to a suspended work platform and maintaining tension in all six cables, the load is kinematically constrained. More-over, the suspended work platform resists perturbing forces and moments with a mechanical stiffness deter-mined by the angle of the cables, the suspended weight, and the elasticity of the cables. Based on these concepts, the RoboCrane is a revolutionary new type of robot crane that can control the position, velocity, and force of tools and heavy machinery in all six degrees of freedom (x, y, z, roll, pitch, and yaw).NIST research into Stewart Platforms also produced an innovative structure from which to suspend the Rob-oCrane work platform. An octahedral tubular structure, containing the three upper support points necessary to sus-pend the work platform, provides exceptional structural stiffness in a lightweight frame. By connecting the struc-ture’s legs in an octahedron configuration, forces and torques incurred by the work platform are translated into pure compressions and tensions in the legs. With only slight bending moments in each of the structure’s legs (due to self-weight), the RoboCrane’s octahedron structure can be made extremely lightweight compared to conventional gantry structures. This fact, along with the RoboCrane’s ability to lift very heavy loads, produces a much higherThis work was prepared by U.S. Government employees as part of their official duties and is therefore a work of the U.S. Government and not subject to copyright. Equipment listings in this paper do not imply a recommendation by NIST nor that the equipment is best for the purpose.RCS-Based RoboCrane IntegrationRoger Bostelman, Adam Jacoff, Nicholas Dagalakis, James AlbusIntelligent Systems DivisionNational Institute of Standards and TechnologyGaithersburg, Maryland 20899lift-to-weight ratio than conventional serial link manipula-tors. Also, the RoboCrane’s stable structure is well suited for mobility. By affixing independent wheeled vehicles under each of the structure’s three feet, the RoboCrane can be made to traverse rough terrain. This was demonstrated using a 2-meter, radio controlled prototype [4].The 6 m RoboCrane prototype has been subjected to a variety of performance measurements and computer simu-lations. Experimental tests were conducted to verify its functional work volume, static loading capability, and load positioning accuracy. These experimental results com-pared favorably to associated computer analysis [2].The RoboCrane work platform has been equipped with tools such as a gripper, grinder, welder, saw, and inspec-tion equipment (stereo vision and laser scanner). Thesetools have been used to demonstrate a variety of tasks. Each new application has contributed to the overall func-tionality of the Robo-Crane controller and to the design of the human/computer interface. The current controller implementation provides for intuitive and robust control of the RoboCrane through the following control modes: mas-ter/slave; joystick input; operator panel input; prepro-grammed trajectory following (teach programing,graphical off-line programing, or part programing); and sensor based motion compensation. Individual winch con-trol is also possible. Potential application areas for the RoboCrane technology can be found in the construction industry [5], nuclear/toxic waste cleanup, the subsea arena [6], and in planetary exploration [4].NIST has also developed an adaptation of the RoboCrane technology in order to investigate its effectiveness forFIGURE 1. 6-Meter NIST RoboCrane Prototype.SUSPENDED UPPER SUPPORT POINTSFOOTINGWITH PULLEY PAIRSPLATFORM TRIANGULAR WORK(1 of 3)(2 OF 3)TOOLSOCTAHEDRON STRUCUREWITH WINCH PAIRSFIGURE 2. TETRA Platform: Development platform for RCS-based RoboCrane Controllerlong-line suspended loads. This prototype system is called the Tetrahedral Robotic Apparatus (TETRA) (FIGURE 2). TETRA uses a similar Stewart Platform geometry with six cables driven by six winches. However, TETRA’s winches, amps, and computer controller are all mounted on the work platform instead of the supporting structure. This allows existing overhead lifting systems such as bridge cranes, boom cranes, and helicopters to be retrofitted. Also, the TETRA system is designed to allow the crane’s hook to provide most of the lifting force, while the addi-tional six winches and cables provide controlled maneu-verability during cargo acquisition and stabilization of the cargo during transfer. The six cables essentially act as coordinated taglines, providing 6 DOF control, thereby limiting spinning and swinging of the hook and cargo. The TETRA prototype will be used as the development plat-form for the RCS-based RoboCrane controller prior to implementation onto the 6m RoboCrane prototype.2.0 Current Control SystemThe RoboCrane Integration Testbed (RIT) control system currently consists of a controller computer with graphical operator interface, a remote operator/observer interface, and graphical programmer [7]. All these components are connected through a network, as shown in the RoboCrane System Architecture Diagram (FIGURE 3). The Rob-oCrane control system functions in the following control modes:•Master/slave•Joystick input•Operator panel input•Preprogrammed trajectory following- Teach programing- Graphical off-line programing- Part programing•Sensor based motion compensation.The RoboCrane’s available motion types include:•Single joint•Cartesian base frame (default)•Cartesian platform frame•Cartesian offset frame (tool center point)•Constrained motions along vectors•Rotations about vectors•Single axis force control2.1 The ControllerThe current RoboCrane Controller consists of a 64MHz Macintosh Quadra and an electronics rack electrically tied to one another through a Sensor Interface (FIGURE 3). Both the computer and the electronics rack house multiple components. The Sensor Interface converts the followingFIGURE 3. Current RoboCrane System Architecture Diagramto be digitally fed back into the computer:•six phase quadrature signal inputs from winch motor encoders (500 pulses per revolution) •six analog 10-turn potentiometers (pot) on each winch motor through speed reducers•six analog pots making up a Stewart platform (SP) joystick•six analog output tension sensors into digital information.The Sensor Interface outputs (not shown in figure 3) can be digital or analog. They are used to turn on/off linear actuators, tools and lights for a variety of applications such as gripping, grinding, and welding. The Macintosh outputs are analog signals from a 12 bit digital/analog converter board that provides input to the six power amplifiers. Pulse Width Modulated (20kHz) power amplifiers can be set in velocity or torque control modes. These are typically con-figured for velocity mode. For the operator interface, the RoboCrane control panel (FIGURE 4) provides interactive control over RoboCrane functions, settings, and status. Because the operators of these controllers are not expected to be computer literate and might wear protective gloves, a simple and intuitive graphic interface was developed using LabView software. Control modes and motion types can be activated or deactivated by computer mouse actions and/or touch screen actions. For direct manipulation of the RoboCrane platform, a 6 DOF force sensing joystick (Spa-ceball) is input serially through the computer modem port and communicates at 19.2 kbaud.2.2 Remote Operator/Observer InterfaceA Remote Operator/Observer Interface was developed to allow networked communications with the RoboCrane controller. This graphical interface consists of a touch-screen panel which looks similar to the actual RoboCrane controller panel. It communicates with the RoboCrane controller via ethernet and serves as a remote operator control station anywhere the network can reach. In addi-tion, this remote interface can act simply as an observer of RoboCrane operations, without any control functionality. The interface’s front panel buttons allow a remote opera-tor, with the proper permissions, to take control of Rob-oCrane functions or to simply act as a passive observer. The executable code for this Remote Operator/Observer Interface can be run on a computer located anywhere on the network.2.3 The Off-Line Graphical Programmer The Off-Line Graphical Programmer allows for safe and easy generation of platform trajectories (move commands) along with the timely actuation of tools. The Graphical Programmer runs on a Silicon Graphics (SGI) computerFIGURE 4. Robocrane Control Paneland controls the operation of a TGRIP (Teleoperative Graphical Robot Instruction Program) simulation of the RoboCrane workspace. The RoboCrane’s platform and tools are intuitively represented as three dimensional solid models. Graphical simulations of Robo-Crane motions, tool status, and other information is stored in a standard text file and is then made available to the RoboCrane through the network.2.4 Telepresence SystemA telepresence system is currently being developed and integrated into the RoboCrane Integration Testbed. It will include a flexible manipulator, called EMMA (Easily Manipulated Mechanical Armatures). The EMMA arm will be attached to the work platform with a pan-tilt-ver-gence (PTV) head as an end-effector. Dual sets of stereo cameras (foveal and peripheral views) will be installed on the PTV head. These allow a remote operator to have full telepresence into the RoboCrane work volume during operation. The operator will wear a heads-up display to enable continuous visual contact along with access to RoboCrane controller operations.2.5 System ComplexityThe RoboCrane controller consists of a complex assort-ment of over 900 electrical wires that can be difficult, tedious, and time-consuming to troubleshoot. The current configuration requires strong electrical signals to pass through long cables connecting the controller to motors, sensors, and tools. For example, the cables connecting the power amplifiers to the motors measure more than 25m in length. This type of electrical system complexity will be remedied in the RCS-based RoboCrane Control System.3.0 RCS-Based RoboCrane Control SystemIn addition to maintaining the RoboCrane’s current control modes and capabilities, the proposed RCS-based Rob-oCrane Control System (FIGURE 5) contains several tar-geted improvements. Some examples are as follows:•hierarchical, open system architecture withstandard control module interfaces•separate user interface into each level of the controller•more adaptable electronic design which scales well for larger systemsThese and other enhancements will form a modular, reconfigurable control system that will allow the system to be easily optimized for particular applications.3.1 Computer PlatformsThe proposed computer platforms for the RCS-based Rob-oCrane Control System differ from the current system. The Controller hardware will be based on PC-compatible machines running the Windows NT operating system with real-time extensions. This combination will support the RoboCrane’s computing requirements, while maintaining consistency with a de facto industry standard platform. Similarly, the graphical operator interface will be a C++ implementation running on a PC compatible/Windows NT machine. The Off-Line Graphical Programmer will remain on the SGI (running UNIX) due to heavy graphics and rendering requirements. All of these computer subsystems will be connected via ethernet.FIGURE 5. Proposed RoboCrane System Architecture Diagram3.2 Component ChangesOne major change to the physical components of the sys-tem will be the power amplifiers for the winches. In the proposed system, the same six winches will be driven by Controller Area Network amplifiers (CANAMP) instead of conventional power amplifiers. These CANAMPs will be located at each of the winches. The CANAMPs will communicate with the host computer through a CAN interface card residing within the Controller computer. The data transfer rate for the CAN system is up to 1 Mbit/sec. for networks up to 40m long and can contain 0 to 8 bytes of data without segmentation. These CANAMPs will remove the need for RoboCrane’s existing Sensor Inter-face, power amplifiers, power supplies, and isolation trans-formers. In addition, an emergency stop system (not shown in the figure) will be independently connected to the six CANAMPs, thereby eliminating the need for a sep-arate electronics rack.Most other major components, such as the winches, joy-stick and sensors (tension, encoder, potentiometer), will be incorporated into the proposed design. The addition of “jog pots” will be necessary for occasional direct axis con-trol of each winch, because the amplifiers will be co-located with winches. Jog pots will be used during non-computer controlled tasks such as calibration and cable replacement.3.3 Hierarchical Controller ModulesThe core concept behind the RCS-based RoboCrane Con-troller architecture centers around a hierarchical decompo-sition of tasks required to perform a particular application (FIGURE 6) [9]. The basic decomposition of commands can be loosely thought of in terms of time needed to per-form the action. That is, at the bottom most Servo Level ,time (t ) can be considered instantaneous. As one traverses up the hierarchy, each command level requires roughly an order of magnitude greater time (10x) to perform its com-mand.Therefore, the Primitive (Prim) Level would require roughly 10t, the Elemental-Move (E-Move) Level would require 100t , the Task Level would require 1000t , and so on.Similarly, the RCS model supports upward passing of sen-sor data as necessary for a particular module to perform its function. As each decomposed command is successfully performed (or not) a status is sent back up the hierarchy to the appropriate controller module at any given level. At the higher levels of the hierarchy, such as the Task Level , com-mands take longer to perform and status indications return less frequently than at the lower levels. While at the Servo Level , commands, status messages and sensor readings are carried out almost continually.In addition, the concept of a world model database is maintained so that any controller module may access a particular piece of information if and when it is necessary. For instance, the Workcell Level: Robocrane module might need to check the status of a gripper (functional or broken) before agreeing to perform some task that involves part manipulation.An essential part of an open system architecture is the def-inition and standardization of controller module interfaces. This is important because it allows developers of particu-lar modules with enhanced capabilities or experimental algorithms to “plug” their module into the overall control system and test it seamlessly. In the case of the RCS-based RoboCrane controller, the Neutral Manufacturing Lan-guage (NML) [10] will be used to perform all communica-tions between control modules.FIGURE 6. RCS Task Decomposition for the RoboCrane’s Work Platform MotionTIME t10t100t1,000t10,000t3.4 Task NarrativeAs an example, consider the task decomposition of a Rob-oCrane application such as welding (FIGURE 6). For the RoboCrane to perform a welding application, the Rob-oCrane’s Workcell Level:RoboCrane control module must receive a command to “weld a part.” This high level com-mand could be generated by a Shop Level control module (not shown in the figure), just as a part was placed in the RoboCrane work volume. Alternatively, that same com-mand could be issued via the operator interface for that level, as will be the case for the RoboCrane. The “weld a part” command is then decomposed by an intelligent Workcell Level:RoboCrane module which understands what tools are available and knows how to decompose the given command into its constituent elements. As a result, it passes down two simultaneous Task Level commands to its subordinates, Task Level: Platform Motion and Task Level: Weld. As the Task Level: Platform Motion module receives its command, it decomposes it into a series of goal motions and passes them down to the E-Move Level: Trajectory Generator module. This module understands the RoboCrane platform kinematics and can turn goal motions into an appropriate path. If the RoboCrane’s kine-matics were changed or if a completely different robot were used, this is the only controller module that would need to be updated. The E-Move Level: Trajectory Gener-ator module passes down path information that the Prim Level: CAN Interface module knows how to turn into joint goals for each Servo Level: CANAMP module. Then each Servo Level: CANAMP receives its goal information from above and interfaces directly with its specific hardware (a winch in this case). Each Servo Level: CANAMP sends incremental move commands to its winch and listens for feedback from associated sensors (encoder or tachometer). This way, the Servo Level: CANAMP closes the servo loop right at the winch, providing a rapid response. All these servo level actuators are synchronized by the levels above so that the resulting platform motion traces the intended robot trajectory.Meanwhile, the Task Level: Weld module has been waiting for the status of the platform motion to show that it is near the intended weld seam. The Task Level: Weld module may check the world model database to see if a particular point has been passed. Once that status is true, the Task Level: Weld module will begin to deploy the weld tip and at the appropriate time, power the welder on and off to cre-ate a synchronized welding path.4.0 