Asynchronous Transfer Mode: An Introduction

by
Cathy S. Moore
xxx-xx-xxxx
EE6304, Spring 1997
Section 403D
123 Street
City, ST zip code
 ABSTRACT
Asynchronous Transfer Mode (ATM) is quickly evolving to enable the full use of the Broadband
Integrated Services Digital Network (BISDN). ATM will allow various services to utilize the same
transfer network. A Protocol Reference Model (PRM) has been developed for managing the
relationships among the various services and the physical media. ATM is connection-oriented
and converts user data in various formats to uniformly sized cells for transport. The cells are
routed dynamically through the ATM network utilizing virtual circuits and paths. This dynamic
routing helps to avoid lost data in the event of congestion or a service interruption. This paper is
presented as an introduction to ATM. It is intended to give the reader an overview of several
ATM topics.

Asynchronous Transfer Mode: An Introduction

The Broadband Integrated Services Digital Network (BISDN) supports digital transmission at

rates greater than 1.544 Mbps. This service includes the transfer of voice, video, and data

through public and private networks. Asynchronous Transfer Mode (ATM) is being developed as

one of the techniques that will enable the BISDN to transport this wide variety of services. ATM

standards are evolving under the guidance of the International Telecommunications Union-

Telecommunications Standards Sector (ITU-T). These standards are being developed to enable

services requiring large bandwidths such as distributed supercomputing and telemedicine and

services requiring a smaller bandwidth such as voice to operate in a cost effective manner on

the same network. The standards also define the protocols required to interface other network

services such as Switched Multimegabit Data Services (SMDS). ATM standards are written in

such a way that services that are in use today and new services that are under development can

use the same network.

ATM combines circuit switch routing of public telephone networks, packet switching of
1
private data networks, and the asynchronous multiplexing of a packet switch. It is a cell

switching and multiplexing technique that supports switching in public and private networks.

Constant transmission delay and guaranteed capacity, two benefits of circuit switching, are
2
combined with the flexibility and efficiency of handling intermittent packet type data. ATM is

connection-oriented and converts all incoming data into a 53 byte cell. The cell consists of a five

1
Gunnar Karlson, “Asynchronous Transfer of Video,” IEEE Communications Magazine,
Volume 34, Number 8 (August, 1996), 118.
2
“Asynchronous Transfer Mode,” (November 14, 1995): p. 6, online, Internet, February 26,
1997.

2
byte header and 48 bytes of user or control data. Uniform cell size enables multiple types of

networks to interface with the ATM network.

1.0 ATM CONNECTION TYPES

There are two types of connections available at this time with ATM, point-to-point and point-

to-multipoint. Point-to-point connections can be unidirectional or bidirectional. Point-to-multipoint

can be unidirectional only. Multipoint-to-multipoint is not available yet. There is no method for a

receiver to identify the cells from individual sources since the cells would be interleaved from

multiple sources. This prohibits proper reassembly of the cells into the proper data frames at the

receiver.

2.0 ATM MULTIPLEXING

ATM uses asynchronous multiplexing instead of synchronous. In synchronous time division

multiplexing (TDM) users are pre-assigned to time slots. In ATM time slots are assigned only

when a user has data to send. TDM is inefficient in relation to ATM in two respects. In TDM an

idle code is transmitted in a time slot in which there is no user traffic. In ATM idle codes are not

required when there is no user data to send. ATM does, however, use idle cells to adapt the rate

of the ATM cells to the physical transmission medium. The idle cells are discarded at the

receiver and are not processed in the same manner as user data. This is more efficient than

synchronous TDM since in that method idle time slots are sent and processed as user data. ATM

transport is an advantage for the user since he pays only for the cells he sends and not for a

dedicated channel he may not fully utilize.

Also, in synchronous TDM if a user has a lot of data to send he must wait until his time slot

arrives even if all of the other timeslots are empty. With ATM a user sends data when he needs

to send. This is certainly more efficient and is an advantage for the network provider since the

network is used to process and transport data instead of idle codes.

3
3.0 ATM QUALITY OF SERVICE

One important concept developed in ATM is quality of service (QOS). When an endstation

connects to the ATM network it establishes its requirements for the quality of the connection.

