大规模互联网端到端行为评价指标研究

目前,Internet端到端行为的研究仅仅局限在包延迟、包丢失等测量指标的研究上,但如何根据测量指标对链路的性能进行评价却没有做深入研究。文章针对链路特性提出了两个评价指标:稳定系数和总体评价系数,对链路的稳定性和总体性能进行了评价。其中,稳定系数是评价网络链路稳定性的一个系数,总体评价系数是体现链路稳定性和链路速率的综合特性的一个系数,它反映了目前链路给用户的总体使用效果。接着文章介绍了在实际测量中用到的笔者自己编写的测量工具Posip。该工具采用指数间隔进行测量采样,并解决了一般ping工具中的“应答包丢失”问题。最后,文章使用工具posip对我国Internet教育网和科技网中的43条链路进行了测量和分析,并使用两种评价指标对所测链路进行了评价,取得了较好的效果。     

基于信息熵的IP网端到端行为分析与建模

具有开放、分布式、不协作、异构、无中心控制等特点的Internet复杂巨系统的管理、容量规划、新一代网络体系结构设计与分析和性能预测都离不开对网络行为的充分理解。而端到端行为作为网络行为的一个重要组成部分,具有一定的研究价值。该文利用信息熵原理建立用于分析端到端整体宏观行为的信息熵模型,该模型能很好地反映端到端整体宏观行为与链路上各节点的状态概率之间关系,根据该模型可以分析端到端链路上各节点之间的相互作用关系以及它们是如何引起端到端整体宏观行为的。最后,给出了该模型的有效性和稳定性定量分析的判别式。     

链路带宽测量方法改进

链路带宽是网络性能分析，容量优化规划的基本指标，链路带宽测量常用的方法是VPS（variable packet size),但VPS具有误差累计和背景流量影响的缺陷，对VPS方法进行改进，提出和实现一个任意链路带宽测量方法PTVS（packet train with variable size)，消除逐跳测量造成的误差累计和背景流量影响，测量实验表明，PTVS具有精确，高效，迅速的特点，PTVS可测量的其他性能指标还包括RTT，单向延迟，丢包率，端到端瓶颈带宽以及链路利用率。     

基于仿真分析的端到端最小包时延测量方法

端到端最小包时延作为反映端到端路径拓扑特征的基本指标得到广泛应用，但目前缺少对最小时延测量方法的研究。以仿真为手段定量分析了在不同路径长度下最小时延的可测性，并建立了反映探测包数量与路径长度关系的线性方程。以此为基础，提出一种基于仿真分析的最小时延测量方法。在互联网的实际测量表明该方法能以较小的测量开销获得较准确的最小时延测量结果。     

路径差异敏感的包散射策略机制

数据中心网络中随机包散射设计策略有效地提升了链路的多路径利用率,但是容易造成乱序问题。介绍了由数据包乱序所引起的“伪丢失”现象,设计了基于分类路径的自适应包散射策略（Cparps）,根据路径的队列长度对路径进行分类,并将不同数据流隔离在不同类型的路径上进行随机散射。实验结果表明,在对称和非对称拓扑下Cparps协议的乱序次数和数据包个数都相对较少,说明Cparps可以有效缓解数据包的乱序。     

一种基于网络编码的应用层多播算法

研究了对多播网络进行网络编码的方法,提出了一种基于网络编码的应用层多播算法.该算法在计算网络拓扑时考虑了链路的花费.源端和-中间节点使用随机线性编码方法进行编码,在目的端进行解码操作使得目的端能从乱序的信息和部分丢失的信息中恢复出原始数据,提高了网络的可靠性.通过对ns-2的扩展并进行仿真实验,结果证明了基于网络编码的应用层多播算法是可以提高网络的吞吐量,并且和网络中最大的吞吐量比较接近.在信息块不是很大的情况下,编码延迟率的增长是在一定的范围内的.     

基于区域道路实况数据的交通行为谱分析方法

针对国内外有关交通行为谱研究的特征指标和评价指标不完善,且不能定量分析等问题,设计并定义了相应的特征指标和评价指标以建立完整的、能够定量化的分析区域交通行为数据的交通行为谱体系。首先基于交通行为特征,采用改进的层次分析法（AHP）对交通秩序类型进行了分类;其次采用多数据融合的实时系统集成（RTSI）算法对某路段交通安全性进行综合评判。最后开发了交通行为谱分析工具,该工具能根据交通实况数据计算区域路段的交通安全指数,较为完备地分析该路段内的交通行为。     

