Figure. Cell switching principles: (a) routing schematic; (b) VP routing; (c) VC routing. (a) PCI =,,, 4 4 PCI =, 4 4 6 PCI = 6, Link/Port RT Link/Port RT Link/Port RT In Port PCI 4 Out Port PCI 4 6 Port In PCI 4 Out Port PCI In Port PCI 6 Out Port PCI 4 (b) VC VC VP VC VC 4 VP VP switch VP VP VC VC 4 VC 5 VC 6 PCI = protocol connection identifier RT = routing table VP = virtual path VC = virtual channel VC 5 VC 6 VP VP VC VC (c) VP VC VC VC switch VC VC VP VP VC VC VC VC 4 VP VP VC VC VC VC VP Pearson Education Limited
Figure. ATM cell formats: (a) user network segment; (b) within network, network network interface. (a) Bits 8 7 6 5 4 (b) 8 7 6 5 4 Header 4 5 6 Payload 5 Octets PTI: user data, no congestion, SDU type user data, no congestion, SDU type user data, congestion, SDU type user data, congestion, SDU type Network control GFC = generic flow control VPI = virtual path identifier VCI = virtual channel identifier PTI = payload type identifier CLP = cell loss priority HEC = header error checksum Pearson Education Limited
Figure. ATM switch architectures: (a) general structure; (b) time-division bus schematic; (c) fully-connected matrix switch. (a) Input ports Output ports Input links Output links (b) IC = input controller OC = output controller N time (c) cell arrival time Matrix switch Routing addresses Pearson Education Limited
Figure.4 Delta switch matrix example. Pearson Education Limited
Figure.5 Batcher Banyan switch matrix. Pearson Education Limited
Figure.6 ATM protocol architecture. Pearson Education Limited
Figure.7 ATM adaption layer: (a) service class relationship; (b) sublayer protocols and their functions. Pearson Education Limited
Figure.8 SAR protocol data unit types: (a) AAL ; (b) AAL. Pearson Education Limited
Figure.9 CS and SAR PDU formats: (a) AAL/4; (b) AAL 5. Pearson Education Limited
Figure. Principle of operation of generic cell rate algorithm. Time, t T = /PCR CDVT (i) Cell Cell (ii) Cell Cell Cells meeting contract (iii) Cell Cell (iv) Cell Cell Cell violating contract t PCR = peak cell rate t + T CDVT = cell delay variation time Pearson Education Limited
Figure. ATM LAN schematic. Pearson Education Limited
Figure. ATM LAN routing example: (a) network segment; (b) example routing table entries. (a) 4 (b) In Out In Out RCU : Port VPI VCI Port VPI VCI Port VPI VCI Port VPI VCI SC CC CLS SC CC CLS Pearson Education Limited
Figure. Continued (b cont.) In Out In Out RCU : Port VPI VCI Port VPI VCI Port VPI VCI Port VPI VCI SC CC CLS SC CC CLS In Out In Out SW : Port VPI VCI Port VPI VCI Port VPI VCI Port VPI VCI SC CC CLS SC CC CLS 4 5 6 4 5 6 = In Out In Out SW : Port VPI VCI Port VPI VCI Port VPI VCI Port VPI VCI SC CC CLS SC CC CLS CLS' 4 5 6 Z 4 4 Y 4 Y 6 Z 4 4 Y 4 Y 6 Z 4 5 6 Z = Y = /5 Z = 64 In Out In Out CLS: VPI VCI VPI VCI VPI VCI VPI VCI 6 Y Y Y Y 6 = Y = /5 SCP: Calls in progress Signaling channel VPI VCI Call channels In Out Port VPI VCI Port VPI VCI Call type details 4 5 Y Y SC = signaling channel CLS = workstation/cls channel CC = call channel CLS' = CLS/server channel Pearson Education Limited
Figure. LAN emulation: (a) terminology and networking components; (b) unicast protocol architecture; (c) multicast protocol architecture. Pearson Education Limited
Figure. Continued Pearson Education Limited
Figure.4 Protocol architecture to support classical IP over an ATM LAN. Pearson Education Limited
Figure.5 DQDB/MAN network architectures: (a) single-site MAN; (b) dual-site private network; (c) wide area multiple MAN network. (a) > 5 km ( miles) Looped bus MAN (4/45/4/55 Mbps) Dual contradirectional buses Access node/customer network interface unit = bridge/router (b) Main site IGW Subnetwork routers IGW Remote site Looped bus (4/55 Mbps) Interconnecting leased circuit (4/45 Mbps) Open bus (4/45 Mbps) = private branch exchange IGW = isochronous gateway (c) City/MAN A City/MAN B Intercity duplex circuits (4/45/4/55 Mbps) MSS = MAN switching system City/ MAN C Pearson Education Limited
Figure.6 DQDB architectures: (a) open bus; (b) looped bus; (c) example reconfigured looped bus networks. (a) Read Write Bus A Head of Bus A Head of Bus B Bus B Write Read (b) Head of Bus B Head of Bus A = slot generator = bus terminator Bus A Bus B Bus B (c) Bus A Bus A Head of Bus B Bus B Bus B Bus A Head of Bus B Head of Bus A Head of Bus A Bus A (i) Link failure (ii) Node failure Pearson Education Limited
Figure.7 DQDB protocol architecture: (a) layer functions; (b) example physical layer convergence function. Pearson Education Limited
Figure.8 DQDB access control principles: (a) request/busy bits; (b) request counters; (c) queuing mechanism. Pearson Education Limited
Figure.9 Flowchart of the algorithm used to control the transmission and reception of segments on busa of a dual bus DQDB subnetwork. Pearson Education Limited
Figure. Bandwidth balancing: (a) unfairness effect; (b) remedial actions; (c) effect on mean access delay. Pearson Education Limited
Figure. Priority access control: (a) no segments waiting; (b) segments queued at priority. Pearson Education Limited
Figure. Slot and segment formats: (a) slot header; (b) connectionless data segment format. Pearson Education Limited
Figure. SMDS internetworking protocol architectures: (a) bridges; (b) routers. Pearson Education Limited
Figure.4 Frame transmission overheads: (a) initial MAC PDU format; (b) frame segmentation. Pearson Education Limited