These requirements are known as QOS parameters and include the required bandwidth, average

sustained bandwidth, and burst size.3 ATM devices must adhere to these requirements and they

do so by various methods. Switches may use queues to prevent data bursts, limit the peak data

rate, and smooth jitter. Congestion may be controlled by routing cells through less congested

nodes or switches or by discarding cells if the user agrees. The discard agreement is negotiated

when the service application is made.

4.0 ATM VIRTUAL CIRCUITS AND PATHS

ATM also uses virtual circuits and paths (VC’s and VP’s) extensively. A VC is a bidirectional

logical connection between the ends of a communication connection. A VP is a bidirectional

logical grouping of VC’s that have the same destination. The VC’s and VP’s are used to transport

cells from one ATM entity to the next. Their use will be explained later.

5.0 ATM CELL STRUCTURE

Each ATM cell is 53 bytes in length. The first five bytes form the cell header while the last

48 bytes carry user or control data. The information in the cell header is used to establish

connections and route cells. The cell header uses one of the two formats defined by the ITU-T.

The formats are the User-Network Interface (UNI) and the Network-Node Interface (NNI). The

UNI defines the interface between the user and the network while the NNI defines the interface

between ATM networks and ATM nodes. In the UNI header format there are six fields that form a

five byte header. They are as follows:

1) Generic Flow Control (GFC)- This field is four bits in length and provides local functions such

as identifying multiple stations that share the same ATM interface. It provides flow control at the

3
Ibid., p. 9.

4
UNI for traffic originating at the user and directed toward the network.4 It does not control

network-to-network or network-to-user flow.

2) Virtual Path Identifier (VPI)- This field is eight bits in length and is used with the next field to

identify the next destination of the cell as it is routed through the ATM network. This yields 256

(28 = 256) possible VP’s.

3) Virtual Circuit Identifier (VCI)- This field is 16 bits in length and it also identifies the next

destination of the cell as it progresses through the network. This provides 65,536 (216 = 65,536)

possible VC’s.

4) Payload Type (PT)- This field is three bits in length. The first bit indicates whether a cell has

user data or ATM control data. If bit one is set to 1 the cell contains ATM control data that wiIl be

used for management functions. If bit one is set to 0 the cell contains user data. When bit one is

set to 0 bit two is used to indicate congestion. A congested switch will set bit two to 1 to inform

an end node that it is congested.5 The third bit is used in some applications to indicate if this is

the last cell in a user frame.

5) Congestion Loss Priority (CLP)- This field is one bit in length and indicates if a cell can be

discarded if it encounters extreme congestion in the network. The value of this bit depends on

the QOS parameters requested by the user when the service is requested. If the CLP bit is set to

1 the cell may be discarded during congestion. Cells with the CLP bit set to 0 have higher priority

and are not discarded if possible.6 Cells in an application such as video coding may be dropped

without degrading the video quality. Also, this bit could be set for those user cells transmitted in
7
excess of the negotiated rate.

6) Header Error Control (HEC)- This field is eight bits in length and uses the polynomial

x8 + x2 + x + 1 to perform a cyclic redundancy check (CRC).8 The polynomial is only applied to

the first four bytes of the header. The contents of this field is the resulting eight bit CRC.

4
Prem Chandran, “Asynchronous Transfer Mode,” p. 3, online, Internet, February 26, 1997.
5
“Cell Switching (ATM),” p. 4, online, Internet, April 3, 1997.
6
“Asynchronous Transfer Mode,” op. cit., p. 5.
7
“Cell Switching (ATM),” op. cit., p. 4.
8
“Asynchronous Transfer Mode,” op. cit., p. 4.

5
 The NNI header is the same as the UNI header except that there is no GFC field. GFC is

only used for traffic originating at a user and transmitted toward the network. It is not required at

the NNI. Instead those four bits are added to the VPI field for a total of 12 bits. This increases the

number of VP’s from 256 to 4,096 (212 = 4,096).