包延迟测量中的数据有效性判定与误差估算

Internet上的包延迟特性对于Internet的行为研究来说是最为复杂也是最为重要的一环。该文对Internet中的包延迟测量方法进行了深入研究，提出了自己开发的延迟测量工具Posip。Posip采用Poisson采样方法进行延迟测量，避免了普通周期性测量带来的负面效应。另外Posip还采用超时等待方式避免了 Ping工具中的“应答包丢失”现象。同时，对测量数据的处理，提出了测量数据有效性判定规则，并提出了一种延迟测量中线路误差的测量方法。使用上述方法在我国教育网和科研网上43条链路上进行了实际测量和验证，并取得了良好的效果。     

一种测量任意链路可用带宽的方法

可用带宽测量对于网络行为分析、网络服务质量(quality of service,简称QoS)的验证等有很重要的作用.现有可用带宽测量工作主要集中在端到端路径可用带宽测量,仅提供路径上承压链路(tight link)的信息,而不能提供其他关键链路的信息.为此,提出一种新颖的链路可用带宽测量算法LinkPPQ(trains of pairs of packet-quartets used to measure available bandwidth of arbitrary links),它采用由四探测分组结构对构成的探测序列,能够测量网络中任意链路的可用带宽,并跟踪该链路上背景流的变化.在仿真环境和实际网络环境下研究了LinkPPQ 的性能.仿真结果表明,在几种不同背景流场景下,对于具有单狭窄链路的路径和具有多狭窄链路的路径,LinkPPQ 都能够对各个链路的可用带宽进行有效的测量.绝大多数情况下测量误差小于30%,且具有较好的测量平稳性.实验网的实验结果也表明,LinkPPQ 可以准确测量以下几种情况下的链路的可用带宽:a) 从容量为10Mbps 的链路准确地测量一条100Mbps 链路的可用带宽;b) 准确测量容量10 倍于紧邻其后狭窄链路的容量的链路的可用带宽;c) 准确测量具有多狭窄链路的路径上各狭窄链路的可用带宽.     

系统域网络基于消减策略网络断层扫描方法

系统域网络是高性能计算机、数据中心的重要组成部分,当前系统域网络存在网络规模庞大,内部链路繁多,网络流量行为复杂和各种应用对网络性能状况敏感度高等特点,致使采用传统网络断层扫描方法进行性能测量的计算复杂度呈指数级增长。针对上述问题提出了一种基于消减策略的网络断层扫描方法(network tomography based on reduction strategy,NTRS)。该方法提出了预处理原则对实际物理拓扑进行策略约束,充分利用内部链路已知的性能信息缩小性能测量网络区域和关键链路覆盖,依据测量结果计算链路性能协方差,筛除性能状况较好的链路集合,实现链路数量的有效约简,进而很大程度上提高了网络诊断的准确性。通过模拟实验验证了NTRS方法的有效性,实验结果表明该方法能缩小链路性能参数的测量规模,降低计算的复杂度。     

An optical burst reordering model for time-based and random selection assembly strategies

Contention resolution schemes in optical burst switched networks (OBS) as well as contention avoidance schemes delay burst delivery and change the burst arrival sequence. The burst arrival sequence usually changes the packet arrival sequence and degrades the upper layer protocols performance, e.g., the throughput of the transmission control protocol (TCP).In this paper, we present and analyze a detailed burst reordering model for two widely applied burst assembly strategies: time-based and random selection. We apply the IETF reordering metrics and calculate explicitly three reordering metrics: the reordering ratio, the reordering extent metric and the TCP relevant metric. These metrics allow estimating the degree of reordering in a certain network scenario. They estimate the buffer space at the destination to resolve reordering and quantify the number of duplicate acknowledgements relevant for investigations on the transmission control protocol.We show that our model reflects the burst/packet reordering pattern of simulated OBS networks very well. Applying our model in a network emulation scenario, enables investigations on real protocol implementations in network emulation environments. It therefore serves as a substitute for extensive TCP over OBS network simulations with a focus on burst reordering.     

Metrics for MT evaluation: evaluating reordering

Translating between dissimilar languages requires an account of the use of divergent word orders when expressing the same semantic content. Reordering poses a serious problem for statistical machine translation systems and has generated a considerable body of research aimed at meeting its challenges. Direct evaluation of reordering requires automatic metrics that explicitly measure the quality of word order choices in translations. Current metrics, such as BLEU, only evaluate reordering indirectly. We analyse the ability of current metrics to capture reordering performance. We then introduce permutation distance metrics as a direct method for measuring word order similarity between translations and reference sentences. By correlating all metrics with a novel method for eliciting human judgements of reordering quality, we show that current metrics are largely influenced by lexical choice, and that they are not able to distinguish between different reordering scenarios. Also, we show that permutation distance metrics correlate very well with human judgements, and are impervious to lexical differences.     