Figure.5 Connectionless working over wide area ATM networks: (a) ATM MAN switching network; (b) broadband ISDN. Pearson Education Limited
Summary Figure.6 Broadband ATM networks, chapter summary. Broadband ATM networks Cells Cell switching Statistical multiplexing Asynchronous transfer mode (ATM) Cell format Switch architectures Non-blocking/blocking Self-routing Protocol architecture ATM adaptation layer (AAL) (AAL,, /4, 5) ATM layer (CBR, VBR(RT/NRT), ABR, UBR) ATM LANs ATM MANs (DQDB) Architecture Call processing (LAN emulation, IP-over-ATM) MAC protocol Protocol architecture (SMDS/CBDS) ATM WANs Pearson Education Limited
Example. A segment of an ATM network is shown in Figure.(a). The numbers alongside each RCU/switch are the port identifiers. Assume that semipermanent VCs are to be set up by network management between stations A, B, C and D, firstly, to the SCP for on-demand calls for both signaling and information transfer and secondly, to the CLS/LES for the cell streams relating to connectionless calls. Also, assume that a separate VC is required to connect the server to the CLS/LES. Derive typical routing table entries for RCU and RCU and SW and SW to provide these connections assuming VP-only switching is used within the network switches and, within the RCUs, the VPI/VCI field is used to identify specific calls/stations. Answer: A suitable set of routing table entries is given in Figure.(b). Note the following points when interpreting the entries: The SCP, CLS, and server are all connected to their switches by separate transmission lines. For on-demand calls, two separate VCs are shown: one between each workstation and the SCP for the signaling messages associated with a call the signaling channel (SC) and the other for the cell streams associated with the call the call channel (CC). Since the latter are to be semipermanent, they are shown set up between each workstation and SW. Alternatively, they could be set up on demand by the SCP between each pair of workstations involved in a call. For connectionless traffic, a separate VC is required between each workstation and the CLS and also between the CLS and the server. The SCP, CLS, and server all use the combined VPI/VCI in the cell header to identify the cells relating to specific calls/server transactions. Pearson Education Limited
. Continued On the station side of each RCU, the VCI field identifies the port number and hence station within each virtual path. Also, in this example, only three virtual paths are required per RCU rather than per station. This allows the approach to be scaled to large installations. Within the network, all switching is carried out using VPIs only. To set up an on-demand call, the SCP creates entries in the routing table of SW to link the port/vci of the calling party to that of the called party. For connectionless traffic, when relaying the cell streams received from each station to the server, the CLS assigns a new VPI/VCI. Also, in order to relay the cell streams in the reverse direction, it maintains a table that maps the incoming VPI/VCI from the stations to those used for communicating with the server. When responding to a request, the server uses the same VPI/VCI values for the cells making up the response as were used in the request. Pearson Education Limited
Example. Derive a flowchart showing the steps taken by the queued arbitrated function to effect the transmission of a set of queued segments produced by the MAC convergence function on a single bus of a dual-bus DQDB subnetwork. Answer: A flowchart showing the steps to control the transmission of segments on bus A is given in Figure.9. Consider the following points when interpreting the figure: On receipt of a full slot on either bus, the queued arbitrated function simply passes the contents of the slot payload directly to the MAC convergence function. This function determines whether the segment is intended for this node. Only a single segment can be queued for transmission by the queued arbitrated function at one time. Hence only after this function has transmitted a segment does it return to the output queue of the MAC convergence function to determine whether another segment is awaiting transmission. Pearson Education Limited
Example. A 5-octet MAC frame is to be transferred across an SMDS/DQDB subnetwork. Stating clearly any assumptions you make, derive the number of queued arbitrated slots that are required to carry out the transfer and hence the total number of overhead octets involved. Answer: With reference to Figure.4(b): MAC convergence protocol: adds octets to make the frame 5 octets which is an integral multiple of 4 octets; assuming a header extension and CRC are not used, a 4-octet header and a 4-octet trailer are added to create a (5 + 4 + 4) 54-octet IMPDU; total overheads = + 4 + 4 = octets; after segmentation, the IMPDU requires DMPDUs: containing a full complement of 44 octets and one with octets; total overheads are 4 for each of the DMPDUs (= 5) plus for the part-full EOM DMPDU. Queued arbitrated (QA) sublayer: a further 5 octets are added to each 48-octet DMPDU to create QA slots; total overheads = 5 = 65 octets. Pearson Education Limited
. Continued Physical layer: a further 4-octet header is added to each QA slot; total overheads = 4 = 5 octets. In summary: QA slots required = ; total overheads = + 5 + + 65 + 5 = octets. Pearson Education Limited