6.0 BISDN PROTOCOL REFERENCE MODEL

A Protocol Reference Model (PRM) has been defined for BISDN for ATM. The model has

three layers that are similar to the first three layers of the Open Systems Interconnection (OSI)

Reference Model. The physical layer of BISDN is roughly equivalent to OSI layer one and

performs bit level functions such as timing for those services that require timing.9 The ATM layer

is similar to the lower edge of OSI layer two. It generates the cell header. The ATM Adaptation

Layer (AAL) is similar to the upper edge of layer 2 and layer three of the OSI Reference Model. It

adapts various services to ATM cells. Above the AAL there are higher layer protocols

representing the traditional transports and applications of the OSI Reference Model.10 The AAL

also provides service dependent functions to those upper layers of the OSI Reference Model.

The BISDN PRM also uses three planes. The three planes are user, control, and

management. The user plane provides for user information to flow along with its associated

control fields for flow control and error recovery. The control plane includes all connection control

functions such as signaling functions required for connection setup, supervision, and release.

The management plane provides both layer and plane management functions. The layer

management function performs layer specific management functions while plane management

coordinates the functions of the entire system.11 See Figure 1 for a model of the layers and

planes in the BISDN PRM.

9
Srinivas Reddivalam, “Asynchronous Transfer Mode Tutorial,” p. 3, online, Internet,
February 20, 1997.
10
“Asynchronous Transfer Mode,” op. cit., p. 5.
11
Tarek N. Saadawi, Mostafa H. Ammar, and Ahmed El Hakeem, Fundamentals of
Telecommunications Networks (New York: John Wiley and Sons, Inc., 1994), p. 389.

6
6.1 Physical Layer

The physical layer contains two sublayers, the Physical Medium (PM) and Transmission

Convergence (TC) sublayers, and is common to all services. ATM can be transported using

fiber, coaxial cable, or twisted pair. The PM sublayer provides the bit transmission functions

including bit alignment and timing, line coding, and electrical/optical conversion.12 The physical

connection to the transmission medium is handled at this sublayer.

The TC sublayer has five functions. The first is generation and recovery of the transmission

frame. This function places cells in the proper format for the physical medium in use. The

second function is to adapt the received data to the ATM cell flow. When receiving data from the

user this transmission frame adaptation function adapts the received payload to the ATM cell

structure. When transmitting data toward the user, this function removes the ATM cell structure.

The third function is the cell delineation function which enables the ATM receiver to recover cell

boundaries. The fourth function is header error control. This is where the eight bit CRC is

formed and placed in the HEC field of the cell header in the transmit direction. At the receiver

the error control code is generated again on the first four bytes of the cell header and compared

to the CRC value that was received in the HEC field. If the value matches, processing continues.

If not, the cell is discarded. This prevents cells from reaching incorrect destinations if there are

errors in the VCI or VPI fields. This reduces requests for retransmission, helps to control

congestion, and ensures high speed data transport.13 If cells are discarded or there are errors the

ATM node or switch does not request retransmission. The application must initiate the

retransmission request.

The last function of the TC sublayer is cell rate decoupling. This function inserts idle cells in

the transmit direction to adapt the rate of the ATM cells to the payload capacity of the

transmission system. It also discards idle cells in the receive direction. As stated in Section 2.0

ATM Multiplexing these idle cells are not dependent on user data as in synchronous TDM. See

Figure 2 for a diagram of the Physical Layer.

12
Reddivalam, op. cit., p. 3.
13
Laura McInerney, “A Layered Approach,” Telephony, (February 10, 1997), 18.

7
6.2 ATM Layer

The ATM layer is also common to all services. It handles the functions of the cell header

independent of the type of user data or physical media. This maintains flexibility in the ATM

layer. With the exception of the HEC value the ATM layer generates and extracts the cell header

in the transmit and receive directions respectively. The HEC value is obtained from the TC

sublayer of the physical layer and is placed in the cell header at this layer.14

The ATM layer also multiplexes cells from individual VC’s and VP’s and different types of

physical media into one cell stream in the transmit direction. In the receive direction, the cell

stream is divided into individual cell flows with respect to the VC’s or VP’s of the cells. Since this

layer is independent of the physical transport media cells from various types of services and

media can be combined into one cell stream for transmission. They can be recovered and placed

in their respective formats at the receiver.