Integration ProcedureThe RCS-based RoboCrane controller will support the current RoboCrane control modes, motion types, and tar-geted improvements listed in section 3.0 above. However, it is being developed on the TETRA platform (FIGURE 2) prior to integration into the RoboCrane Testbed for several tactical reasons. First, it is preferable to maintain Rob-oCrane demonstrations of tools and applications while the RCS-based controller is being developed. Second, because the TETRA and RoboCrane platforms differ slightly, adapting the controller to RoboCrane will provide a good first test of the controller’s modularity and portability. Third, TETRA’s compact design provides easy access to all hardware and will simplify development and testing of the new controller.TETRA’s current controller configuration consists of six conventional amplifiers driving each of its six winches. Feedback from each winch encoder and tachometer are input into an eight axis PMAC motion controller board housed within an onboard PC compatible computer. The onboard PC also houses a CANPC interface board, which sends digital signals through a “token ring” style network to each CANAMP located at each winch. TETRA’s cur-rent configuration is wired to allow easy switching between using the conventional amplifiers and the CAN-AMP system to power the winches. When the CANAMPs are in service, the servo motion control loop is imple-mented between the intelligent CANAMP and the winch, instead of sending motor feedback signals all the way to a centralized motion controller board. This will be particu-larly beneficial for systems such as RoboCrane where this length can be 30m or more. TETRA’s CANAMP system will provide a functional example of the reduced elec-tronic design complexity that was mentioned previously. Both TETRA’s onboard controller and offboard operator interface computers will use the Windows NT operating system, supplemented with real-time extensions which provide a 1ms cycle time. The onboard controller will incorporate RCS templates, NML communication proto-cols, shared memory, and other concepts consistent with an RCS controller [9]. The graphical user interface will be developed in C++.NIST has contracted with an industry partner, Advanced Technology and Research Corp. (ATR), to work closely in development of the RCS-based RoboCrane Controller aimed at producing a commercially available system. ATR will provide control module development and consulting, similar to their efforts on NIST’s Enhanced Machine Con-troller [9], a 4-axis machine tool controller currently in use in a General Motors manufacturing facility.5.0 SummaryBoth the RoboCrane and TETRA prototypes are based on the Stewart platform parallel-link manipulator. They use cables as the parallel links and winches as the actuators. They can control the position, velocity, and force of tools and heavy machinery in all six degrees of freedom (x, y, z, roll, pitch, and yaw).The current RoboCrane control system consists of a com-puter Controller, a Remote Operator/Observer Interface, and an Off-line Graphical Programmer connected by a net-work. The Controller itself can support a variety of control modes and motion types. The Remote Operator/ObserverInterface can be used from any computer located on the network. It allows for full remote control of all RoboCrane functionality or can be used simply to observe RoboCrane operations.Although the current RoboCrane controller is functional, it does not provide a standard, easily modifiable, open sys-tem, nor does it use a hierarchical control architecture. So, ISD is focusing on development of an RCS-based Rob-oCrane Control System that supports all the existing con-trol modes and provides improvements. These include standardization of controller module interfaces, separate user interfaces, more efficient electronic design, and oth-ers. A commercially available Controller Area Network (CAN) is also being incorporated to allow tighter servo loop control and limit wiring complexity.The RCS-based RoboCrane Controller is being developed on the TETRA system to allow continued demonstration of the RoboCrane while controller work advances. TETRA’s control can easily switch between conventional amplifiers and CANAMPs to allow stepwise development of the controller. NIST has contracted with an industry partner (ATR) to work closely in development of the RCS-based RoboCrane controller with the intention of produc-ing a commercially available system that could be readily reconfigured for different applications.6.0 References[1] Albus, J., McCain, H., Lumia, R., “NASA/NBS Stan-dard Reference Model for Telerobot Control System Architecture (NASREM),” NIST Technical Note 1235, National Institute of Standards and Technology, July, 1987.[2] Albus, J. S., Bostelman, R. V., Dagalakis, N. G., “The NIST ROBOCRANE, A Robot Crane,” Journal of Robotic Systems, July 1992.[3] Stewart, D., “A Platform with Six Degrees of Free-dom,” Proceedings of the Inst. of Mechanical Engineering, Volume 180(15), Part I:371-386, 1965-1966[4] Bostelman, R.V, Albus, J. S., et. al., “A Stewart Plat-form Lunar Rover,” Engineering Construction and Opera-tions in Space IV Proceeding, Albuquerque, NM, Feb 26-Mar. 3, 1994.[5] Bostelman, R., Albus, J., Dagalakis, N., Jacoff, A., “RoboCrane Project: An Advanced Concept for Large Scale Manufacturing,” Association for Unmanned Vehi-cles Systems International Proceedings., Orlando, FL, July 15-19, 1996.[6] Bostelman.R.V., Albus, J.S., “Stability of an Underwa-ter Work Platform Suspended from an Unstable Refer-ence,” Engineering in Harmony with the Ocean Proceedings, October 1993.[7] Dagalakis, N. G., Albus, J.S., Bostelman, R.V., Fiala, J., “Development of the NIST Robot Crane Teleoperation Controller,” Robotics and Remote Handling Proceedings, Fifth Topical Meeting, Knoxville, TN, April 1993.[8] Bostelman, R.V., Albus, J.S., Dagalakis, N. G., “A Robotic Crane System Utilizing the Stewart Platform Con-figuration,” International Symposium on Robotics and Manufacturing Proceedings, Santa Fe, NM, November 10-12, 1992.[9] Lumia, Ronald, “The Enhanced Machine Controller Architecture,” International Symposium on Robotics and Manufacturing Proceedings, Maui, HI, August 14-18, 1994.[10] Shackleford, Will, “The NML Programmer’s Guide,” /proj/rcs.lib/nml.html, 1995.。
英语优秀说课课件ppt
01
Course Introduction
Course objectives
01
Improving English listening and speaking skills
02
Enhancing English reading and writing abilities
03
Developing cultural awareness and communication skills
02
Teaching method
Task Based Teaching Method
Practice oriented
The Task Based Teaching Method emphasizes learning English through completing tasks, enabling students to master language skills in practical use. This method emphasizes active participation and collaborative learning among students, which helps to cultivate their practical application abilities.
03
Teaching content
Vocabulary and Grammar
Solid foundation
This section focuses on explaining the application of English vocabulary and grammar, and through examples and exercises, helps students deeply understand the basic elements of English, laying a solid foundation for subsequent learning.
一种机制四种方法英文
一种机制四种方法英文One mechanism, four methods (in English):1. Mechanism: Feedback- Method 1: Surveys - collecting data through questionnaires and interviews.- Method 2: Focus groups - gathering a group of individuals to discuss and provide feedback on a specific topic.- Method 3: User testing - observing users as they interact with a product or service and collecting their feedback.- Method 4: Social media monitoring - keeping an eye on social media platforms to analyze user sentiments and collect feedback.2. Mechanism: Incentives- Method 1: Rewards and incentives - offering tangible or intangible rewards to individuals for providing feedback.- Method 2: Contests and competitions - organizing contests where users can participate and win prizes by providing feedback.- Method 3: Loyalty programs - creating loyalty programs where users can earn points or rewards for providing feedback.- Method 4: Referral programs - incentivizing users to refer others to provide feedback by offering rewards for successful referrals.3. Mechanism: Communication and Collaboration- Method 1: Online forums and communities - creating online platforms where users can discuss and collaborate in providing feedback.- Method 2: Feedback events and workshops - organizing events or workshops where users can come together and provide feedback through interactive sessions.- Method 3: Collaborative platforms - using collaborative platforms that allow multiple users to share and provide feedback on a common platform.- Method 4: Feedback sessions with stakeholders - engaging key stakeholders in feedback sessions to gather their insights and suggestions.4. Mechanism: Data Analysis- Method 1: Quantitative analysis - analyzing numerical data collected through surveys, questionnaires, and other quantitative methods.- Method 2: Qualitative analysis - analyzing textual data such asfeedback, comments, and reviews for themes, patterns, and insights.- Method 3: Sentiment analysis - using natural language processing techniques to analyze the sentiments expressed in feedback and reviews. - Method 4: Data visualization - representing feedback data through visualizations like charts, graphs, and heatmaps for better understanding and interpretation.。
multivariable feedback control analysis and design
Multivariable Feedback Control Analysis and Design1. IntroductionIn modern control systems, it is often necessary to control multiple variables simultaneously in order to meet specific performance requirements. This is known as multivariable control. Multivariable feedback control analysis and design involves the study of techniques for designing control systems that can handle multiple variables simultaneously. In this article, we will explore various aspects of multivariable feedback control analysis and design.2. Multivariable Control SystemsMultivariable control systems are systems that have multiple inputs and multiple outputs. These systems are typically more complex than single-input single-output (SISO) systems because the interactions between different variables can complicate the control design process. Understanding the characteristics and behavior of multivariable control systems is crucial for their effective analysis and design.2.1 System IdentificationBefore designing a multivariable control system, it is important to identify the dynamic behavior of the system. System identification techniques can be used to determine the mathematical models that describe the relationships between inputs and outputs of the system. This involves collecting input-output data and using various modeling techniques such as empirical modeling, transfer function modeling, or state-space modeling.2.2 Control ObjectivesIn multivariable control, there are often multiple conflicting control objectives that need to be satisfied simultaneously. These control objectives can include stability, desired transient response, disturbance rejection, and tracking of setpoints. Balancing theseobjectives and designing controllers that achieve them is a central aspect of multivariable control analysis and design.2.3 Interactions and CouplingOne of the key challenges in multivariable control is the presence of interactions and coupling between the different variables. Interactions occur when changes in one variable affect the behavior of another variable. These interactions can make it difficult to design controllers that do not interfere with each other. Understanding and mitigating interactions is essential for effective multivariable control.2.4 Controller StructureThe selection of an appropriate controller structure is critical to the success of multivariable control design. There are various types of controller structures that can be used, such as decentralized control, centralized control, and decentralized control with optimization. Each structure has its advantages and disadvantages, and the choice depends on the specific requirements of the control problem.3. Multivariable Control AnalysisMultivariable control analysis involves studying the behavior and performance of multivariable control systems. It aims to provideinsights into the system’s dynamics, stability, and robustness.3.1 Stability AnalysisStability is a fundamental requirement for any control system. In multivariable control, stability analysis becomes more complex due to the interactions and couplings between variables. Stability analysis techniques such as eigenvalue analysis, Nyquist stability criterion, and pole placement methods can be used to investigate and ensure thestability of a multivariable control system.3.2 Performance AnalysisPerformance analysis involves evaluating the performance of a multivariable control system in terms of its response to inputs and disturbance rejection. Performance measures such as rise time, settling time, overshoot, and steady-state error can be used to assess the system’s performa nce. Analysis techniques like frequency response analysis, time response analysis, and sensitivity analysis can provide valuable insights into the system’s performance characteristics.3.3 Robustness AnalysisRobustness analysis is concerned with the ability of a multivariable control system to withstand uncertainties and variations in the system’s parameters. Robust control techniques aim to designcontrollers that can provide satisfactory performance over a range of operating conditions and system uncertainties. Sensitivity analysis, robust stability analysis, and the use of optimal control techniques are commonly employed for robustness analysis in multivariable control.3.4 Interaction AnalysisInteraction analysis is crucial for understanding and managing the interactions between variables in a multivariable control system. Interaction measures such as relative gain array (RGA) and condition number matrix (CN) can be used to quantify the strength and direction of interactions. Analysis of interaction patterns can help in choosing appropriate control strategies and gain scheduling techniques to minimize interactions.4. Multivariable Control DesignMultivariable control design involves designing controllers that achieve the desired control objectives while taking into account the system’s dynamics, interactions, and constraints.4.1 Decentralized Control DesignDecentralized control design involves designing individual controllers for each variable in a multivariable control system. Each controller operates based on its local measurements and controls its respective variable. Decentralized control can be advantageous when theinteractions between variables are weak, and the system can be effectively decoupled.4.2 Centralized Control DesignCentralized control design aims to design a single controller that regulates all variables simultaneously. This approach considers the interactions between variables explicitly and can achieve better overall control performance. However, centralized control can be computationally complex and may require accurate modeling of the entire system.4.3 Decentralized Control with OptimizationDecentralized control with optimization is an intermediate approach that combines the advantages of both decentralized and centralized control. It involves designing decentralized controllers for individual variables while optimizing their performance collectively. This approach can provide a good balance between performance and complexity.4.4 Controller Tuning MethodsOnce the controller structure is selected, tuning methods are used to determine the controller parameters. Various tuning methods are available, such as PID tuning, gain scheduling, pole placement, and optimization-based methods. Each method has its advantages and limitations, and the choice depends on the specific control problem and requirements.5. ConclusionMultivariable feedback control analysis and design are essential for the effective control of systems with multiple variables. This article discussed key aspects of multivariable control systems, including systemidentification, control objectives, interactions and coupling, and controller structure selection. It also explored multivariable control analysis techniques, such as stability analysis, performance analysis, robustness analysis, and interaction analysis. Furthermore, the article covered various multivariable control design approaches like decentralized control, centralized control, and decentralized control with optimization. By understanding and applying these concepts and techniques, engineers can design robust and efficient multivariable control systems to meet desired control objectives.。
优秀生培养方法英语作文