Packet reordering in high-speed networks and its impact on high-speed TCP variants

Several recent Internet measurement studies show that the higher the packet sending rate, the higher the packet-reordering probability. This implies that recently proposed high-speed TCP variants are more likely to experience packet reordering than regular TCP in high-speed networks, since they are designed to achieve much higher throughput than regular TCP in these networks. In this paper, we first study the characteristics of packet reordering in high speed networks. Second, we verify the impact of packet reordering on high speed TCP variants and evaluate the effectiveness of the existing reordering-tolerant TCP enhancements using simulations. Our simulation results demonstrate that high-speed TCP variants perform poorly in the presence of packet reordering, and existing reordering-tolerant algorithms can significantly improve the performance of high-speed TCP variants.     

基于NewReno拥塞控制机制的TCP分组乱序影响分析

报文乱序是互联网中普遍存在的现象,深入理解互联网报文乱序的行为属性和影响,对于设计高性能网络协议和新型网络设备体系结构、优化网络整体效能具有重要的意义。本文建立了NewReno拥塞控制算法的工作模型,通过实验验证了模型的有效性。基于该模型,分析报文乱序对TCP连接的影响,发现乱序报文对TCP连接吞吐率的影响显著,而且乱序报文的影响与报文丢失率紧密相关:当报文丢失率较低时,报文乱序会导致TCP连接的吞吐率急剧下降;当丢失率较高时,报文乱序的影响比较小。这些结论可为报文乱序友好的网络协议和新型网络设备体系结构的设计提供指导。     

一种采用流切割实现报文保序的负载均衡算法

网络链路过载或链路失效时,使用负载均衡技术可以避免网络发生拥塞。负载的分派粒度决定了负载均衡系统的均衡性能。分派粒度越细,均衡效果越理想。基于包水平粒度的负载分派可以实现理想的均衡性能,但是会造成同一TCP业务流中报文乱序;基于流水平的分派可以保证报文不乱序,但均衡效果不理想。提出了按照报文段粒度分派负载的FSLB算法。仿真实验表明,该算法可避免报文乱序并能达到较理想的均衡效果。     

Modelling and robust stability of networked control systems with packet reordering and long delay

In this article, modelling and robust stability of networked control systems (NCS) are discussed. Considering the existence of packet reordering and network-induced delay, a new mathematical model of NCS whose network-induced delay is longer than one sampling period is obtained, which can fully describe packet reordering and effectively eliminate the impact of packet reordering on the performance of NCS such that the newest control input can be executed by the actuator. Based on this model, the time-varying NCS is converted into an uncertain discrete linear system with multi-step delay in terms of matrix theory. Furthermore, a sufficient condition for robust stability of NCS is presented. Linear matrix inequality approach has been employed to solve the controller design problems. Numerical examples are compared with previous schemes to demonstrate the effectiveness of the proposed method.     

一种MPLS的故障恢复方案*

在备份路径预建立的前提下,提出了一种MPLS的故障恢复方案--EB,该方案在现有的Bridge模型基础上引入了Backpressure信令以解决分组失序问题.理论性能评估及仿真实验均表明,EB较好地继承了Bridge方案原有的优势,并且解决了分组失序问题.经综合评估,EB有着良好的恢复性能.     

Sorting Reordered Packets with Interrupt Coalescing

TCP performs poorly in networks with serious packet reordering. Processing reordered packets in the TCP-layer is costly and inefficient, involving interaction of the sender and receiver. Motivated by the interrupt coalescing mechanism that delivers packets upward for protocol processing in blocks, we propose a new strategy, Sorting Reordered Packets with Interrupt Coalescing (SRPIC), to reduce packet reordering in the receiver. SRPIC works in the network device driver; it makes use of the interrupt coalescing mechanism to sort the reordered packets belonging to the same TCP stream in a block of packets before delivering them upward; each sorted block is internally ordered. Experiments have proven the effectiveness of SRPIC against forward path reordering.     

Minimum information rate for observability of linear systems with stochastic time delay

This paper investigates the state estimation problem for linear time-invariant systems where sensors and controllers are geographically separated and connected via stationary memoryless uncertain digital communication channels with the information rate limitation. Such channels introduce data packet dropout, stochastic time delay, and data packet reordering. In the TCP/IP model, the transport layer provides a reliable service on top of an unreliable communication channel. However, the performance of such systems is still affected by packet reordering and stochastic time delay. In particular, we present an optimal encoding and decoding scheme to guarantee observability by employing the minimum information rate, and derive a necessary and sufficient condition on the delay distribution and information rate for observability. Illustrative examples are given to demonstrate the effectiveness of the proposed scheme.