通信网理论基础（工程硕士）：Ch1 Networks and Layering

1、通信网理论基础（工程硕士）第二讲1zhanglinCourse Description在通信网络的性能分析中，需要对进入不同通信网络的各种流量给予建模较为深入、系统的掌握不同流量的建模方法应用随机服务系统或排队论进行分析和模拟；应用排队论对网络的性能进行分析，通过蒙特卡洛罗模拟完成数据采集和验证等；对整个网络，需要进行拓扑结构分析和流量安排、优化等。较好掌握一些拓扑结构分析和网络流量安排、优化的算法。2zhanglin3zhanglinCourse DescriptionNetworks and layering IntroductionOSI 7-layer architecturePoin

2、t-to-Point Protocols and LinksThe Data Link Layer: ARQ ProtocolsDelay ModelsQuery Theory & Littles theoremM/M/1, , M/M/m, M/G/1, queues etc.Multiple AccessThe Aloha protocol4zhanglinCourse DescriptionLocal Area NetworkHigh Speed Switch SchedulingData routingBroadcast routing & Spanning treesRouting

3、algorithmsFlow and Congestion ControlHigh layer ProtocolsTCP/IPATM Network5zhanglin6zhanglin234567891011121314Grade CompositionTotal 100.0%Quiz and discussion 2030%Textbook exercises Final Exam 7080%with book7zhanglinCh1 Networks and LayeringKnown and unknown Known and unknown（Feb., 12， 2002, Donald

4、 Rumsfeld ）As we know, There are known knowns. There are things we know we know. We also know There are known unknowns. That is to say We know there are some things We do not know. But there are also unknown unknowns, The ones we dont know We dont know. “据我们所知，有已知的已知，有些事，我们知道我们知道；我们也知道，有 已知的未知，也就是说，

5、有些事，我们现在知道我们不知道。但是，同样存在未知的未知有些事，我们不知道我们不知道。” 10/1/20229zhanglin已知与未知:美前国防部长拉姆斯菲尔德回忆录美国历史上最年轻的国防部长（ 1975-1977年 ），到最年长的国防部长（ 2001-2006年 ），Historical Overview10/1/2022zhanglin10Traditionally, the computer communication network is only a computer with remote terminals. Many of the interesting problems a

6、ssociated with data networks will not arise if we keep these centralized systems, such as the distributed control of the system, the relaying of messages over multiple communication links, the sharing of communication links between many users and processes. Historical Overview(cont.)10/1/2022zhangli

7、n11The reality is that the communication subnet of the network is central. Keep this picture of external devices around a communications subnet in mind will make it easier both to understand network layering later in this chapter and to understand the issues of distributed network control throughout

8、 the course.Network Coverage AreasWide Area Networks (WANS)Span large areas (countries, continents, world)Use leased phone lines (expensive!)1980s: 10 Kbps, 2000s: 2.5 GbpsUser access rates: 56Kbps 155 Mbps typicalShared comm links: switches and routersE.g, IBM SNA, X.25 networks, InternetLocal Area

9、 Networks (LANS)Span office or buildingSingle hop (shared channel) (cheap!)User rates: 10 Mbps 1 GbpsE.g., Ethernet, Token rings, Apple-talkMetro Area networks (MANS)Storage area networks10/1/2022zhanglin12Quiz 1: what is Internet, why it is called as the Network of network? (Is it like the Zeus, hi

10、sgodof gods?)Answer 1: From a more practical viewpoint, a network of network is much more complex than a single network. The problem is that each constituent subnet has its own conventions and control algorithms (i.e, protocols) for handling data, and the gateways and bridges must deal with this inh

11、omogeneity. We discuss this problem later after developing some understanding of the functioning of individual subnets. Network ServiceSynchronousSession appears as a continuous stream of traffic (e.g, voice)Usually requires fixed and limited delaysAsynchronousSession appears as a sequence of messag

12、esTypically burstyE.g., Interactive sessions, file transfers, emailConnection oriented servicesLong sustained sessionOrderly and timely delivery of packetsE.g., Telnet, FTPConnectionless servicesOne time transaction (e.g., email)QoS10/1/2022zhanglin13Connection vs. SessionConnectionSessionPhysical p

13、ath from client to instanceTypically between a client process and either a dedicated server or a dispatcherConnection vs. sessionA connection may have zero, one or more sessionsEach session is separate and independentLogical entity that exists in the instanceMessage arrival rate and variability of a

14、rrivalsSession Holding timeExpected message length and length distributionAllowable DelayReliabilityMessage and packet ordering10/1/2022zhanglin14Connection（连接）是一个物理概念，是指一个通过网络建立的客户端和专有服务器或调度器之间的一个网络连接Session（会话）是一个逻辑概念，它存在于实例中，一个Connection可以拥有多个Session，也可以没有Session，同一个Connection上的多个Session之间不会相互影响S

15、ession10/1/2022zhanglin15Some gross characteristics of Sessions:1, Message arrival rate and variability of arrivals2, Session Holding time3, Expected message length and length distribution4, Allowable Delay5, Reliability6, Message and packet orderingSwitching TechniquesCircuit SwitchingDedicated res