To cultivate outstanding students,it is essential to focus on a holistic approach that encompasses academic excellence,personal development,and social skills.Here are some effective methods for nurturing excellence in students:1.Encourage Curiosity and Creativity:Foster an environment where students are encouraged to ask questions,explore new ideas,and think creatively.This can be done through projectbased learning and discussions that challenge their perspectives.2.Set High Expectations:High expectations can motivate students to strive for excellence.Teachers should believe in their students abilities and push them to achieve their full potential.3.Provide Personalized Learning:Recognize that each student has unique learning needs and styles.Personalize the learning experience to cater to these differences,using differentiated instruction and adaptive learning strategies.4.Promote Critical Thinking:Incorporate activities that require students to analyze, evaluate,and synthesize information.This can be achieved through debates,case studies, and problemsolving exercises.5.Emphasize the Importance of Hard Work and Perseverance:Teach students that success comes from dedication and consistent e examples of successful individuals who have overcome challenges through hard work.6.Offer Support and Resources:Provide students with the necessary resources and support to excel in their studies.This includes access to quality educational materials, tutoring,and counseling services.7.Cultivate a Growth Mindset:Encourage students to view challenges as opportunities for growth rather than as obstacles.Teach them that intelligence and abilities can be developed over time.8.Encourage Extracurricular Activities:Participation in sports,arts,and clubs can help students develop leadership skills,teamwork,and time management,which are valuable for their overall development.9.Foster a Collaborative Learning Environment:Promote group work and collaboration, allowing students to learn from each other and develop social skills.10.Provide Regular Feedback:Offer constructive feedback on students work to helpthem understand their strengths and areas for improvement.This should be done in a timely and supportive manner.11.Teach Effective Study Skills:Equip students with strategies for effective studying, such as time management,notetaking,and test preparation techniques.12.Model Excellence:As educators,set an example of excellence in your own work and behavior.Show passion for learning and a commitment to personal growth.13.Create a Safe and Inclusive Environment:Ensure that the classroom is a place where all students feel safe,respected,and included,regardless of their background or abilities.14.Involve Parents and the Community:Engage parents and the wider community in the educational process.Their support can provide additional resources and opportunities for students.15.Celebrate Successes:Recognize and celebrate the achievements of students,both big and small,to boost their confidence and motivation.By implementing these methods,educators can create a supportive and challenging environment that nurtures excellence in students,preparing them for success in their academic and future endeavors.。
A Robust Optimization Approach to Inventory theory
A Robust Oபைடு நூலகம்timization Approach to Inventory Theory
Sloan School of Management and Operations Research Center, Massachusetts Institute of Technology, E53-363, Cambridge, Massachusetts 02139, dbertsim@ Department of Industrial and Systems Engineering, Lehigh University, Mohler Building, Bethlehem, Pennsylvania 18015, aurelie.thiele@
1. Introduction
Optimal supply chain management has been extensively studied in the past with much theoretical success. Dynamic programming has long emerged as the standard tool for this purpose, and has led to significant breakthroughs as early as 1960, when Clark and Scarf (1960) proved the optimality of base-stock policies for series systems in their landmark paper. Although dynamic programming is a powerful technique as to the theoretical characterization of the optimal policy for simple systems, the complexity of the underlying recursive equations over a growing number of state variables makes it ill suited for the computation of the actual policy parameters, which is crucial for real-life applications. Approximation algorithms have been developed to address those issues. These include stochastic approximation (see Koshner and Clark 1978) and infinitesimal perturbation analysis (IPA) (see Glasserman 1991, Ho and Cao 1991), where a class of policies, e.g., base-stock, characterized by a set of parameters, is optimized using simulation-based methods (see Fu 1994, Glasserman and Tayur 1995, Kapuscinski and Tayur 1999). IPA-based methods assume knowledge of the underlying probability distributions and restrict their attention to cer150
Feedback on second language studnets writing
State-of-the-Art ArticleFeedback on second language students’writing Ken Hyland Institute of Education,University of London k.hyland@Fiona Hyland University of Hong Kongfhyland@hkucc.hku.hkFeedback is widely seen as crucial for encouraging and consolidating learning,and this significance has also been recognised by those working in thefield of second language(L2)writing.Its importance is acknowledged in process-based classrooms,where it forms a key element of the students’growing control over composing skills,and by genre-oriented teachers employing scaffolded learning techniques.In fact,over the past twenty years,changes in writing pedagogy and research have transformed feedback practices,with teacher written comments often supplemented with peer feedback,writing workshops, conferences,and computer-delivered feedback.But while feedback is a central aspect of ESL/EFL writing programs across the world,the research literature has not been unequivocally positive about its role in writing development,and teachers often have a sense that they are not making use of its full potential.In this paper we examine recent research related to feedback on L2 students’writing,focusing on the role of feedback in writing instruction and discussing current issues relating to teacher written and oral feedback,collaborative peer feedback and computer-mediated feedback.Feedback has long been regarded as essential for the development of second language(L2)writing skills,both for its potential for learning and for student motivation.In process-based,learner-centred classrooms,for instance,it is seen as an important developmental tool moving learners through multiple drafts towards the capability for effective self-expression.From an interactionist perspective it is regarded as an important means of establishing the significance of reader responses in shaping meanings (Probst1989).In genre classrooms feedback is a key element of the scaffolding provided by the teacher to build learner confidence and the literacy resources to participate in target communities.In fact,over the past twenty years,changes in writing pedagogy and insights gained from research studies have transformed feedback practices,with teacher written comments now often combined with peer feedback,writing workshops,oral-conferences,or computer-delivered feedback.Summative feedback, focusing on writing as a product,has generally been replaced or supplemented by formative feedback which points forward to the student’s future writing and the development of his or her writing processes.But while feedback is a central aspect of L2writing programs across the world,the research literature has not been unequivocally positive about its role in writing development,and teachers often have a sense that they are not making use of its full potential.Many questions relating to feedback remain unanswered or only partially addressed:Does it make a difference to students’writing?If so,in what areas?What is the best way of delivering feedback?Can error correction and form focused feedback have long term benefits on students’writing?Can technology play a greater part in delivering feedback?What role can peer feedback play in writing development?How far does culture play a part in student responses to feedback?How can teacher feedback enhance students’ability to independently reflect on their writing?What are the implications of feedback for teacher control and text appropriation?This paper reviews recent research which addresses these questions by focusing onKen Hyland is Professor of Education and Head of the Centre for Academic and Professional Literacies at the Institute of Education,University of London. He has considerable experience teaching and researching first and second language writing and has published over one hundred articles and ten books on writing and applied linguistics.Recent titles are T eaching and researching writing(Longman,2002),Second language writing(Cambridge University Press,2003), Genre and second language writing(University of Michigan Press,2004)and Metadiscourse (Continuum,2005).Professor Ken Hyland,School of Culture,Language and Communication,Institute of Education,University of London,20Bedford Way, London WC1H0AL,UK.k.hyland@Fiona Hyland has a Ph.D.from Victoria University of Wellington,New Zealand,on the subject of feedback in second language writing and is a lecturer in the Faculty of Education at the University of Hong Kong,where she coordinates the ELT specialism of the Masters in Education programme.She has25years of experience teaching and researching in the areas of TESOL and teacher education,and has published a number of papers on the subject of feedback to ESL writers.Dr Fiona Hyland,Faculty of Education,University of Hong Kong, Pok Fu Lam Road,Hong Kong.fhyland@hkucc.hku.hkLang.T each.39,77–95.doi:10.1017/S0261444806003399Printed in the United Kingdom c 2006Cambridge University Press77K.Hyland &F .Hyland■teacher written and oral feedback,peer conferencing and computer-mediated feedback.The volume of this research means that we are forced to focus on L2learners of English,although the issues are common to studies of learners of other languages.1.T eacher written feedbackDespite increasing emphasis on oral response and the use of peers as sources of feedback,teacher written response continues to play a central role in most L2and foreign language (FL)writing classes.Many teachers feel they must write substantial comments on papers to provide a reader reaction to students’efforts,to help them improve as writers and to justify the grade they have been given (K.Hyland 2003).Research in the 1980s and early 1990s,however,began to question the effectiveness of teacher feed-back as a way of improving students’writing.Early re-search on native English speakers (L1)suggested that much written feedback was of poor quality and was frequently misunderstood by students,being vague,inconsistent and authoritarian,overly concerned with error and often functioning to appropriate,or take over,student texts by being too directive (e.g.Knoblauch &Brannon 1981;Connors &Lunsford 1993;see also Ferris (2003:chapter 1)for a review).While Zamel (1985)painted a similarly bleak picture in L2contexts,it is important to note that feedback research was in its infancy at that time and ideas of best practice in both giving feedback and designing studies to describe it were fairly rudimentary.More recent empirical research suggests that feedback does lead to writing improvements and this section highlights this research.1.1Responding to errorA substantial amount of the research on teacher written feedback in L2writing contexts has been concerned with error correction and whether this be-nefits students’writing development.Research in this area has sought to explore whether error correction is effective and what strategies and treatments teachers use for error correction,and to discover the effects correction has on students’immediate revisions and their longer term development as writers.One line of argument,influenced by process theories,claims that feedback on error to L2students is discouraging and generally fails to produce any improvements in their subsequent writing (Robb,Ross &Shortreed 1986;Kepner 1991;Sheppard 1992;Polio,Fleck &Leder 1998;Fazio 2001).In a well-known summary of this literature,Truscott (1996)saw very little benefit in this kind of feedback and argued strongly that it was the responsibility of teachers to change student attitudes regarding what they should expect from teacher response by adopting a ‘correction-free approach’in their classrooms (Truscott 1996,1999).Recently,he responded to Chandler’s (2003)experimental study,which found that the accuracy of students’writing improved significantly over a semester when they corrected their errors after feedback than when they did not.Truscott (2004:342)queried Chandler’s findings and reiterated that error correctionjust be ineffective,but even harmful to and their overall writing quality,arguing time spent dealing with errors in class is better spent on additional writing practice.T eachers have been reluctant to follow this advice,however,as they are acutely aware that accuracy in writing is important to academic and professional audiences and that so-called ‘L2errors’often stigmatize writers (Horowitz 1986;Johns 1995;James 1998).They also feel that they should respond to the needs of students themselves,who see error-free work as very important.Research on student preferences has consistently found that students expect teachers to comment on their written errors and are frustrated if this does not happen (Cohen &Cavalcanti 1990;Leki 1991;Hedgcock &Lefkowitz 1994;Cumming 1995;Ferris 1995;F .Hyland 1998;Ferris &Roberts 2001;Lee 2004).In addition,it has been argued that the research on the lack of effectiveness of error feedback is not nearly as conclusive as Truscott claims (e.g.Polio 1997;Ferris 1999,2002,2003,2004).Statements that teachers’error feedback is often incomplete,arbitrary and inaccurate,for instance (e.g.Zamel 1985;Connors &Lunsford 1993),have simply not been consistently demonstrated empirically .Clearly,teacher variation is a crucial issue in both feedback and in the design of classroom research and little can be said on the basis of a few studies.Lee’s (2004)finding that half of her sample of Hong Kong teachers corrected errors inaccurately ,for example,was based on a de-contextualised teacher correction task,while Ferris’(2006)more naturalistic in situ study found teacher feedback to be overwhelmingly accurate.Further research into the linguistic knowledge and backgrounds of teachers whose first language is not English may go some way to help explain such variations in feedback practices.In fact,it is difficult to draw clear conclusions and generalizations from the literature as a result of varied populations,treatments and research designs.As we shall discuss below ,written feedback is more than marks on a page,yet research procedures often remove it from the real classrooms and teacher-student relationships within which it occurs (see F .Hyland &K.Hyland 2001;Goldstein 2005).Moreover,while marking mechanical errors can be frustrating,the view that there is no direct connection between correction and learning is greatly overstated.Master (1995),for instance,found that corrective feedback was effective when combined with classroom discussions.Fathman &Whalley (1990)found positive effects for rewriting from78■Feedback on L2students’writingfeedback on both grammar and content and Ferris (2006)discovered that about80%of students in her L2sample were able to successfully edit errors marked by teachers in a subsequent draft,with only10% making incorrect changes.But demonstrating that a student can utilize teacher feedback to successfully edit from one draft of a paper to the next tells us little about the learner’s successful acquisition of the linguistic features addressed by the feedback(Truscott 1996).The few studies that have looked beyond the immediate corrections in a subsequent draft,how-ever,have noted improvements in students’language accuracy(Polio et al.1998;F.Hyland2003;Chandler 2003).Ferris(2006),for instance,showed that students made statistically significant reductions in their total number of errors over a semester infive major grammar categories with a particular reduction in verb and lexical errors.These results underline the importance of general language proficiency and metalingusitic awareness in writing development and support Y ates and Kenkel’s(2002)argument that both error correction and its effectiveness must be seen in the context of a student’s evolving mastery of overall text construction.It is also worth pointing out that many studies of feedback on error have ignored how language acquisition occurs,although the influence of feed-back on the learner’s long term writing development fits closely with the SLA research(Goldstein,p.c). SLA studies indicate that second language acquisition takes place gradually over time and that mistakes are an important part of the highly complex develop-mental process of acquiring the target language.In fact,there may be a U-shaped course of development (Ellis1997)where learners are initially able to use the correct forms,only to regress later,before finally using them according to the target language norms(e.g.Doughty&Long2003).W e cannot, in other words,expect that a target form will be acquired either immediately or permanently after it has been highlighted through feedback.Even though explicit feedback can play an important role in second language acquisition,it needs time and repetition before it can help learners to notice correct forms,compare these with their own interlanguage and test their hypotheses about the target language. Attempting to establish a direct relationship between corrective feedback and successful acquisition of a form is,therefore,over-simplistic and highly problematic(e.g.Carson2001;Ferris2003).While feedback alone will not be responsible for improvement in language accuracy,it is likely to be one important factor.One key variable here is the type of error feedback that is given,and a number of researchers have compared direct feedback,where the teacher makes an explicit correction,with indirect forms where he or she simply indicates that an error has been made by means of an underline,circle,code,etc.The role of explicitness in student uptake,or response to feedback,is important as while indirect error feedback may encourage learner reflection and self-editing(Lalande1982),lower proficiency students may be unable to identify and correct errors even when they have been marked for them(Ferris &Hedgcock2005).Findings on feedback type have been conflicting, largely due to the widely varying student populations, types of writing and feedback practices examined and the diverse research designs lande (1982),for instance,reported a reduction in student errors with indirect feedback and Robb et al.