next up back


A. Introduction
 
 

A.1. What is ATM?

ATM network. What's that? That's the first question came to my head when the first time I heard about it. ATM here is not Auto Teller Machines, instead it stands for Asynchronous Transfer Mode.

The term Asynchronous Transfer Mode (ATM) strictly speaking refers to "a generic mode of data transfer in which units of data being transmitted are not time related to each other." I'm not going to explain in very detail about ATM and how it works in this paper. It takes pages to explain all the procedures, the  layers  and the products of ATM.  Just to brief you through on what is ATM, it is just like a method of delivery using US mail, where one can choose to send 1st class, overnight, 2 day delivery, etc. and can ask for certified mail.  When information needs to be communicated, the sender NEGOTIATES a "requested path" with the network for a connection to the destination. When setting up this connection, the sender specifies the type, speed and other attributes of the call, which determine the end-to-end quality of service.

Using ATM, information to be sent is segmented into fixed length cell, transported to and re-assembled at the destination. The
ATM cell has a fixed length of 53 bytes.The cell is broken into two main sections, the header and the payload. The payload (48 bytes) is the portion which carries the actual information-either voice, data, or video. The Header (5 bytes) is the addressing mechanism.

Maybe now you have a clearer idea on what is ATM, but what are the reasons why people came out with the idea of ATM network. Check this out. 
 

A.2. How ATM exists?

There are many reasons why people came out with the idea of ATM network.

All the reasons stated above have forced  ATM to rise. ATM was originally proposed as the switching structure of B-ISDN (Broadband-Integrated Service Digital Network) but has taken on its own role outside of B-ISDN.

                                               
B. Advantages Of Asynchronous Transfer Mode
    (ATM)
 
 
ATM was first proposed by Bellcore, the research facility for AT&T in the U.S., and a few other telecommunications
companies in Europe. It was originally described as "fast packet switching with short fixed length packets." It has a fixed size of packets to ensure the same transmitted voice quality as in Synchronous Transfer Mode (STM) networks. STM is used to transfer packetized voice and data across long distances by today's telecommunications backbone networks. It uses circuit switching. Before any data is transferred, a connection is established between two end points, and then is removed after communication stops. The end points are actually connected for the entire time even if no data is being sent or received. Basically, data is transferred across STM networks by dividing the bandwidth of the STM links (T1 and T3 links) into units of
transmission called time-slots or buckets.

On the other hand, in ATM networks, two end points are connected together via an identifier called the "Virtual Circuit
Identifier" (VCI). The header of the fast packet transports the VCI. The fast packet itself is still carried in a similar bucket as
before, except it does not have a label or designation for the bucket. VCI labels are controlled by network nodes, and are
quite random as connections are established and eliminated. They are defined on a link-by-link basis between terminal and
switching systems and between switching systems which sometimes change the values of VCIs. The address part of the VCI is
never more than 24 bits.

In looking more closely at an ATM cell or packet as specified by the U.S. T1S1 subcommittee (ANSI sponsored), it is
composed of 53 bytes. Five bytes make up the header, and 48 are the payload. Out of those 5 header bytes, 24 bits are for
the VCI label, 8 are for control, and 8 are for header checksum. The 48 bytes of payload may contain a 4 byte ATM
adaptation layer (AAL) with 44 bytes of real data; this all depends on a bit in the control field of the header. Fragmentation and
reassembly of cells into larger packets at the source and then the destination are affected by this optional adaptation layer. The
ATM packet specified by the European ETSI committee is also composed of 53 bytes. Yet, within the 5 byte header, there is
a difference in the number of bits of each of its subcategories. Nevertheless, ATM transmission of these 53 bytes will be
regulated   throughout the world.
 
            <---------------- 5 bytes -------------------->|<---------- 48 bytes--- --------->|
            ----------------------------------------------------------------------------------------
            | VCI Label  | control  | header checksum  | optional adaptation  | payload |
            |  24 bits      |  8 bits    |          8 bits          |      layer 4 bytes      |44 or 48 |
            -----------------------------------------------------------------------------------------

With ATM being the international standard for cell relay, it combines the efficiency of packet switching and the reliability of
circuit switching. This is done by dividing all of the information into very small cell units and transferring and switching these units
at super-high speed. Most networks are either circuit-oriented, for transmitting delay-sensitive (isochronous) information such
as video or voice, or packet-oriented, for high-speed data delivery. Circuit switching guarantees end-to-end delivery and
response times but can waste expensive bandwidth. Packet switching makes the most out of its use of bandwidth but has variable packet delivery times, therefore, making it unsuitable for isochronous transmission. In analyzing different designs for advanced packet switch fabrics, IBM has come up with one named Switch-on-a-Chip. It is based on "a single chip switch element from which larger, self-routing single-stage or multistage switch fabrics can be constructed. Since the switch design can be scaled, multiple chips can be connected in series or in parallel to create large switch fabrics with higher performance. None of the packets are discarded this way" (IBM). Most of the new advanced packet switches are based on hardware as opposed to their earlier software-based counterparts. Another way IBM is trying to invoke user interest is in focusing on existing equipment and existing voice-grade, unshielded twisted-pair (UTP-3) wiring for ATM transmission. IBM will simply be reusing what is already there.