Each entity that is part of or accesses an ATM network has some type of address. The

address may be an ATM address or an address that is used in the application such as a Media

Access Control (MAC) address used in the Institute of Electrical and Electronics Engineers

(IEEE) 802 Local Area Network (LAN) specification.15 At an ATM node or switch the ATM layer

obtains the address or Service Access Point (SAP) identifiers from the next layer, the AAL, and

translates them into VCI’s and VPI’s. At the ATM layer the VCI’s and VPI’s are used to determine

the next destination of the cell. When the destination is determined the VCI and VPI values are

changed before the cell is transmitted to the next node or switch.

GFC is another function of the ATM layer. As explained in Section 5.0 ATM Cell Structure it

supports control of ATM traffic flow in a customer network. The information is contained in the

GFC field of the cell header. See Figure 2 for a diagram of the ATM Layer.

6.3 AAL Layer

14
Reddivalam, op.cit., p. 3.
15
“Asynchronous Transfer Mode,” op. cit., p. 9.

8
 Since the various BISDN applications do not require the same functions, the applications

are placed in categories of service classes and the third layer of the PRM, the AAL, handles the

functions of the various services. This layer provides the link between the various applications

and ATM functions.

The AAL supports the higher layer functions of the user and control planes. It translates

between larger service data units (SDU’s) of the upper layer processes and ATM cells through

two sublayers, the Segmentation and Reassembly (SAR) sublayer and Convergence Sublayer

(CS). The SAR receives cells from the upper layer protocols (NetWare, AppleTalk, or Internet

Protocol for example) and breaks them into 48 byte segments to form the payload of the ATM

cell.16 At the receiver this sublayer reassembles the contents of the cells into data units to be

delivered to the higher layer protocols.17 The CS is for message identification and clock recovery

for those services that require it.

There are four different AAL classes, defined for different services.18 Class A is for constant

bit rate, connection-oriented services that require timing. Voice is an example. Class B is for

variable bit rate, connection-oriented services that require timing. Video is an example. Class C

is for variable bit rate, connection-oriented services that do not require timing. X.25 is an

example. Class D is for variable bit rate, connectionless services that do not require timing. LAN

interconnection is an example.19 Although ATM is connection-oriented, connectionless

applications are not excluded from using ATM for transport. The higher layers determine the

connection orientation for each application.

Four AAL types support the four classes. AAL1 and 2 correspond to Classes A and B

respectively. AAL 3/4 and 5 can be used for Classes C and D.20 See Figure 2 for a diagram of

the AAL.

16
Ibid., p. 6.
17
Reddivalam, op. cit., p. 6.
18
Ibid., p. 6.
19
Saadawi, op. cit., pp. 394-395.
20
Ibid., pp. 394-395.

9
7.0 ATM SIGNALING

Signaling is another major function for ATM. When an endpoint device wants to establish a

connection with another endpoint device, the transmitting endpoint device sends a signaling

packet to its ATM switch. The packet contains the address of the receiving endpoint device along

with the QOS parameters. The address is translated to the proper ATM address by the ATM

layer. The signaling packet is examined by the switch and if there is a table entry for the

endpoint device and the QOS parameters can be met, the switch establishes a VC on the input

link and forwards the request to the interface for the endpoint device as specified in the table.

The request may be sent through several ATM nodes or switches prior to reaching the endpoint

device. Each node or switch in the path examines the signaling packet and routes it to the next

node or switch if the QOS parameters can be met.

The VC is being built as the signaling packet is forwarded. If any node or switch cannot

meet the QOS parameters the request is rejected and a rejection message is returned to the

originator. This includes the endpoint at the destination. When the signaling packet arrives at the

endpoint and if the QOS parameters can be met the endpoint device responds with an accept

message. That message traverses back to the originator via the VC that has just been

established. The originator of the request receives the accept message from its directly

connected ATM switch along with the VCI and VPI values the originator should use.21 The

routing table entries at each node are written at the connection establishment phase for each

connection.22

This is an example of what makes ATM connection-oriented. A connection must be

established before any user data is transmitted. This not only establishes the link, it also helps

control congestion. This connection sequence prevents user data from entering the network

before there is a path to its intended destination. This helps to ensure there is a physical

connection available, that the QOS parameters can be met, that no errors have occurred in the

transmission of the cell headers during setup, and that the cells will arrive at their proper

21
“Asynchronous Transfer Mode,” op. cit., p. 13.
22
Chandran, op. cit., p. 2.