(1986) discovered minimal long-term gains in accuracy compared with direct feedback practices.In a textual study of over5,000teacher comments,Ferris(2006) found that students utilized direct feedback more consistently and effectively than indirect types,partly as it involves simply copying the teacher’s suggestion into the next draft of their papers.However,less explicit forms of feedback also led to accurate revisions most of the time and this occurred whether underlined errors were coded or not.Ferris notes, however,that students’short-term ability to edit some types of errors which were directly marked by feedback did not always translate into long-term improvement,while indirect feedback seemed to help them develop more over time.While this may be a discouragingfinding for many teachers looking for evidence that their students are becoming more proficient writers,the importance of immediate improvement of drafts cannot be underestimated. Another issue is whether different types of errors ‘respond’differently to error treatment.Ferris(1999) suggests that some errors,such as problems with verbs,subject-verb agreement,run-ons,fragments, noun endings,articles,pronouns,and possibly spelling,can be considered‘treatable’,because they ‘occur in a patterned,rule-governed way’.In contrast,errors such as word choice and word order are‘untreatable’,in that‘there is no handbook or set of rules students can consult to avoid orfix those types of errors’(1999:6).T eachers tend to mark ‘treatable’errors indirectly and‘untreatable’errors directly(Ferris2006)and this is probably because they believe that students are unable to self-correct untreatable errors marked indirectly(Ferris2006). Moreover,while students seem to be able to improve their language accuracy through feedback on form if they are taught the rules governing directly‘treatable’errors(Ferris1999),idiosyncratic errors are more amenable to indirect feedback techniques,such as locating the type of error and asking students to correct it themselves(Ferris&Roberts2001). Overall,students appear to attend to teacher error corrections and in most cases use them to make accurate changes in their texts.This seems to facilitate student writing improvement both in the short term and over time,although it must be admitted79K.Hyland &F .Hyland■that longitudinal studies rarely span more than one semester.Improvements seem to be more likely if feedback is related to instruction and if indirect feedback methods are used.This last point might relate to SLA research which suggests that students need to invest more effort in processing the input they receive and are forced to notice discrepancies in their own work and the correct pattern they are trying to employ (see Gass &Selinker 2001;Mitchell &Myles 2004).1.2Teacher stances and feedback practicesAnother key area of investigation has been the stance teachers take towards students’texts and the relationship they build with their learners when giving feedback.It has long been recognised that teachers approach texts with a number of different purposes in mind and that these may change with different assignments,different students and different drafts (Bates,Lane &Lange 1993).Thus commentary on a draft is likely to serve more immediate pedagogical goals than that given on a final product,for instance,and process approaches mandate that teachers should comment on ideas in earlier drafts and on grammar in later drafts (e.g.Zamel 1985).Several researchers have observed,however,that because meaning is only realised through language,the content-form distinction creates a false dichotomy (K.Hyland 2003;Ferris &Hedgcock 2005),and research has shown that varying this recommended ‘meaning before form’pattern seems to make little difference to the quality of final products (Ashwell 2000).Outside the language class,of course,feedback is less concerned with the development of writing proficiency and more with appraising how students have processed content:writing is merely a medium by which students are judged on what they know of specific subject knowledge.T eachers,moreover,do not simply respond to grammar or content but have other purposes.T eachers adopt various commenting strategies which vary according to the type of essay assigned,the point of the semester in which feedback is given,and the proficiency of the student (Ferris 1997;Ferris et al.1997).Ferris et al.(1997),for instance,distinguished eight broad functions of response in 1500teacher comments,ranging from ‘Asking for unknown information’to ‘Giving information on ideas’.More simply,F .Hyland &K.Hyland (2001)and K.Hyland &F .Hyland (2006b)collected these purposes under the overarching functions of praise,criticism and suggestions,where suggestion and criticism are opposite ends of a continuum ranging from a focus on what is done poorly to a plan of action for its improvement.W e should also note that written feedback is not purely informational,for although the commentary may facilitate writing development it will only be effective if it engages with the writer.One increasingly studied social factor of this kind has been the ways teachers seek to structure activities through a course to ensure that students are able to interpret and use their comments effectively.Several studies have shown that feedback is not simply disembodied reference to student texts but an interactive part of the whole context of learning,helping to create a productive interpersonal relationship between the teacher and individual students (F .Hyland 1998;Conrad &Goldstein 1999;F .Hyland &K.Hyland 2001).F .Hyland (1998,2000a)and K.Hyland &F .Hyland (2006b),for instance,observed a close relationship between written and oral feedback and instruction,finding that the points made through explicit teaching were picked up and reinforced by written feedback and then recycled in both peer and student-teacher oral interactions.One way of establishing this link has been to encourage students to revise papers based on feedback and incorporate both final versions and drafts in a portfolio (Song &August 2002;Hamp-Lyons 2006).This allows teachers to observe students’development,improve their own practices by judging students’uptake of feedback,and create a closer relationship with learners.In addition,by using mechanisms such as cover sheets (Leki 1990),questionnaires,or writer autobiographies (Goldstein 2004)teachers are able to reveal student feedback preferences to enhance the effectiveness of their commentary and the students’revisions.In a longitudinal study,K.Hyland &F .Hyland (2006b)found that teachers did not simply mark a text but used information about the student to contextualise the writing being done,the strengths and weaknesses of the individual student,and his or her explicit requests for particular kinds of help.T o assist them,the two teachers studied not only made considerable use of the data provided in the cover sheets that students submitted with each piece of writing,but also considered the personality and possible response of the individual student to specific feedback points.This vividly illustrates that teachers do not give feedback in a vacuum but create a context for their remarks,making use of what they know of the writer to create an interpersonal link and target feedback to their personality and needs.Another aspect of the teacher-learner relationship now beginning to receive attention is how teachers select appropriate language and style in their feedback to construct the kinds of relationships which can facilitate a student’s writing development (F .Hyland &K.Hyland 2001;K.Hyland &F .Hyland 2006b).Essentially ,teachers have to weigh their choice of comments to accomplish informational,pedagogic,and interpersonal goals simultaneously while taking account of likely student reactions.Negative feedback may have a detrimental effect on writer confidence80■Feedback on L2students’writingwhile premature and gratuitous praise can confuse students and discourage revisions(F.Hyland&K. Hyland2001).T eachers often praise frequently as a means of building students’confidence,but students expect to receive constructive criticism rather than simple platitudes(Ferris1995;F.Hyland1998).T eachers also seek to mitigate the full force of their criticisms and suggestions by the use of hedges, question forms,and personal attribution and to be more reticent about criticising students’ideas than their language choices(F.Hyland&K.Hyland 2001;K.Hyland&F.Hyland2006b).However, while mitigation may foster a cooperative pedagogical environment,its indirect approach also carries the very real danger that students may miss the point of the comment or misinterpret the feedback(F.Hyland 2001b;K.Hyland&F.Hyland2006b).The choice between promoting positive affect or confronting writing weaknesses may be decided by targeting some errors only and leaving others to later drafts or assignments,but this certainly presents a serious dilemma for writing teachers who have to rely mainly on their experience and knowledge of their students.1.3Student views on teacher feedback Attempts have been made tofind out more about students’perspectives on teacher response,mainly through questionnaire research.Surveys of students’feedback preferences generally indicate that ESL students greatly value teacher written feedback and consistently rate it more highly than alternative forms such as peer and oral feedback(Radecki&Swales 1988;Leki1991;Enginarlar1993;Saito1994;Ferris 1995;Zhang1995).Although most surveys show that students want teacher feedback to highlight their grammatical errors,some indicate that they also want teachers to give them feedback on the content and ideas in their writing.(Hedgcock&Lefkowitz 1994,1996).Studies also suggest that students like to receive written feedback in combination with other sources,including conferences(Arndt1993; Hedgcock&Lefkowitz1994)and are positive about receiving indirect feedback on errors,giving them clues rather than corrections since they recognize that it encourages them to be more active in their use of feedback(Arndt1993;Saito1994;F.Hyland2001a). While most research has focused on feedback given by EFL or ESL writing teachers at undergraduate or pre-university levels,recent research has focused on feedback to L2graduate students,particularly by disciplinary faculty.Prior’s(1995,1998)studies,for instance,looked at feedback from peers and faculty on writing in graduate geography and sociology courses,showing how the writing tasks and the faculty’s response were shaped by the experiences, activities and goals the participants brought to and created within the seminars.Riazi’s(1997)study of four Iranian graduate students in education showed that the students viewed feedback as important for improving their understanding of their discipline,but also saw form-based comments as a way of developing their L2.It may be,however,that students receive fewer form-focused comments than they wish.Zhu’s (2004)survey,for example,suggests that faculty saw themselves primarily as providers of content-based summative feedback and regarded formative feedback on writing as the job of writing instructors.Leki(2006)has looked at feedback given by faculty to L2graduate students in a US university,analysing the written comments made by disciplinary faculty on student assignments and interviewing students to investigate their opinions about the value of written feedback in their development of disciplinary literacy. Most students reported that they found the feedback very useful but many also said they would have liked even more,especially feedback helping them to identify problems and giving them information about academic and disciplinary expectations.They also wanted feedback on how native speakers would express the same ideas,suggesting that–like Riazi’s students–they wanted their feedback to have a dual content/language focus.There is a need for more investigations to address questions on how L2 instructors and disciplinary faculty can work together more closely to meet students’needs.1.4Impact of teacher written feedbackon students’writingAlthough L2students themselves are positive about teacher written feedback,its contribution to writing development is still unclear,both in terms of its immediate impact on revisions to drafts and of the longer term development of their writing skills. Studies suggest that students may ignore or misuse teacher commentary when revising drafts.Sometimes they misunderstand it(Ferris1995;Conrad& Goldstein1999),or they understand the problems pointed out but are unable to come up with a suitable revision(Ferris1997;Conrad&Goldstein1999), and sometimes this causes them to simply delete the offending text to avoid the issues raised(F.Hyland 1998).There is also the issue of whether student revisions in response to feedback improve their writing.Research is not conclusive on this as it is difficult to claim a direct causal relationship between feedback and revision since both take place within a complex of contextual factors which can influence the extent and success of revision after feedback(Conrad&Goldstein1999;Goldstein 2006).Research into feedback,however,‘has largely been non-contextual and non-social,focused largely on texts and conducted within a linear model of teacher respond and student revise’(Goldstein2001: 77).81K.Hyland &F .Hyland■Much of the research on error correction reflects experimental or analytical research techniques that ignore classroom realities and the preferences of students (e.g.Frantzen 1995;Polio et al.1998;Ferris &Roberts 2001;Chandler 2003).Such approaches have their place,but they cannot capture the impact that wider classroom,institutional and personal contexts have on the ways feedback is given,understood and negotiated between participants.Sociocultural perspectives on learning therefore see knowledge and understanding not as things that can be handed down but as constructed through interactive processes (Riazi 1997;Murphy 2000;Goldstein 2006;Villamil &de Guerrero 2006).T eachers respond to students in their comments as much as texts,and experienced teachers often tailor their feedback to suit each student,considering their backgrounds,needs and preferences as well as the relationship they have with them and the ongoing dialogue between them (Ferris et al.1997;F .Hyland 1998,2001b).It may be,in other words,that what is effective feedback for one student in one setting is less so in another.Context is a combination of factors related to the institution and writing programme as well as those that teachers and students bring to the interaction (F .Hyland 1998,2000a;K.Hyland &F .Hyland 2006b).Goldstein (2004,2005),for instance,suggests that contextual factors can include socio-political issues that influence teacher status and morale,available resources and class size,institutional attitudes towards L2writers,exams,and program philosophies about feedback.Similarly,teacher factors such as attitudes towards particular students or the content of their texts,and student factors like reactions to teacher feedback and their investment in the course can have an impact on feedback and revision.These factors have not been systematically incorporated into most feedback research designs,however,and we know little about their potential impact on feedback and student revision.Attempts have been made,however,to link aspects of teacher feedback with student revision.Ferris (1997),for instance,looked at 110first and second drafts of papers by ESL tertiary students and found that three quarters of substantive teachers’commentsQ1on the drafts were used by the students and they tended to improve student papers.V ery few of the changes (less than 5%)actually had a negative effect,but revisions based on comments in the question form were judged as having mixed effects.Longer and more detailed text specific comments were also found to result in more positive revisions.Researchers have also tried to find out what kinds of comments are most effective.Ferris &Hedgcock (2005)suggest text-specific commentary is most likely to encourage revision.They also make the point that marginal comments have more immediacy and make it easier for students to locatethe source of a problem and revise appropriately whereas end comments can be more useful for writing development,since they summarize major problems.Marginal comments are also considered to be more motivating since they show the reader actively engaged with the writer’s text (Goldstein 2004).Conrad &Goldstein (1999)also question the importance often attributed to comment type in revision success.Their findings from three case studies suggest that choosing to phrase comments as questions,declaratives or imperatives had far less impact on the success of revisions than the type of problem identified in the feedback.Problems dealing with facts and details in the content were revised successfully about half the time,while those dealing with argumentation and analysis were revised successfully only 10%of time,a finding supported by Ferris (2001).Problems dealing with facts and details in the content were revised successfully about half the time,while those dealing with argumentation and analysis were revised successfully only 10%of time,a finding supported by Ferris (2001).A final key issue of students’responses to teacher feedback is that of ‘text appropriation’,or the idea that ownership of writing can be ‘stolen’from a writer by the teacher’s comments.L1writing researchers have suggested that writers might follow directive comments too closely and lose the oppor-tunity to develop as writers by merely rewriting their texts to reflect their teachers’preoccupations (Knoblauch &Brannon 1984).These concerns have been raised in ESL discussions of feedback but have been rigorously questioned by Reid who suggested that text appropriation was ‘largely a mythical fear of ESL writing teachers’(1994:275).She pointed out the danger of confusing helpful intervention with appropriation and urged ESL teachers to focus instead on their roles as ‘cultural informants and as facilitators for creating the social discourse community in the ESL writing classroom’(p.275).Appropriation and the socio-political aspects of giving feedback do,however,continue to be debated in the literature.F .Hyland (2000a)describes episodes where teachers overrode student decisions on use of feedback,raisingto not just to the ownership of writing to the ownership and control of the revision processes,supporting Hall’s (1995)point that appropriation may involve both concepts of good and bad writing products and good and bad writing behaviour.Recently ,the concept of appropriation has been redefined with the suggestion that appropriation can go in two directions.Appropriation of teacher feedback can be an active strategy used by novice academic writers as they develop their own voices and their familiarity with different genres.Tardy (2006)illustrates the complexity of this in her description of Chatri,an engineering doctoral student,whose sense of agency and ownership of his writing developed as82。