16、ourcePacket SwitchingShared resourcesVirtual circuitsDatagrams10/1/2022zhanglin16Circuit SwitchingEach session is allocated a fixed fraction of the capacity on each link along its pathDedicated resourcesFixed pathIf capacity is used, calls are blockedE.g., telephone networkAdvantages of circuit swit

17、chingFixed delaysGuaranteed continuous deliveryDisadvantagesCircuits are not used when session is idleInefficient for bursty trafficCircuit switching usually done using a fixed rate stream (e.g., 64Kbps)Difficult to support variable data rates10/1/2022zhanglin17Problems with Circuit SwitchingMany da

18、ta sessions are low duty factor (bursty),(message transmission time)/(message inter-arrival time) 1Same as: (message arrival rate) * (message transmission time) Synchronous (e.g., voice)Packet switching = Asynchronous (e.g., Data)Virtual circuits = Connection orientedDatagram = Connectionless10/1/20

19、22zhanglin22Circuit vs. packet switchingCan circuit switched network be used to support data traffic?Can packet switched network be used for connection oriented traffic (e.g., voice)?Need for Quality of service (QoS) mechanisms in packet networksGuaranteed bandwidthGuaranteed delaysGuaranteed delay

20、variationsPacket loss rateEtc.10/1/2022zhanglin237 Layer OSI Reference Model10/1/2022zhanglin24Layering, or layered architecture, is a form of hisrachical modularity that is contral to data network design. The concept of modularity is as old as engineering. The work module is used to refer either to

21、 a device or to a process within some computer system. The module performs a given function in support of the overall function of the system.10/1/2022zhanglin25What is unusual about the layered architecture for data networks is that the black boxes at the various layers are in fact distribued black

22、boxes. LayersPresentation layerProvides character code conversion, data encryption, data compression, etc.Session layerObtains virtual end to end message service from transport layerProvides directory assistance, access rights, billing functions, etc.Standardization has not proceeded well here, sinc

23、e transport to application are all in the operating system and dont really need standard interfacesFocus: Transport layer and lower10/1/2022zhanglin26Physical LayerProvide a virtual link for transmitting a sequence of bits between any pair of nodes, called as Virtual Bit Pipe. Propagation delaysTime

24、 it takes the signal to travel from the source to the destinationSignal travel approximately at the speed of light, C = 3x108 meters/secondE.g.,LEO satellite: d = 1000 km = 3.3 ms prop. DelayGEO satellite: d = 40,000 km = 1/8 sec prop. DelayEthernet cable: d = 1 km = 3 s prop. DelayTransmission erro

25、rsSignals experience power loss due to attenuationTransmission is impaired by noiseSimple channel model: Binary Symmetric ChannelP = bit error probability Independent from bit to bitIn reality channel errors are often burstyzhanglin27Quiz 2: Is it better to have a physical layer that transmits R bit

26、s/sec with an error rate of 10-6 or a modem that transmits 2R bits/sec with an error rate of 10-4? Answer 2: This cannot be answered without some knowledge of how errors are eliminated at higher layers of the architecture. Conversely, decisions on how and where to eliminate errors at higher layers s

27、hould depend on the error rate and error characteristics at the physical layer.Link LayerCovert the unreliable bit pipe at layer 1 into a higher levelVirtual communication link for sending packets asynchronously but error-free in both directions over the link Responsible for error-free transmission

28、of packets across a single linkFramingDetermine the start and end of packetsError detectionDetermine which packets contain transmission errorsError correctionRetransmission schemes (Automatic Repeat Request (ARQ)zhanglin28The MAC SublayerThe appropriate layers for multi-access communication are some

29、what different from those in networks of point-to-point links. As in the example Figure, the physical medium is accessed by all three users, each of whom hears the transmitted signals of the others. The DLC sublayers sees virtual point-to-point bit pipes below it. The MAC sublayer sees a multiaccess

30、 bit pipe, and the modems access the actual channel. zhanglin29Network LayerThe network layer module accepts incoming packets from the transport layer and transit packets from the DLC layerIt routes each packet to the proper outgoing DLC or (at the destination) to the transport layerTypically, the n

31、etwork layer adds its own header to the packets received from the transport layer. The network layer module uses the header of incoming packet, along with stored information at the module, to accomplish routing and flow control.10/1/2022zhanglin30Each node contains one networkLayer module plus oneLi

32、nk layer module per linkflow control, or congestion controlThe other major function of the network layer, along with routing. Some authors make a distinction between flow control and congestion control, viewing the first as avoiding sending data faster than the final destination can absorb it, and t

33、he second as avoiding congestion within the subnet.Fundamentally, congestion occurs when the users of the network collectively demand more resources than the network (including the destination sites) has to offer. The control of packet flow into the network must be done in such a way as to prevent c

34、ongestion and also to provide equitable service to the various sessions. zhanglin31One might hope that the high link capacities that will be available in the future will make it possible to operate networks economically with low utilization, thus making flow control unnecessary. It is called over-provisioning（超量配置）.Unfortunately, this view appears overly simplis