10
destinations. This aids in controlling potential network congestion by not sending cells that

cannot arrive at their proper destinations or meet transmission requirements.

When the connection is established the cells can begin to flow toward their destinations.

Since ATM uses VC’s and VP’s all cells associated with the signaling packet follow the same

path as that packet.23 Cells are relayed at intermediate nodes or switches in the network by

forwarding cells from one ATM entity to another. Cells can be relayed from one VP to another or

one VC to another, either in the same or different VP. This switch from an incoming link to an

outgoing link is done by first reading the incoming VPI and VCI fields. The second step is to

perform a table lookup to find the correct outgoing link and determine the new VPI’s and VCI’s.

Finally, the cell is delivered to the corresponding outgoing link with the new header information.

The cell is received at the incoming port at the next node or switch and the process is repeated

until the cell arrives at its intended endpoint.

Connection release is similar in that a disconnect packet is sent when a user disconnects.

As the disconnect packet traverses the network the VC’s and VP’s used in building the

connection are released.

8.0 ATM SWITCH

One of the main functions of the ATM switch is to receive cells on a port and switch those

cells to the proper output port.24 The ATM switch uses the VPI and VCI fields of the cell header

to identify the next network segment the cell needs to access on its way to its final destination.

The ATM switch is a composite of a VP switch and a VP/VC switch. In a VP switch the VCI’s

bound for the same intermediate destination are multiplexed into VP’s. The ATM switch can

perform an operation on a single VP and affect numerous VC’s. For instance, many VC’s bound

for the same intermediate node or switch can be placed in the same VP. When the VP is

switched all of the VC’s are switched simultaneously using only the VPI. The VCI’s pass through

the VP switch unchanged and have no significance in routing the cells. Since only the VP was

23
“Routing,” (1996): p. 1, online, Internet, April 3, 1997.
24
“Asynchronous Transfer Mode,” op. cit., p. 9.

11
switched only the VPI is changed. Since the VCI’s remain the same the number of fields that

must be changed as the cells pass through the switch is reduced. Switching at the VP level and

only changing the VPI’s reduces the amount of processing time required by the switch and

improves its performance.25 An example is provided in Section 9.0 Routing.

The VP/VC switch switches cells using the VCI’s as well as the VPI’s. In the VP/VC switch

the VP’s are classified as terminating or not terminating in this switch. The switch checks the VPI

to determine whether this VP terminates at this node or not. If not, this is a VP switch and the

VPI is changed to one used as an output link and the cell is switched to that port. If the VP

terminates at this switch it is a VP/VC switch and the VCI is used to determine the new VPI and

VCI to be assigned for the output link. The values of both are changed in the latter case and the

cell is switched to its output port. An example is provided in Section 9.0 Routing.

Another major function of the switch is congestion control. The switch may police traffic to

determine the amount of congestion. This involves measuring the traffic flow and comparing it to

the agreed upon QOS. Since the number of cells arriving at a switch bound for the same

destination may exceed the bandwidth available at the output port, the switch may queue cells

until this condition, known as contention, subsides. If the contention lasts too long the buffer for

the queue may not be large enough and some cells may be discarded, depending on the state of

the CLP bit. If the cells are discard eligible and the congestion rate is exceeded any ATM switch

handling the cells can discard them.

Another method for congestion control is to ensure that cells are destined for the desired

endpoint. This concern for congestion is one reason that the cell header is checked for accuracy

as the cell makes its way through the network. Congestion control is a major concern for ATM

switch designers. For instance, if one cell of a Fiber Distributed Data Interface (FDDI) frame is
26
routed to the incorrect destination and discarded, 93 cells must be retransmitted. This leads to

an exponential increase in congestion.