数据挖掘专业词汇中英对照
1、Bilingual 双语Chinese English bilingual text 中英对照2、Data warehouse and Data Mining 数据仓库与数据挖掘3、classification 分类systematize classification 使分类系统化4、preprocess 预处理The theory and algorithms of automatic fingerprint identification system (AFIS) preprocess are systematically illustrated.摘要系统阐述了自动指纹识别系统预处理的理论、算法5、angle 角度6、organizations 组织central organizations 中央机关7、OLTP On-Line Transactional Processing 在线事物处理8、OLAP On-Line Analytical Processing 在线分析处理9、Incorporated 包含、包括、组成公司A corporation is an incorporated body 公司是一种组建的实体10、unique 唯一的、独特的unique technique 独特的手法11、Capabilities 功能Evaluate the capabilities of suppliers 评估供应商的能力12、features 特征13、complex 复杂的14、information consistency 信息整合15、incompatible 不兼容的16、inconsistent 不一致的Those two are temperamentally incompatible 他们两人脾气不对17、utility 利用marginal utility 边际效用18、Internal integration 内部整合19、summarizes 总结20、application-oritend 应用对象21、subject-oritend 面向主题的22、time-varient 随时间变化的23、tomb data 历史数据24、seldom 极少Advice is seldom welcome 忠言多逆耳25、previous 先前的the previous quarter 上一季26、implicit 含蓄implicit criticism 含蓄的批评27、data dredging 数据捕捞28、credit risk 信用风险29、Inventory forecasting 库存预测30、business intelligence（BI）商业智能31、cell 单元32、Data cure 数据立方体33、attribute 属性34、granular 粒状35、metadata 元数据36、independent 独立的37、prototype 原型38、overall 总体39、mature 成熟40、combination 组合41、feedback 反馈42、approach 态度43、scope 范围44、specific 特定的45、data mart 数据集市46、dependent 从属的47、motivate 刺激、激励Motivate and withstand higher working pressure个性积极，愿意承受压力.敢于克服困难48、extensive 广泛49、transaction 交易50、suit 诉讼suit pending 案件正在审理中51、isolate 孤立We decided to isolate the patients.我们决定隔离病人52、consolidation 合并So our Party really does need consolidation 所以，我们党确实存在一个整顿的问题53、throughput 吞吐量Design of a Web Site Throughput Analysis SystemWeb网站流量分析系统设计收藏指正54、Knowledge Discovery（KDD）55、non-trivial(有价值的）--Extraction interesting (non-trivial(有价值的), implicit（固有的）, previously unknown and potentially useful) patterns or knowledge from huge amounts of data.56、archeology 考古57、alternative 替代58、Statistics 统计、统计学population statistics 人口统计59、feature 特点A facial feature 面貌特征60、concise 简洁a remarkable concise report 一份非常简洁扼要的报告61、issue 发行issue price 发行价格62、heterogeneous (异类的)--Constructed by integrating multiple, heterogeneous (异类的)data sources63、multiple 多种Multiple attachments多实习64、consistent（一贯）、encode（编码）ensure consistency in naming conventions,encoding structures, attribute measures, etc.确保一致性在命名约定，编码结构，属性措施，等等。
An Adaptive Virtual Queue (AVQ) Algorithm for Active Queue Management
An Adaptive Virtual Queue(A VQ)Algorithm for ActiveQueue ManagementSrisankar S.Kunniyur,Member,IEEE,and R.Srikant,Senior Member,IEEEAbstract—Virtual Queue-based marking schemes have been recently proposed for AQM(Active Queue Management)in Internet routers.We consider a particular scheme,which we call the Adaptive Virtual Queue (A VQ),and study its following properties:stability in the presence of feed-back delays,its ability to maintain small queue lengths and its robustness in the presence of extremely shortflows(the so-called web mice).Using a linearized model of the system dynamics,we present a simple rule to design the parameters of the A VQ algorithm.We then compare its performance through simulation with several well-known AQM schemes such as RED, REM,PI controller and a non-adaptive virtual queue algorithm.With a view towards implementation,we show that A VQ can be implemented as a simple token bucket using only a few lines of code.Keywords—Active Queue Management,Internet congestion control, ECN Marking1IntroductionIn the modern day Internet,there has been a strong demand for QoS and fairness amongflows.As a result,in addition to the sources,the links are also forced to play an active role in conges-tion control and avoidance.Random Early Discard(RED)[1] was originally proposed to achieve fairness among sources with different burstiness and to control queue lengths.RED allows for dropping packets before buffer overflow.Another form of congestion notification that has been discussed since the advent of RED is Explicit Congestion Notification(ECN)[2].ECN has been proposed to allow each link to participate in conges-tion control by notifying users when it detects an onset of con-gestion.Upon detecting incipient congestion,a bit in the packet header is set to one for the purpose of notifying the user that a link on its route is experiencing congestion.The user then reacts to the mark as if a packet has been lost.Thus,the link avoids dropping the packet(thereby enhancing goodput)and still man-ages to convey congestion information to the user.To provide ECN marks or drop packets in order to control queue lengths or provide fairness,the routers have to select pack-ets to be marked in a manner that conveys information about the current state of the network to the users.Algorithms that the routers employ to convey such information are called Active Queue Management(AQM)schemes.An AQM scheme might mark or drop packets depending on the policy at the router.In this paper,we use the term“marking”more generally to refer to any action taken by the router to notify the user of incipient congestion.The action can,in reality,be ECN-type marking or dropping(as in RED)depending upon the policy set for the S.S.Kunniyur is with the Dept.of Electrical and Systems Engineering,University of Pennsylvania.Email:kunniyur@R.Srikant is with the Department of General Engineering and Coordinated Science Lab-oratory,University of Illinois at Urbana-Champaign.Email:rsrikant@ Research supported by DARPA Grant F30602-00-2-0542and NSF Grants ANI-9813710 and NCR-9701525.router.As in earlier work on studying AQM schemes[3,4,5], this distinction is blurred in the mathematical analysis to allow for the development of simple design rules for the choice of AQM parameters.However,our simulations consider marking and dropping schemes separately.Designing robust AQM schemes has been a very active re-search area in the Internet community.Some AQM schemes that have been proposed include RED[1],a virtual queue-based scheme where the virtual capacity is adapted[6,7],SRED[8], Blue[9],Proportional Integral(PI)controller[4],REM[10], a virtual queue based AQM scheme[11](which we refer to as the Gibbens-Kelly Virtual Queue,or the GKVQ scheme)among others.While most of the proposed AQM schemes detect con-gestion based on the queue lengths at the link(e.g.,RED),some AQM schemes detect congestion based on the arrival rate of the packets at the link(e.g.,virtual queue-based schemes)and some use a combination of both(e.g.,PI).Also,most of the AQM schemes involve adapting the marking probability(as noted be-fore we use the term marking to refer to both marking and drop-ping)in some way or the other.An important question is how fast should one adapt while maintaining the stability of the sys-tem?Here the system refers jointly to the TCP congestion con-trollers operating at the edges of the network and the AQM schemes operating in the interior of the network.Adapting too fast might make the system respond quickly to changing net-work conditions,but it might lead to large oscillatory behavior or in the worst-case even instability.Adapting too slowly might lead to sluggish behavior and more losses or marks than desired, which might lead to a lower throughput.In this paper,we start by presenting an implementation of a virtual-queue based AQM scheme,namely the Adaptive Virtual Queue(A VQ).The motivation behind the A VQ algorithm is to design an AQM scheme that results in a low-loss,low-delay and high utilization operation at the link.We then discuss a method-ology forfinding the fastest rate at which the marking probabil-ity adaptation can take place,given certain system parameters like the maximum delay and the number of users,so that the system remains stable.We note that the marking probability in A VQ is implicit,no marking probability is explicitly calcu-lated and thus,no random number generation is required.On the other hand,we replace the marking probability calculation with the computation of the capacity of a virtual queue.Moti-vated by the success of the analysis and design of other AQM schemes in[3,4,5],we consider a single router accessed by many TCP sources with the same round-trip time(RTT)and use a control-theoretic analysis to study the stability of this system. The A VQ algorithm maintains a virtual queue whose capac-ity (called virtual capacity )is less than the actual capacity of the link.When a packet arrives in the real queue,the virtual queue is also updated to reflect the new arrival.Packets in the real queue are marked/dropped when the virtual buffer over-flows.The virtual capacity at each link is then adapted to en-sure that the total flow entering each link achieves a desired utilization of the link.This was originally proposed in [6]as a rate-based marking scheme.In the absence of feedback de-lays,it was shown in [7]that a fluid-model representation of the above scheme,along with the congestion-controllers at the end-hosts,was semi-globally asymptotically stable when the link adaptation is sufficiently slow.An appealing feature of the A VQ scheme is that,in the absence of feedback delays,the system is fair in the sense that it maximizes the sum of utilities of all the users in the network [7].Combining this with a result in [6]which shows that a TCP user with an RTT of d r can be approxi-mated by a user with a utility function −1d 2r x r ,where x r is the rate of the TCP user,shows that the network as a whole converges to an operating point that minimizes r −1d 2r xr.This utility func-tion called the potential delay was introduced as a possible fair-ness criterion in [12].The throughput under this utility functionis given by 1/d r √p r ,where p r is the loss probability seen by User r which is consistent with the models in [3,4,5].While we use this simplified model for analysis in the paper,our sim-ulations in ns-2use TCP-Reno,including slow-start,time-out,fast retransmit,etc.A slightly more refined utility function is used in [13]and the results in this paper can be easily modified to incorporate that utility function.We start with a fluid-model representation of the TCP flow-control,along with the A VQ scheme that was proposed in [6].However,here we explicitly consider the feedback delay due to the RTT of each user and thus,we obtain a set of delay-differential equations.We linearize this system and obtain con-ditions for local stability in terms of the round-trip delay,the number of users,the utilization of the link and a smoothing pa-rameter in the update equation of the A VQ scheme.The rest of the paper is organized as follows:in Section 2,we present an implementation of the A VQ algorithm and provide design rules for the stability of the A VQ and TCP together.In Section 3,we provide detailed ns-2simulations to validate our design rules and also compare the A VQ algorithm with RED,REM,GKVQ and the PI controllers.The PI controller is somewhat similar to A VQ in that it adapts the marking probability in a manner similar to the virtual capacity adaptation in the A VQ scheme,but it depends on the queue size at the link.As a result,for small buffers the system tends to perform poorly.Also,since the marking probability is directly modified and this update has to be slow enough for system stability,the scheme exhibits slug-gishness when short flows are introduced.This is the subject of simulations in Section 3.Since the analytical model does not capture the discrete packet behavior of the network in ad-dition to the slow-start and the timeout characteristics of TCP,the simulations in Section 3are intended to demonstrate the per-formance of the A VQ algorithm under these realistic conditions.In Section 4,we provide theoretical justification for the design rules in Section 2.The design rules in Section 4are less restric-tive than the design rules in [14].Conclusions are provided in Section5.− Real Queue − Virtual QueueBefore packet arrival After packet arrivalFigure 1:A VQ:When Virtual Queue is notfull− Real Queue Before packet arrivalAfter packet arrival− Virtual Queue− Marked PacketFigure 2:A VQ:When the incoming packet is dropped from the Virtual Queue2The A VQ AlgorithmLet C be the capacity of a link and γbe the desired utilization at the link.The A VQ scheme,as presented in [6,7],at a router works as follows:•The router maintains a virtual queue whose capacity ˜C≤C and whose buffer size is equal to the buffer size of the real queue.Upon each packet arrival,a fictitious packet is enqueued in the virtual queue if there is sufficient space in the buffer (See Figure 1).If the new packet overflows the virtual buffer,then the packet is discarded in the virtual buffer and the real packet is marked by setting its ECN bit or the real packet is dropped,depending upon the conges-tion notification mechanism used by the router (See Fig-ure 2).•At each packet arrival epoch,the virtual queue capacity is updated according to the following differential equation:˙˜C=α(γC −λ),(1)where λis the arrival rate at the link and α>0is thesmoothing parameter.The rationale behind this is that mark-ing has to be more aggressive when the link utilization ex-ceeds the desired utilization and should be less aggressive when the link utilization is below the desired utilization.We now make the following observations.No actual enqueue-ing or dequeueing of packets is necessary in the virtual queue,we just have to keep track of the virtual queue length.Equation (1)can be thought of as a token bucket where tokens are gener-ated at rate αγC up to a maximum of C and depleted by each arrival by an amount equal to αtimes the size of the packet (See Figure 3).DefineαCapacityVirtual capacityTokens are generated at the ratetokens are removedα(Desired Utilization)When a packet of size b arrives, b Figure 3:A VQ:Token bucket implementation of the virtual ca-pacity adaptationB =buffer sizes =arrival time of previous packet t =Current timeb =number of bytes in current packetV Q =Number of bytes currently in the virtual queueThen,the following pseudo-code describes an implementa-tion of A VQ scheme:The A VQ AlgorithmAt each packet arrival epoch doV Q ←max(V Q −˜C (t −s ),0)/∗Update Virtual QueueSize ∗/If V Q +b >BMark or drop packet in the real queue elseV Q ←V Q +b /∗Update Virtual Queue Size ∗/endif˜C =max(min(˜C +α∗γ∗C ∗(t −s ),C )−α∗b,0)/∗Update Virtual Capacity ∗/s ←t /∗Update last packet arrival time ∗/We note the following features of the A VQ scheme:1.The implementation complexity of the A VQ scheme is com-parable to RED.RED performs averaging of the queue length,dropping probability computation and random num-ber generation to make drop decisions.We replace these with the virtual capacity calculation in A VQ.2.A VQ is a primarily a rate-based marking,as opposed to queue length or average queue length based marking.This provides early feedback,the advantages of which have been explored by Hollot et al [4,5],which was also mentioned in Kelly et al [15].3.Instead of attempting to regulate queue length as in RED,PI controller or recent versions of REM,we regulate uti-lization.As we will see in simulations,this is more robust to the presence of extremely short flows or variability in the number of long flows in the network.The reason is that,when utilization is equal to one,variance introduced by the short flows seems to lead to an undesirable transient behav-ior where excessively large queue lengths persist over longperiods of time.4.Unlike the GKVQ algorithm [11],we adapt the capacityof the virtual queue.A fixed value of ˜Cleads to a utiliza-tion that is always smaller than ˜C/C and it could be muchsmaller than this depending on the number of users in the system.Our marking mechanism is also different in that we do not mark until the end of a busy period after a con-gestion episode.5.There are two parameters that have to be chosen to imple-ment A VQ:the desired utilization γand the damping fac-tor α.The desired utilization γdetermines the robustness to the presence of uncontrollable short flows.It allows an ISP to trade-off between high levels of utilization and small queue lengths.Both the parameters αand γdetermine the stability of the A VQ algorithm and we provide a simple design rule to choose these parameters.The starting point for the analysis of such a scheme is the fluid-model of the TCP congestion-avoidance algorithm as pro-posed in [6].A theoretical justification of how a stochastic discrete-time equation can be approximated by a fluid-model is shown in [16].We then incorporate the virtual capacity update equation with this model and study the stability of the entire system under linearization.Consider a single link of capacity C with N TCP users travers-ing it.Let the desired utilization of the link be γ≤1,and let d be the round-trip propagation delay of each user (we assume that all users have the same round-trip propagation delay).Let x i (t )be the flow-rate of user i at time t.We will model the TCP usersusing the −1d 2x utility function as proposed in [6].For the sake of simplicity and tractability,we will neglect slow-start and time-outs when modeling the TCP users.We will later show through simulations that even with slow-start and timeouts,stability is maintained.Let p (.,.)be the fraction of packets marked at the link.The fraction of packets marked (i.e.,p (.,.))is a function of the total arrival rate at the link as well as the virtual capacity of the link.The congestion-avoidance algorithm of TCP useri can now be represented by the following delay differential equation:˙x i =1d −βx i (t )x i (t −d )p (N j =1x j (t −d ),˜C (t −d )),(2)where β<1and ˜Cis the virtual-capacity of the link.A βvalue of 2/3would give us the steady-state throughput of TCP as 1d32p ∗,where p ∗is the steady-state marking probability which is consistent with the results in [17].Hence,we will use β=2/3,in all our calculations.Also,note that on substitutingx i ≈W id ,where W i is the window-size of user i,we recover the TCP window control algorithm [6,3].The update equation at each link can now be written as:˙˜C=α(γC −λ),(3)where λ= Nj =1x j is the total flow into the link and α>0is the smoothing parameter.Note that αdetermines how fast one adapts the marking probability at the link to the changing network conditions.We will present a design rule that specifies how to choose αfor a given feedback delay (d ),utilization (γ)and a lower bound on the number of users (N ).In fact,as we will show in Section 4,one can derive bounds on any of the four parameters α,γ,N or d,given the other three using the same design rule.However,in practice,it would seem most natural to choose αgiven the other three parameters.Let x ∗i ,λ∗,˜C ∗and p ∗denote the equilibrium values of x i ,λ,˜Cand p (λ,˜C ).The equilibrium point of the non-linear TCP/AQM model is given by:ix ∗i=λ∗=γC x ∗i=γC N p ∗=p (γC,˜C∗)=N 2β(dγC ).Let us assume thatλ(t )=λ∗+δλ(t )˜C(t )=˜C∗+δ˜C (t ).The linearized version of the non-linear TCP/AQM model can now be written as:˙δλ=−K 11δλ(t )−K 12δλ(t −d )+K 2δ˜C (t −d )(4)˙δ˜C =−αδλ(t ),(5)whereK 11:=N γCd 2,K 12:=N γCd 2+βγC 2N ∂p (γC,˜C ∗)∂λ,and K 2:=βγC 2N ∂p (γC,˜C ∗)∂˜C .For analytical tractability,we assume thatp (λ,˜C )=max {0,(λ−˜C )}λ.(6)We will now state the main result of this paper which serves as the design for the A VQ algorithm.The proof of this result is given in Section 4.Theorem 2.1Suppose that the feedback delay d,number of users N,and the utilization γ,are given.Let ˆαbe given by:ˆα=min N γCd ,π2dK 2π24d −K 212+K 211 .(7)Then,for all α<ˆα,the system is locally stable.3SimulationsThe above theorem shows that by choosing αaccording to (7),one can guarantee the local stability of the TCP/AQM scheme.However,the fluid-model does not take into account the discrete packet behavior of the network as well as the inherent nonlin-earities in the TCP algorithm.As a result,it becomes impor-tant to verify the analytical results using simulations in which the nonlinearities of the system are taken into account.In thissection,we conduct experiments that simulate various scenarios in the network and show that the A VQ algorithm performs as predicted by the analytical model in all the experiments.Even though each experiment shows that A VQ results in small queues,low loss and high utilization at the link,it is important to note that each experiment simulates a different scenario and the per-formance of A VQ is tested in this scenario.In this section,we use the packet-simulator ns-2[18]to simu-late the adaptive virtual queue scheme.We show that the simu-lation results agree with the convergence results shown in the previous section.In particular,we select an α,using Theo-rem 2.1that will ensure stability for a given round-trip delay d,and a lower bound on the number of users,N.We then present a single set out of many experiments that we did to show that αindeed stabilizes the system even in the presence of arrivals and departure of short connections.We then compare this scheme with many other AQM schemes.3.1Simulation SetupThroughout this section,we consider a single link of capacity 10Mbps that marks or drops packets according to some AQMscheme.For A VQ,unless otherwise stated,we let γ,the desired utilization,be 0.98.We use TCP-Reno as the default transport protocol with the TCP data packet size set to 1000bytes.Each TCP connection is placed in one of five classes which differ only in their round-trip propagation delays.Class 1has a round-tripdelay of 40ms,Class 2has a round-trip delay of 60ms,Class 3has a round-trip delay of 80ms,Class 4has a round-trip delay of 100ms and Class 5has a round-trip delay of 130ms.The buffer size at the link is assumed to be 100packets.In the first five experiments,we assume that the link markspackets and thus,any packet loss is due to buffer overflow.Inthese experiments,we demonstrate that the A VQ scheme achieves high utilization and low packet loss.Further,the algorithm re-sponds quickly to changing network conditions such as vary-ing number of TCP flows.In the first experiment,we studythe convergence properties of the A VQ algorithm both in the absence and in the presence of short flows.In the remaining experiments,we compare the A VQ algorithm with other AQM schemes like RED,REM,PI and the Gibbens’and Kellys’Vir-tual Queue (GKVQ).In the second experiment,we compare the performance of various AQM schemes in the presence oflong-lived flows.In the third experiment,we compare the tran-sient behavior of the AQM schemes when long-lived flows aredropped and added to the network.In the fourth experiment,we compare the AQM schemes in the presence of short flows in the network.In the fifth experiment,we study the sensitivity of the A VQ algorithm to the smoothing parameter (α)as well as to the desired utilization parameter (γ).In the last experiment,we compare the A VQ scheme with other schemes when the linkdrops packets (as opposed to marking)to indicate congestion.Again,the A VQ scheme is shown to have smaller queue lengths compared to other schemes.We design the A VQ controller for a maximum delay of d =ing the design rule in Theorem 2.1,any α<0.8,will ensure stability.In the experiments,to account for non-20406080100120140160180200450500550600650700Time (in seconds)V i r t u a l C a p a c i t y (i n p a c k e t s /s e c o n d )Virtual capacity vs time for AVQFigure 4:Experiment 1.Evolution of the virtual capacity with time for the A VQ schemeTable 1:Experiment 1.Mean and the standard deviation of the queue size before and after the introduction of short flows.Before After Short Flows Short FlowsAverage Queue Size 12.1719.19Standard Deviation 10.2813.44linearities in the system,we let αbe 0.15.In all experiments,we consider two types of flows:FTP flows that persist throughout the duration of the simulations and FTP flows of 20packets each (to model the short flows).Experiment 1:In this experiment,we study the convergence properties and buffer sizes at the queue for the A VQ scheme alone.A total of 180FTP flows with 36in each delay class persist throughout the duration of the simulations,while short flows (of 20pack-ets each)arrive at the link at the rate of 30flows per second.The short flows are uniformly distributed among the five delay classes.To simulate a sudden change in network conditions,we start the experiment with only FTP flows in the system and intro-duce the short flows after 100seconds.The evolution of the vir-tual capacity is given in Figure 4.After an initial transient,the virtual capacity settles down and oscillates around a particular value.Note that the oscillations in the virtual capacity are due to the packet nature of the network which is not captured by the analytical model.At 100seconds,there is a drop in the virtual capacity since the A VQ algorithm adapts to the changing num-ber of flows.Beyond 100seconds,the virtual capacity is lower than it was before 100seconds since the links marks packets ag-gressively due to the increased load.The queue length evolution for the system every 100ms is given in Figure 5.Except dur-ing transients introduced by load changes,the queue lengths are small (less than 20packets).At 100seconds,the queue length jumps up due to the short flows.However,the system stabilizes and the queue lengths are small once again.Table 1gives the average and the standard deviation of the queue length before and after the introduction of short flows.Note that there isFigure 5:Experiment 1.Queue length vs time for the A VQschemea small increase in the average queue length as well as in the standard deviation due to the addition of short-flows in the sys-tem.Another important performance measure is the number of packets dropped due to buffer overflow in the system.Since ECN marking is used,we expect the number of packets lost due to buffer overflow to be small.Indeed only 10out of roughly 250,000packets are dropped.These drops are primarily due to the sudden additional load brought on by the short flows.An-other performance measure that is of interest is the utilization of the link.The utilization was observed to be 0.9827,which is very close to the desired utilization of 0.98.We note that the apparent discrepancy between Figure 5,where the queue length never reaches the buffer size of 100packets,and the fact that there are 10dropped packets is due to the fact that the queue length is sampled only once every 100ms to plot the graphs.We will now compare the A VQ scheme with other AQM schemes that have been proposed.Since there are many AQM schemes in the literature,we will compare the A VQ scheme with a rep-resentative few.In particular,we will compare the A VQ scheme with1.Random Early Discard (RED)proposed in [1].In our ex-periments,we use the “gentle”version of RED.Unless oth-erwise stated,the parameters were chosen as recommended in /floyd/REDparameters.txt .2.Random Early Marking (REM)proposed in [10].The REM scheme tries to regulate the queue length to a desired value (denoted by qref )by adapting the marking probability.The REM controller marks each packet with a probabil-ity p which is updated periodically (say,every T seconds)asp [k +1]=1−φ−µ[k +1],where φis a arbitrary constant greater than one and µ[k +1]=max(0,µ[k ]+γ(q [k +1]−(1−α)q [k ]−αqref )),and αand γare constants and q [k +1]is the queue length at the k +1sampling instant.Since REM is very sensitive to φ,we will use the values as recommended in [10].3.The PI controller proposed in[5].The PI controller markseach packet with a probability p which is updated periodi-cally(say,every T seconds)asp[k+1]=p[k]+a(q[k+1]−qref)−b(q[k]−qref),where a>0and b>0are constants chosen according tothe design rules given in[5].4.The virtual queue based AQM scheme(GKVQ)proposedin[11].In this scheme,the link maintains a virtual queuewithfixed capacity˜C=θC,and buffer size˜B=θB,whereθ<1,and B is the buffer capacity of the originalqueue.Whenever the virtual queue overflows,all pack-ets in the real queue and all future incoming packets aremarked till the virtual queue becomes empty again.Notethat this scheme cannot be used in the case where the linkdrops the packets instead of marking them because the through-put would be very bad due to aggressive dropping.Asin[11],we will useθ=0.90in all our simulations usingthe GKVQ.Experiment2:In this experiment,we compare the performance of the various AQM schemes assuming that the link“marks”packets and in the presence of long-lived FTPflows only.The queue size at the link is set to100packets.The desired queue length for the REM scheme and the PI scheme is set at50packets and the minthresh and the maxthresh for the RED(with gentle turned on)scheme are set at37and75packets respectively.Recall that the desired utilization of the link is set to be0.98for the A VQ scheme. Since we use an average queue length of50packets for REM and the PI controller,it is natural to attempt to regulate the queue length to50for the A VQ scheme also.However,the A VQ does not directly attempt to control queue size.Thus,for the A VQ scheme,we drop every packet that arrives when there are al-ready50packets in the real queue.Note that this is the worst-case scenario for the A VQ scheme,since when ECN marking is used,the natural primary measure of performance is packet loss. We summarize our simulation results below:•Packet Losses and Link Utilization:The losses incurred byall the schemes are shown in Figure6as a function ofthe number of FTPflows.The A VQ scheme has fewerlosses than any other scheme except the GKVQ even athigh loads.The loss rate for GKVQ and A VQ are com-parable;however,the GKVQ marks packets more aggres-sively than any other scheme and thus has lower utilization.Figure7shows the utilization of the link for all the AQMschemes.Note that,the utilization of GKVQ is as low as75%.This can once again be attributed to the aggressivemarking strategy of GKVQ.RED also results in a poor uti-lization of the link.Our observation has been that when in-creasing the utilization of RED(by tuning its parameters),the packet losses at the link also increases.REM and PIhave a utilization of1.0as the queue is always non-empty.For the A VQ scheme,we required a desired utilization of0.98and we can see that the A VQ scheme tracks the desiredutilization quite well.Thus,the main conclusion from thisexperiment is that the A VQ achieves low loss with high uti-lization.Figure6:Experiment2.Losses at the link for varying number of FTP connections for the different AQM schemes•Responsiveness to changing network conditions:The objec-tive of this experiment is to measure the response of eachAQM scheme when the number offlows is increased.Twenty new FTP users are added every100s till the total num-ber of FTP connections reach180and the average queuelength over every100s is computed.Schemes that have along transient period will have an increasing average queuelength as new users are added before the scheme is able toconverge.The average queue length(over each100secondinterval)of each scheme as the number of users increase isshown in Figure8.We see from thefigure that PI and REMhave higher average queue lengths than the desired queuelength.On the other hand,A VQ,GKVQ and RED havesmaller queue sizes.This is due to the fact that REM andPI apparently have a long transient period and new usersare added before the queue length converges.The aver-age queue length over each100s interval is used to capturepersistent transients in this experiment for studying the re-sponsiveness of the AQM schemes to load changes.Thisexperiment shows that A VQ is responsive to changes innetwork load and is able to maintain a small queue lengtheven when the network load keeps increasing. Experiment3:In this experiment,we compare the responsiveness of the AQM schemes whenflows are dropped and then introduced later on. Specifically,we only compare REM and the PI controller(since these are only ones among those that we have discussed that attempt to precisely regulate the queue length to a desired value) with the A VQ controller.Unless otherwise stated,all the system parameters are identical to Experiment2.The number of FTP connections is140at time0.0At time100,105FTP connections are dropped and at time150a new set of105FTP connections is established.We plot the evolution of the queue size for each of the AQM scheme.Figure9shows the evolution of the queue size for PI as theflows depart and arrive.Note that the desired。
MetacognitiveFeedback：元认知反馈