25
Ibid., p. 11.
26
Ibid., p. 13.

12
9.0 ATM ROUTING

Routing in ATM networks is performed by using routing tables for VC’s and VP’s. VC links

are defined by the routing table entries of two nodes or switches connected by point-to-point

physical links. To establish a connection between two ATM entities the VCI’s and VPI’s to be

used by the transmitter and receiver are assigned at connection setup. The routing tables of the

intermediate nodes or switches along the path are updated during the connection setup as

explained earlier.

VP’s are semi-permanent connections and the routing tables for the VP switch are preset by

network management functions. Each VP has a defined bandwidth, however, and that limits the

number of VC’s that can be multiplexed.27 VPI’s are usually used to route cells between two

nodes that originate, delete, or terminate VP’s. VCI’s are usually used at the end nodes to

distinguish between connections. A VC and a VP would be identical in a single hop.28

Table 1 and Figure 3 help to illustrate how the routing table, VP, and VP/VC switch interact.

A table in the switch maps input ports to output ports based on the VPI and VCI fields in the cell

header. As cells are routed to the output ports the VPI and possibly the VCI values are changed.

For example, according to Table 1 cells that enter the switch on input port 1, VPI 4, VCI 7 are

processed through the VC switch and will be transmitted via output port 2, VPI 6, VCI 8. Both the

VCI and VPI are changed. Cells that enter the switch on input port 1, VPI 5, VCI 9 are processed

through the VP switch. According to the table, those cells would be output on port 3, VPI 10, VCI

9. Only the VPI is changed.

One of the major functions of routing in an ATM network is to reroute VC’s and VP’s to

account for changes in network operating conditions such as congestion and link or node

failures. Rerouting may be required to meet QOS parameters. To maintain routing in ATM

networks, three attributes are desired:

1) Quick changes in the bandwidth allocated for a VC or VP.

2) Quick detection of failures and fast rerouting of VP’s with failed components.

27
Chandran, op. cit., p. 4.
28
Ibid., p. 5.

13
3) Flexible means of adding and deleting VP’s to adjust to variability of traffic at different times.

Rerouting decisions can be made at any node or switch and distributed as necessary to other

nodes or switches via network control overhead.29

One example of those attributes is the manner in which network failures are handled.

Although there are several methods for rerouting during failures they all involve the use of

backup VP’s. In the event of a failure at or between nodes a VP alarm indication signal (VP-AIS)

is sent downstream to the destination endpoint of the VP as notification of the failure. If the

failure affects only the originating-terminating direction of transmission the terminating endpoint

sends a VP far-end receive failure (VP-FERF) to the originating endpoint via the same VP. The

originating endpoint is then advised that there is a failure and cells must be rerouted to the

backup VP.

If, however, the failure is bidirectional sending the VP-FERF on the same VP is to no avail.

Instead a backup VP must be activated and the VP-FERF must be sent on the backup. In either

the unidirectional or bidirectional failure the backup VP can only be activated if it meets QOS

parameters for the traffic.

When traffic is rerouted one potential problem is that cells may not arrive in the proper

sequence. Rerouted cells could reach the VP destination ahead of the cells that were on the

working VP prior to the failure. One solution is to share the last link from the ATM node or switch

to the destination endpoint between the working and backup VP’s and reroute both directions of

the VP at the endpoints. This ensures that no multipath conditions exist and cell sequence is

maintained.30

This routing scheme using backup VP’s updates the routing tables at the VP endpoints when

the alarm cells are terminated. For the link on which the alarm cell was received all

corresponding VPI entries are replaced with the VPI of the backup VP.31 The VCI’s remain the

same. This allows faster processing and restoral since the VCI’s are not changed.

29
“Asynchronous Transfer Mode,” op. cit., p. 7.
30
Ender Ayanoglu and Richard Gitlin, “Broadband Network Restoration,” IEEE
Communications Magazine, Volume 34, Number 7 (July, 1996), 101.
31
Ibid., p. 106.

14
 Similar routing changes occur when nodes or switches encounter congestion. Routing tables

enable nodes and switches to reroute cells dynamically by using alternate routes.