Metacognitive FeedbackWhy Is This Strategy Useful?Some students are weighed down by the cognitive load associated with learning and organizing new knowledge, particularly in mathematics. Metacognition has been identified as a critical process that supports student learning and problem solving. Computers can serve as an effective aid in learning process and in developing mathematical reasoning because they provide students with immediate feedback about their progress and success in the task. The use of animated “agents” in computer software extends the cognitive scaffolding provided by various computer tools and representations (e.g., searchable text, simulations, concept maps, etc.) by embedding them in productive and motivating social-constructive interactions (e.g., peer teaching, collaboration, and assessment). In those types of interactions, the student teaches the animated agent and sharpens his or her conceptual understanding from the interactive experience.Description of StrategyTeachable Agents are software environments in which students teach a computer “agent” using well-structured visual representations. Sometimes, these “agents” will have a name (in the study below, one was “Betty’s Brain”). An important feature of these types of computer environments is that students can monitor how the computer agent answers questions, and can provide corrections if the computer makes mistakes. Students are encouraged to “learn-by-teaching” through the use of three primary strategies:(1) Teach the computer agent by using a concept map(2) Query the computer agent with your own questions to see if the computer understood(3) Quiz the computer with tests to see how well the computer agent doesAlthough this strategy is grounded in a specific computer environment, the principles of metacognitive feedback and “learn-by-teaching” can be extended to other contexts. However, metacognitive feedback must be accompanied by content cues in order for students to benefit. Students have to be taught and given enough opportunities to practice metacognitive strategies in socially engaging and relevant ways.Research EvidenceOne correlational study provides support for this approach. This study included 39 students in two 5th grade classrooms. The study was designed to compare the effects of the different types of feedback. Results showed that guided Metacognitive feedback better prepares the student for future learning tasks even in situations where the metacognitive support is removed. However, guided feedback with Metacognitive cues but no content information does not help novice learners with low prior knowledge. Students have to be taught and given enough opportunities to practice Metacognitive strategies in socially engaging and relevant ways.Sample Studies Supporting this StrategyTan, J., Biswas, G., & Schwartz, D.L. (n.d.). Feedback for Metacognitive Support in Learning by Teaching Environments. Available at:/papers/p828-cogsci.pdf.Past research on feedback in computer-based learning environments has shown that corrective feedback helps immediate learning, whereas guided and metacognitive feedback help in gaining deep understanding and developing the ability to transfer knowledge. Feedback becomes important in discovery learning environments, where novice students are often overwhelmed by the cognitive load associated with learning and organizing new knowledge while monitoring their own learning progress. We focus on feedback mechanisms in teachable agent systems to help improve students’ abilities to monitor their agent’s knowledge, and, in the process, their own learning and understanding. Our studies demonstrate the effectiveness of guided metacognitive feedback in preparing students for future learning.Additional ResourcesThe Teachable Agents Groups at Vanderbilt University. /The AAA Lab @ Stanford. /teachable.html#。
ACM-GIS%202006-A%20Peer-to-Peer%20Spatial%20Cloaking%20Algorithm%20for%20Anonymous%20Location-based%
A Peer-to-Peer Spatial Cloaking Algorithm for AnonymousLocation-based Services∗Chi-Yin Chow Department of Computer Science and Engineering University of Minnesota Minneapolis,MN cchow@ Mohamed F.MokbelDepartment of ComputerScience and EngineeringUniversity of MinnesotaMinneapolis,MNmokbel@Xuan LiuIBM Thomas J.WatsonResearch CenterHawthorne,NYxuanliu@ABSTRACTThis paper tackles a major privacy threat in current location-based services where users have to report their ex-act locations to the database server in order to obtain their desired services.For example,a mobile user asking about her nearest restaurant has to report her exact location.With untrusted service providers,reporting private location in-formation may lead to several privacy threats.In this pa-per,we present a peer-to-peer(P2P)spatial cloaking algo-rithm in which mobile and stationary users can entertain location-based services without revealing their exact loca-tion information.The main idea is that before requesting any location-based service,the mobile user will form a group from her peers via single-hop communication and/or multi-hop routing.Then,the spatial cloaked area is computed as the region that covers the entire group of peers.Two modes of operations are supported within the proposed P2P spa-tial cloaking algorithm,namely,the on-demand mode and the proactive mode.Experimental results show that the P2P spatial cloaking algorithm operated in the on-demand mode has lower communication cost and better quality of services than the proactive mode,but the on-demand incurs longer response time.Categories and Subject Descriptors:H.2.8[Database Applications]:Spatial databases and GISGeneral Terms:Algorithms and Experimentation. Keywords:Mobile computing,location-based services,lo-cation privacy and spatial cloaking.1.INTRODUCTIONThe emergence of state-of-the-art location-detection de-vices,e.g.,cellular phones,global positioning system(GPS) devices,and radio-frequency identification(RFID)chips re-sults in a location-dependent information access paradigm,∗This work is supported in part by the Grants-in-Aid of Re-search,Artistry,and Scholarship,University of Minnesota. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on thefirst page.To copy otherwise,to republish,to post on servers or to redistribute to lists,requires prior specific permission and/or a fee.ACM-GIS’06,November10-11,2006,Arlington,Virginia,USA. Copyright2006ACM1-59593-529-0/06/0011...$5.00.known as location-based services(LBS)[30].In LBS,mobile users have the ability to issue location-based queries to the location-based database server.Examples of such queries include“where is my nearest gas station”,“what are the restaurants within one mile of my location”,and“what is the traffic condition within ten minutes of my route”.To get the precise answer of these queries,the user has to pro-vide her exact location information to the database server. With untrustworthy servers,adversaries may access sensi-tive information about specific individuals based on their location information and issued queries.For example,an adversary may check a user’s habit and interest by knowing the places she visits and the time of each visit,or someone can track the locations of his ex-friends.In fact,in many cases,GPS devices have been used in stalking personal lo-cations[12,39].To tackle this major privacy concern,three centralized privacy-preserving frameworks are proposed for LBS[13,14,31],in which a trusted third party is used as a middleware to blur user locations into spatial regions to achieve k-anonymity,i.e.,a user is indistinguishable among other k−1users.The centralized privacy-preserving frame-work possesses the following shortcomings:1)The central-ized trusted third party could be the system bottleneck or single point of failure.2)Since the centralized third party has the complete knowledge of the location information and queries of all users,it may pose a serious privacy threat when the third party is attacked by adversaries.In this paper,we propose a peer-to-peer(P2P)spatial cloaking algorithm.Mobile users adopting the P2P spatial cloaking algorithm can protect their privacy without seeking help from any centralized third party.Other than the short-comings of the centralized approach,our work is also moti-vated by the following facts:1)The computation power and storage capacity of most mobile devices have been improv-ing at a fast pace.2)P2P communication technologies,such as IEEE802.11and Bluetooth,have been widely deployed.3)Many new applications based on P2P information shar-ing have rapidly taken shape,e.g.,cooperative information access[9,32]and P2P spatio-temporal query processing[20, 24].Figure1gives an illustrative example of P2P spatial cloak-ing.The mobile user A wants tofind her nearest gas station while beingfive anonymous,i.e.,the user is indistinguish-able amongfive users.Thus,the mobile user A has to look around andfind other four peers to collaborate as a group. In this example,the four peers are B,C,D,and E.Then, the mobile user A cloaks her exact location into a spatialA B CDEBase Stationregion that covers the entire group of mobile users A ,B ,C ,D ,and E .The mobile user A randomly selects one of the mobile users within the group as an agent .In the ex-ample given in Figure 1,the mobile user D is selected as an agent.Then,the mobile user A sends her query (i.e.,what is the nearest gas station)along with her cloaked spa-tial region to the agent.The agent forwards the query to the location-based database server through a base station.Since the location-based database server processes the query based on the cloaked spatial region,it can only give a list of candidate answers that includes the actual answers and some false positives.After the agent receives the candidate answers,it forwards the candidate answers to the mobile user A .Finally,the mobile user A gets the actual answer by filtering out all the false positives.The proposed P2P spatial cloaking algorithm can operate in two modes:on-demand and proactive .In the on-demand mode,mobile clients execute the cloaking algorithm when they need to access information from the location-based database server.On the other side,in the proactive mode,mobile clients periodically look around to find the desired number of peers.Thus,they can cloak their exact locations into spatial regions whenever they want to retrieve informa-tion from the location-based database server.In general,the contributions of this paper can be summarized as follows:1.We introduce a distributed system architecture for pro-viding anonymous location-based services (LBS)for mobile users.2.We propose the first P2P spatial cloaking algorithm for mobile users to entertain high quality location-based services without compromising their privacy.3.We provide experimental evidence that our proposed algorithm is efficient in terms of the response time,is scalable to large numbers of mobile clients,and is effective as it provides high-quality services for mobile clients without the need of exact location information.The rest of this paper is organized as follows.Section 2highlights the related work.The system model of the P2P spatial cloaking algorithm is presented in Section 3.The P2P spatial cloaking algorithm is described in Section 4.Section 5discusses the integration of the P2P spatial cloak-ing algorithm with privacy-aware location-based database servers.Section 6depicts the experimental evaluation of the P2P spatial cloaking algorithm.Finally,Section 7con-cludes this paper.2.RELATED WORKThe k -anonymity model [37,38]has been widely used in maintaining privacy in databases [5,26,27,28].The main idea is to have each tuple in the table as k -anonymous,i.e.,indistinguishable among other k −1tuples.Although we aim for the similar k -anonymity model for the P2P spatial cloaking algorithm,none of these techniques can be applied to protect user privacy for LBS,mainly for the following four reasons:1)These techniques preserve the privacy of the stored data.In our model,we aim not to store the data at all.Instead,we store perturbed versions of the data.Thus,data privacy is managed before storing the data.2)These approaches protect the data not the queries.In anonymous LBS,we aim to protect the user who issues the query to the location-based database server.For example,a mobile user who wants to ask about her nearest gas station needs to pro-tect her location while the location information of the gas station is not protected.3)These approaches guarantee the k -anonymity for a snapshot of the database.In LBS,the user location is continuously changing.Such dynamic be-havior calls for continuous maintenance of the k -anonymity model.(4)These approaches assume a unified k -anonymity requirement for all the stored records.In our P2P spatial cloaking algorithm,k -anonymity is a user-specified privacy requirement which may have a different value for each user.Motivated by the privacy threats of location-detection de-vices [1,4,6,40],several research efforts are dedicated to protect the locations of mobile users (e.g.,false dummies [23],landmark objects [18],and location perturbation [10,13,14]).The most closed approaches to ours are two centralized spatial cloaking algorithms,namely,the spatio-temporal cloaking [14]and the CliqueCloak algorithm [13],and one decentralized privacy-preserving algorithm [23].The spatio-temporal cloaking algorithm [14]assumes that all users have the same k -anonymity requirements.Furthermore,it lacks the scalability because it deals with each single request of each user individually.The CliqueCloak algorithm [13]as-sumes a different k -anonymity requirement for each user.However,since it has large computation overhead,it is lim-ited to a small k -anonymity requirement,i.e.,k is from 5to 10.A decentralized privacy-preserving algorithm is proposed for LBS [23].The main idea is that the mobile client sends a set of false locations,called dummies ,along with its true location to the location-based database server.However,the disadvantages of using dummies are threefold.First,the user has to generate realistic dummies to pre-vent the adversary from guessing its true location.Second,the location-based database server wastes a lot of resources to process the dummies.Finally,the adversary may esti-mate the user location by using cellular positioning tech-niques [34],e.g.,the time-of-arrival (TOA),the time differ-ence of arrival (TDOA)and the direction of arrival (DOA).Although several existing distributed group formation al-gorithms can be used to find peers in a mobile environment,they are not designed for privacy preserving in LBS.Some algorithms are limited to only finding the neighboring peers,e.g.,lowest-ID [11],largest-connectivity (degree)[33]and mobility-based clustering algorithms [2,25].When a mo-bile user with a strict privacy requirement,i.e.,the value of k −1is larger than the number of neighboring peers,it has to enlist other peers for help via multi-hop routing.Other algorithms do not have this limitation,but they are designed for grouping stable mobile clients together to facil-Location-based Database ServerDatabase ServerDatabase ServerFigure 2:The system architectureitate efficient data replica allocation,e.g.,dynamic connec-tivity based group algorithm [16]and mobility-based clus-tering algorithm,called DRAM [19].Our work is different from these approaches in that we propose a P2P spatial cloaking algorithm that is dedicated for mobile users to dis-cover other k −1peers via single-hop communication and/or via multi-hop routing,in order to preserve user privacy in LBS.3.SYSTEM MODELFigure 2depicts the system architecture for the pro-posed P2P spatial cloaking algorithm which contains two main components:mobile clients and location-based data-base server .Each mobile client has its own privacy profile that specifies its desired level of privacy.A privacy profile includes two parameters,k and A min ,k indicates that the user wants to be k -anonymous,i.e.,indistinguishable among k users,while A min specifies the minimum resolution of the cloaked spatial region.The larger the value of k and A min ,the more strict privacy requirements a user needs.Mobile users have the ability to change their privacy profile at any time.Our employed privacy profile matches the privacy re-quirements of mobiles users as depicted by several social science studies (e.g.,see [4,15,17,22,29]).In this architecture,each mobile user is equipped with two wireless network interface cards;one of them is dedicated to communicate with the location-based database server through the base station,while the other one is devoted to the communication with other peers.A similar multi-interface technique has been used to implement IP multi-homing for stream control transmission protocol (SCTP),in which a machine is installed with multiple network in-terface cards,and each assigned a different IP address [36].Similarly,in mobile P2P cooperation environment,mobile users have a network connection to access information from the server,e.g.,through a wireless modem or a base station,and the mobile users also have the ability to communicate with other peers via a wireless LAN,e.g.,IEEE 802.11or Bluetooth [9,24,32].Furthermore,each mobile client is equipped with a positioning device, e.g.,GPS or sensor-based local positioning systems,to determine its current lo-cation information.4.P2P SPATIAL CLOAKINGIn this section,we present the data structure and the P2P spatial cloaking algorithm.Then,we describe two operation modes of the algorithm:on-demand and proactive .4.1Data StructureThe entire system area is divided into grid.The mobile client communicates with each other to discover other k −1peers,in order to achieve the k -anonymity requirement.TheAlgorithm 1P2P Spatial Cloaking:Request Originator m 1:Function P2PCloaking-Originator (h ,k )2://Phase 1:Peer searching phase 3:The hop distance h is set to h4:The set of discovered peers T is set to {∅},and the number ofdiscovered peers k =|T |=05:while k <k −1do6:Broadcast a FORM GROUP request with the parameter h (Al-gorithm 2gives the response of each peer p that receives this request)7:T is the set of peers that respond back to m by executingAlgorithm 28:k =|T |;9:if k <k −1then 10:if T =T then 11:Suspend the request 12:end if 13:h ←h +1;14:T ←T ;15:end if 16:end while17://Phase 2:Location adjustment phase 18:for all T i ∈T do19:|mT i .p |←the greatest possible distance between m and T i .pby considering the timestamp of T i .p ’s reply and maximum speed20:end for21://Phase 3:Spatial cloaking phase22:Form a group with k −1peers having the smallest |mp |23:h ←the largest hop distance h p of the selected k −1peers 24:Determine a grid area A that covers the entire group 25:if A <A min then26:Extend the area of A till it covers A min 27:end if28:Randomly select a mobile client of the group as an agent 29:Forward the query and A to the agentmobile client can thus blur its exact location into a cloaked spatial region that is the minimum grid area covering the k −1peers and itself,and satisfies A min as well.The grid area is represented by the ID of the left-bottom and right-top cells,i.e.,(l,b )and (r,t ).In addition,each mobile client maintains a parameter h that is the required hop distance of the last peer searching.The initial value of h is equal to one.4.2AlgorithmFigure 3gives a running example for the P2P spatial cloaking algorithm.There are 15mobile clients,m 1to m 15,represented as solid circles.m 8is the request originator,other black circles represent the mobile clients received the request from m 8.The dotted circles represent the commu-nication range of the mobile client,and the arrow represents the movement direction.Algorithms 1and 2give the pseudo code for the request originator (denoted as m )and the re-quest receivers (denoted as p ),respectively.In general,the algorithm consists of the following three phases:Phase 1:Peer searching phase .The request origina-tor m wants to retrieve information from the location-based database server.m first sets h to h ,a set of discovered peers T to {∅}and the number of discovered peers k to zero,i.e.,|T |.(Lines 3to 4in Algorithm 1).Then,m broadcasts a FORM GROUP request along with a message sequence ID and the hop distance h to its neighboring peers (Line 6in Algorithm 1).m listens to the network and waits for the reply from its neighboring peers.Algorithm 2describes how a peer p responds to the FORM GROUP request along with a hop distance h and aFigure3:P2P spatial cloaking algorithm.Algorithm2P2P Spatial Cloaking:Request Receiver p1:Function P2PCloaking-Receiver(h)2://Let r be the request forwarder3:if the request is duplicate then4:Reply r with an ACK message5:return;6:end if7:h p←1;8:if h=1then9:Send the tuple T=<p,(x p,y p),v maxp ,t p,h p>to r10:else11:h←h−1;12:Broadcast a FORM GROUP request with the parameter h 13:T p is the set of peers that respond back to p14:for all T i∈T p do15:T i.h p←T i.h p+1;16:end for17:T p←T p∪{<p,(x p,y p),v maxp ,t p,h p>};18:Send T p back to r19:end ifmessage sequence ID from another peer(denoted as r)that is either the request originator or the forwarder of the re-quest.First,p checks if it is a duplicate request based on the message sequence ID.If it is a duplicate request,it sim-ply replies r with an ACK message without processing the request.Otherwise,p processes the request based on the value of h:Case1:h= 1.p turns in a tuple that contains its ID,current location,maximum movement speed,a timestamp and a hop distance(it is set to one),i.e.,< p,(x p,y p),v max p,t p,h p>,to r(Line9in Algorithm2). Case2:h> 1.p decrements h and broadcasts the FORM GROUP request with the updated h and the origi-nal message sequence ID to its neighboring peers.p keeps listening to the network,until it collects the replies from all its neighboring peers.After that,p increments the h p of each collected tuple,and then it appends its own tuple to the collected tuples T p.Finally,it sends T p back to r (Lines11to18in Algorithm2).After m collects the tuples T from its neighboring peers, if m cannotfind other k−1peers with a hop distance of h,it increments h and re-broadcasts the FORM GROUP request along with a new message sequence ID and h.m repeatedly increments h till itfinds other k−1peers(Lines6to14in Algorithm1).However,if mfinds the same set of peers in two consecutive broadcasts,i.e.,with hop distances h and h+1,there are not enough connected peers for m.Thus, m has to relax its privacy profile,i.e.,use a smaller value of k,or to be suspended for a period of time(Line11in Algorithm1).Figures3(a)and3(b)depict single-hop and multi-hop peer searching in our running example,respectively.In Fig-ure3(a),the request originator,m8,(e.g.,k=5)canfind k−1peers via single-hop communication,so m8sets h=1. Since h=1,its neighboring peers,m5,m6,m7,m9,m10, and m11,will not further broadcast the FORM GROUP re-quest.On the other hand,in Figure3(b),m8does not connect to k−1peers directly,so it has to set h>1.Thus, its neighboring peers,m7,m10,and m11,will broadcast the FORM GROUP request along with a decremented hop dis-tance,i.e.,h=h−1,and the original message sequence ID to their neighboring peers.Phase2:Location adjustment phase.Since the peer keeps moving,we have to capture the movement between the time when the peer sends its tuple and the current time. For each received tuple from a peer p,the request originator, m,determines the greatest possible distance between them by an equation,|mp |=|mp|+(t c−t p)×v max p,where |mp|is the Euclidean distance between m and p at time t p,i.e.,|mp|=(x m−x p)2+(y m−y p)2,t c is the currenttime,t p is the timestamp of the tuple and v maxpis the maximum speed of p(Lines18to20in Algorithm1).In this paper,a conservative approach is used to determine the distance,because we assume that the peer will move with the maximum speed in any direction.If p gives its movement direction,m has the ability to determine a more precise distance between them.Figure3(c)illustrates that,for each discovered peer,the circle represents the largest region where the peer can lo-cate at time t c.The greatest possible distance between the request originator m8and its discovered peer,m5,m6,m7, m9,m10,or m11is represented by a dotted line.For exam-ple,the distance of the line m8m 11is the greatest possible distance between m8and m11at time t c,i.e.,|m8m 11|. Phase3:Spatial cloaking phase.In this phase,the request originator,m,forms a virtual group with the k−1 nearest peers,based on the greatest possible distance be-tween them(Line22in Algorithm1).To adapt to the dynamic network topology and k-anonymity requirement, m sets h to the largest value of h p of the selected k−1 peers(Line15in Algorithm1).Then,m determines the minimum grid area A covering the entire group(Line24in Algorithm1).If the area of A is less than A min,m extends A,until it satisfies A min(Lines25to27in Algorithm1). Figure3(c)gives the k−1nearest peers,m6,m7,m10,and m11to the request originator,m8.For example,the privacy profile of m8is(k=5,A min=20cells),and the required cloaked spatial region of m8is represented by a bold rectan-gle,as depicted in Figure3(d).To issue the query to the location-based database server anonymously,m randomly selects a mobile client in the group as an agent(Line28in Algorithm1).Then,m sendsthe query along with the cloaked spatial region,i.e.,A,to the agent(Line29in Algorithm1).The agent forwards thequery to the location-based database server.After the serverprocesses the query with respect to the cloaked spatial re-gion,it sends a list of candidate answers back to the agent.The agent forwards the candidate answer to m,and then mfilters out the false positives from the candidate answers. 4.3Modes of OperationsThe P2P spatial cloaking algorithm can operate in twomodes,on-demand and proactive.The on-demand mode:The mobile client only executesthe algorithm when it needs to retrieve information from the location-based database server.The algorithm operatedin the on-demand mode generally incurs less communica-tion overhead than the proactive mode,because the mobileclient only executes the algorithm when necessary.However,it suffers from a longer response time than the algorithm op-erated in the proactive mode.The proactive mode:The mobile client adopting theproactive mode periodically executes the algorithm in back-ground.The mobile client can cloak its location into a spa-tial region immediately,once it wants to communicate withthe location-based database server.The proactive mode pro-vides a better response time than the on-demand mode,but it generally incurs higher communication overhead and giveslower quality of service than the on-demand mode.5.ANONYMOUS LOCATION-BASEDSERVICESHaving the spatial cloaked region as an output form Algo-rithm1,the mobile user m sends her request to the location-based server through an agent p that is randomly selected.Existing location-based database servers can support onlyexact point locations rather than cloaked regions.In or-der to be able to work with a spatial region,location-basedservers need to be equipped with a privacy-aware queryprocessor(e.g.,see[29,31]).The main idea of the privacy-aware query processor is to return a list of candidate answerrather than the exact query answer.Then,the mobile user m willfilter the candidate list to eliminate its false positives andfind its exact answer.The tighter the spatial cloaked re-gion,the lower is the size of the candidate answer,and hencethe better is the performance of the privacy-aware query processor.However,tight cloaked regions may represent re-laxed privacy constrained.Thus,a trade-offbetween the user privacy and the quality of service can be achieved[31]. Figure4(a)depicts such scenario by showing the data stored at the server side.There are32target objects,i.e., gas stations,T1to T32represented as black circles,the shaded area represents the spatial cloaked area of the mo-bile client who issued the query.For clarification,the actual mobile client location is plotted in Figure4(a)as a black square inside the cloaked area.However,such information is neither stored at the server side nor revealed to the server. The privacy-aware query processor determines a range that includes all target objects that are possibly contributing to the answer given that the actual location of the mobile client could be anywhere within the shaded area.The range is rep-resented as a bold rectangle,as depicted in Figure4(b).The server sends a list of candidate answers,i.e.,T8,T12,T13, T16,T17,T21,and T22,back to the agent.The agent next for-(a)Server Side(b)Client SideFigure4:Anonymous location-based services wards the candidate answers to the requesting mobile client either through single-hop communication or through multi-hop routing.Finally,the mobile client can get the actualanswer,i.e.,T13,byfiltering out the false positives from thecandidate answers.The algorithmic details of the privacy-aware query proces-sor is beyond the scope of this paper.Interested readers are referred to[31]for more details.6.EXPERIMENTAL RESULTSIn this section,we evaluate and compare the scalabilityand efficiency of the P2P spatial cloaking algorithm in boththe on-demand and proactive modes with respect to the av-erage response time per query,the average number of mes-sages per query,and the size of the returned candidate an-swers from the location-based database server.The queryresponse time in the on-demand mode is defined as the timeelapsed between a mobile client starting to search k−1peersand receiving the candidate answers from the agent.On theother hand,the query response time in the proactive mode is defined as the time elapsed between a mobile client startingto forward its query along with the cloaked spatial regionto the agent and receiving the candidate answers from theagent.The simulation model is implemented in C++usingCSIM[35].In all the experiments in this section,we consider an in-dividual random walk model that is based on“random way-point”model[7,8].At the beginning,the mobile clientsare randomly distributed in a spatial space of1,000×1,000square meters,in which a uniform grid structure of100×100cells is constructed.Each mobile client randomly chooses itsown destination in the space with a randomly determined speed s from a uniform distribution U(v min,v max).When the mobile client reaches the destination,it comes to a stand-still for one second to determine its next destination.Afterthat,the mobile client moves towards its new destinationwith another speed.All the mobile clients repeat this move-ment behavior during the simulation.The time interval be-tween two consecutive queries generated by a mobile client follows an exponential distribution with a mean of ten sec-onds.All the experiments consider one half-duplex wirelesschannel for a mobile client to communicate with its peers with a total bandwidth of2Mbps and a transmission range of250meters.When a mobile client wants to communicate with other peers or the location-based database server,it has to wait if the requested channel is busy.In the simulated mobile environment,there is a centralized location-based database server,and one wireless communication channel between the location-based database server and the mobile。
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
is the technique most compatible with typical data requirements: no data should be lost, and the network can take whatever time needed for data delivery. A number of consolidation algorithms have been proposed in 1, 7, 8, 9 . Several design and implementation considerations come into play when developing a consolidation algorithm. The oscillations and transient response of the algorithm are important. The algorithm must also be scalable to very large multicast trees. The implementation complexity, feedback delay, and the overhead of the backward RM cells should not increase with the increase of the number of levels or leaves of the multicast tree. In this paper, we propose a set of consolidation algorithms that aim at providing a fast transient response, while eliminating consolidation noise. We examine the performance of the proposed algorithms, and compare it to the previous ones in complexity, transient response, consolidation noise, and scalability. The remainder of the paper is organized as follows. The next two sections provide an overview of the ABR ow control mechanism, and a summary of the previous work on point-to-multipoint ABR ow control. A discussion of the various design and implementation issues involved is then presented, followed by a description of the speci c underlying switch scheme employed. An explanation and pseudocode of the previously proposed consolidation algorithms, as well as the new ones we propose, is presented next. All the algorithms are then simulated and analyzed under a variety of con gurations. The paper concludes with a discussion of the tradeo s among the algorithms.
2 ABR Flow Control
The available bit rate ABR service for data tra c in ATM networks periodically indicates to sources the rate at which they should be transmitting. The switches monitor their load and compute the available bandwidth, dividing it fairly among active ows. The feedback from the switches to the sources is sent in resource management RM cells which are generated by the sources and turned around by the destinations. The RM cells contain the source current cell rate CCR, in addition to elds that can be used by the switches to provide feedback to the sources. These elds are: explicit rate ER, the congestion indication CI ag, and no increase NI ag. The ER eld indicates the rate that the network can support at this particular instant. Initially, the ER eld is set to a value no greater than the peak cell rate PCR, and the CI and NI ags are clear. Each switch on the path reduces the ER eld to the maximum rate it can support, and sets CI or NI if necessary 5 . A component c is said to be downstream of another component c in a certain connection if c is on the path from c to the destination. In this case, c is said to be upstream of c . RM cells owing from the source to the destination are called forward RM cells FRMs while those returning from the destination to the source are called backward RM cells BRMs. When a source receives a BRM, it computes its allowed cell rate ACR using its current ACR, the CI and NI ags, and the ER eld of the RM cell.
Abstract: ABR tra c management for point-to-multipoint
connections controls the source rate to the minimum rate supported by all the branches of the multicast tree. A number of algorithms have been developed for extending ABR congestion avoidance algorithms to perform feedback consolidation at the branch points. This paper discusses various design options and implementation alternatives for the consolidation algorithms, and proposes a number of new algorithms. The performance of the proposed algorithms and the previous algorithms is compared under a variety of conditions. Results indicate that the algorithms we propose eliminate the consolidation noise caused if the feedback is returned before all branches respond, while exhibiting a fast transient response. Keywords: ATM networks, ABR service category, tra c management, congestion control, multipoint communication
Feedback Consolidation Algorithms for ABR Point-to-Networks
Sonia Fahmy, Raj Jain, Rohit Goyal, Bobby Vandalore, and Shivkumar Kalyanaraman The Ohio State University E-mail: ffahmy,jaing@ Sastri Kota Lockheed Martin Telecommunications Pradeep Samudra Samsung Telecom America, Inc.