ATM is capable of transporting multiple types of services simultaneously on the same

network. All data is placed in cells of uniform size. The BISDN PRM adapts the various

applications to the ATM cells and prepares the cells for transport over different types of physical

media. The cell header contains information concerning cell routing using VCI’s and VPI’s. Cells

from various applications with the same destination can be interleaved to share physical

facilities. This allows network providers to transport different types of services using the same

physical facilities. This is an advantage for network providers in that facilities can be fully

utilized. It is an advantage for end users since they can connect their various networks and only

pay for the data they are sending.

QOS parameters are specified when the service application is made. These parameters

determine the level of quality the user expects for his application and includes the bandwidth

required for the amount of data and rate at which he wishes to send. To maintain the desired

QOS ATM nodes and switches constantly evaluate traffic and other network conditions. Traffic

can be rerouted if there is extreme congestion or a node, switch, or facility failure in the network.

To aid in fast transport of the cells ATM uses VC’s and VP’s. VC’s bound for the same

destination are multiplexed into the same VP. By switching a VP the entire group of VC’s is

switched. This helps to reduce processing time and increases the speed of the switch.

Each ATM node or switch uses the VCI’s and VPI’s to determine the next node or switch the

cell needs to access. The end-to-end VC is built as the connection is setup from node-to-node.

The setup cell must reach the endpoint destination and be acknowledged before user data can

be sent. This aids in congestion control.

Congestion control is a major topic in ATM. The network monitors itself constantly for

failures and congestion. If either is encountered the cells are rerouted. Also, to ensure that ATM

cells arrive at their proper destinations ATM performs error checks and is connection-oriented.

These attributes ensure that cells are not sent to incorrect destinations because of transmission

15
errors that occur and that cells have physical paths to their destinations prior to being placed on

the network.

Since ATM can transport all types of services via different physical interconnections it can

be used anywhere. The network provider can utilize it for transporting numerous types of

services and utilize various physical facilities that are already in place. End users can connect

several types of services to the same network and only pay for the data they send. ATM holds

much promise for the future of all types of services, voice, video, and data.

16
 Figures and Tables

Management Plane

Control Plane User Plane

Plane Management
Higher Higher

Layer Management
Layers Layers

ATM Adaptation Layer

ATM Layer

Physical Layer 1

Figure 1
BISDN Protocol Reference Model
From: “Asynchronous Transfer Mode,” (November 14, 1995), p. 4, online, Internet, February 26,
1997.

A Convergence Sublayer
A
L Segmentation and Reassembly Sublayer

A Generic Flow Control

T Cell Header Generation/Extraction
M Cell VPI/VCI Translation
Cell Multiplexing and Demultiplexing

P Trans Cell Rate Decoupling

H mission HEC Header Sequence Generator/Verification
Y Conver Cell Delineation
S gence Transmission Frame Adaptation
I Sublayer Transmission Frame Generation/Recovery
C
A Physical Bit Timing
L Medium Physical Medium
Sublayer

Figure 2
Layers and Sublayers of the BISDN Protocol Reference Model
From: “Asynchronous Transfer Mode An Overview,” p. 7, online, Internet, February 26, 1997.

17
 Input Output

Port VPI VCI Port VPI VCI

1 4 7 2 6 8
2 6 8 1 4 7
1 5 9 3 10 9
3 10 9 1 5 9

Table 1
Routing Table for Figure 3
From: “Asynchronous Transfer Mode,” (November 14, 1995), p. 10, online, Internet, February 26,
1997.

VC Switch
VCI 7 VCI 8

Port 1 Port 2

VCI 7 VPIVPI
4 4 VPI 6 VPI 6 VCI 8

Port 3
VCI 9
VPIVPI
5 7 VPI 6 VPI 10 VCI 9

VP Switch

Figure 3
VP and VP/VC Switch
From: “Asynchronous Transfer Mode,” (November 14, 1995), p. 11, online, Internet, February 26,
1997.

18
 List of Works Consulted

“Asynchronous Transfer Mode,” (November 14, 1995) 14 pages, Internet, February 26, 1997,
Available http://www.cisco.com/univercd/data/doc/cintrnet/ito/55755.htm#HDTO.

“Asynchronous Transfer Mode An Overview,” 20 pages, Internet, February 26, 1997, Available
http://www.teleware.fi/lukusali/cw_atmov/index.html.

Ayanoglu, Ender and Gitlin, Richard D. “Broadband Network Restoration,” IEEE Communications
Magazine, Volume 34, Number 7 (July, 1996), 110-119.

Bellamy, John. Digital Telephony, Second Edition. New York: John Wiley and Sons, Inc., 1991.

Cabral, James E., Jr. and Kim, Yongmin. “Multimedia Systems for Telemedicine and Their
Communications Requirements,” IEEE Communications Magazine, Volume 34, Number 7
(July, 1996), 20-27.

“Cell Switching (ATM),” (1996), 9 pages, Internet, April 3, 1997, Available

http://literary.com/mkp/new/3689/book/node42.html.

Chandran, Prem. “Asynchronous Transfer Mode,” 8 pages, Internet, February 26, 1997,
Available http://www.ecst.csuchico.edu/~prem/ATM.html.

Clark, Douglas P. and Goodman, William D. “Signaling Challenges in Video Dial Tone
Networks,”
IEEE Communications Magazine, Volume 34, Number 8 (August, 1996), 128-133.

Fendick, Kerry W. “Evolution of Controls for the Available Bit Rate Service,” IEEE
Communications Magazine, Volume 34, Number 11 (November, 1996), 35-39.

Fujii, Nobuo and Yamamura, Tetsuya. “ATM Transport Network Operation System in Japan,”
IEEE Communications Magazine, Volume 34, Number 9 (September, 1996), 70-75.

Green, James Harry. The Irwin Handbook of Telecommunications Second Edition. Burr Ridge,
Illinois: Irwin Professional Publishing, 1992.

Karlson, Gunnar. “Asynchronous Transfer of Video,” IEEE Communications Magazine, Volume

34, Number 8 (August, 1996), 118-126.

Leonardi, Emilio; Neri, Fabio; Gerla, Mario; and Palnati, Prasasth. “Congestion Control in
Asynchronous High-Speed Wormhole Routing Networks,” IEEE Communications Magazine,
Volume 34, Number 11 (November, 1996), 58-69.

Lockwood, J. W. “iPoint: What is ATM,” (1993), 1 page, Internet, February 26, 1997, Available
http://www.ele.auckland.ac.nz/students/tankh/atm/faq/atm.html.

“Low Speed Asynchronous Transfer Mode,” 2 pages, Internet, February 26, 1997, Available
http://www.impact.nl/Whiteppr/gdc/gdcl.html.

McInerney, Laura. “A Layered Approach,” Telephony, (February 10, 1997), 18-24.

Perros, Harry G. and Elsayed, Khaled M. “Call Admission Control Schemes: A Review,” IEEE
Communications Magazine, Volume 34, Number 11 (November, 1996), 82-91.

Reddivalam, Srinivas. “Asynchronous Transfer Mode Tutorial,” 17 pages, Internet, February 20,
1997, Available http://www.cne.gmu.edu/~sreddiva/Texttut.html.

19
Rothauser, Charles H. “Asynchronous Transmission Mode,” NCSA Industrial Partner, Network
Technology Coop Program, 8 pages, Internet, February 26, 1997, Available
http://utrcwww.utc.com/UTRC/Techreports/atm.html.

“Routing,” 1 page, (1996), Internet, April 4, 1997, Available

http://literary.com/mkp/new/3689/book/node41.html.

Saadawi, Tarek N.; Ammar, Mostafa H.; and El Hakeem, Ahmed. Fundamentals of
Telecommunications Networks. New York: John Wiley and Sons, Inc., 1994.

Saito, Hiroshi; Kawashima, Konosuke; Kitazume, Hideo; Koike, Arata; Ishizuka, Mika; and Abe,
Atsushi. “Performance Issues in Public ABR Service,” IEEE Communications Magazine,
Volume 34, Number 11 (November, 1996), 40-48.

20