ATM transmission connections for users are based on well-defined and well-controlled interfaces called "User Network
Interface" (UNI). At the connection setup time, user specifications dictate the requirements for sending and receiving data via
the network. The system will try to stay within these requirements for the duration of the transmission. There may be several
kinds of network interfaces in the future. The ATM Forum is the most responsible for defining interfaces required to operate
and manage ATM networks.

Through an ATM Forum technical committee, General DataComm, has come up with an ATM frame user-network interface
(FUNI) specification. It will "provide ATM User-Network Interface functionality for data transmission at speeds from 56/64
Kb/s to T1/E1 (1.5/2.0 Mb/s); interoperate end-to-end with cell-based ATM UNIs, including signaling, traffic management,
network management, and OAM functionality; deliver efficient utilization of bandwidth at narrowband speeds; and allow low
speed ATM traffic to be carried as part of higher bandwidth physical connections using standard time-division multiplexing"
("Low Speed ATM"). FUNI frame headers also provide congestion notification and cell loss priority flags that have similarities
to flags in UNI cell headers.

Statistical multiplexing is another advantage of ATM. If a good number of connections are quite bursty, then all of them could be assigned to the same link since statistically they might not all burst at the same time. Yet, if some of them do burst
simultaneously, the burst will be cushioned by the available elasticity until it is put in the next free buckets.

Therefore, ATM networks rely heavily on user-supplied information at connection setup time in order to provide the desired
connection for transmission. They can secure a fixed bandwidth for an isochronous (repeating in time such as voice samples)
connection carrying a continuous bit stream and a plesiochronous (variable frequency such as compressed video) connection for a variable bit stream as well as rely on statistical sharing with no specific bandwidth for bursty sources. In ATM networks, the
transmitter and receiver performance are also independent of one another. The transmitter side is constrained by flow control of
the simultaneous connection streams with respect to bandwidth and other user requirements. The receiver side is constrained by asynchronous reception of cells at a variable rate and whether or not the adaptation layer is used. If the AAL layer is used, the
reassembly of these cells into a higher layer protocol data unit (PDU) would also be done in the hardware of the receiver side.

The flow control of ATM packets in internetworking requires different control mechanisms than that of TCP because of the
gigabit/sec capacity of ATM networks. These networks need to be able to relax themselves to a steady state with the help of
fast hardware devices because of extremely dynamic congestive situations. For example, the network state may be
communicated to the UNI by the network very quickly by generating a flow control cell whenever a cell is going to be dropped
on some node because of congestion. It would be better to avoid congestion altogether by detecting its build up early with
internal queue monitoring schemes inside ATM switches and gradually reacting to the situation. No matter how extreme the
congestion may become, one good thing is that cells with the same VCI label will arrive in order at the destination. This is
attributed to a few reasons: there is no store and forwarding--packets travel over a single virtual circuit path, cells with the same
VCI are not switched out of order by ATM switches, and there are no retransmissions at any point in the network.

In trying to determine which protocol layer best fits ATM, it seems that it does not fit nicely into the abstract layered model. This is attributed to the fact that within the ATM network itself, end-to-end connection, flow control, and routing are all done at the
ATM cell level. However, ATM is ideal for switching fixed length packets in hardware over long distances in the Gigabit/sec
range. Thus, it should lie somewhere around the data link layer as in the OSI model. There are only a few higher layer functions
present in ATM such as connectionless services. Another higher layer service necessary for guaranteed delivery would be a
TCP-type transport layer protocol. At the ATM layer, there is no end-to-end reliable delivery service--there are no
retransmissions or acknowledgments of received data. If an error is detected, the cell is simply discarded. This allows very high
transmission rates with little degradation, a positive result of recent optical fiber technology.

One of the important new technologies ATM networks will provide is Broadband ISDN, delivering faster and more varied
kinds of communication services. The present ISDN networks are referred to as Narrowband ISDN (NISDN), where
transmission speeds are based on the basic rate of 64Kb/s, ranging up to 1536/1920Kb/s. For future BISDN transmissions,
transfer speeds will reach up to 155.52Mb/s or 622.08Mb/s, capable of handling potential developments in future
communication networks. BISDN will be available in two broad categories, interactive and distribution. With interactive
services, there will be two-directional communications between terminals. One example is simultaneous communications, as in
telephones, monitoring by remote cameras, or TV conferences. Another example includes non simultaneous, as in
store-and-forward message transferring services or database retrieval services. Distribution services, on the other hand, refer to
TV broadcasting or other one-way services. Some of the characteristics of BISDN will include transmissions that can be
delivered using either a Constant Bit Rate (CBR) or a Variable Bit Rate (VBR). CBR services are provided at a fixed
transmission speed. VBR services will provide transmission as in the cases of video conferencing, where the information to be
transmitted is not constant (it depends on the movement within the picture).

Through ATM technology, BISDN will be able to offer a variety of communication speeds and constant or variable bit rates for a variety of media. It will allow the terminal user to request a certain kind of service quality for communication without making
any fixed assumptions. With the right kind of communications quality for every application, bit error requirements, will differ for
low-speed and high-speed applications. However, many problems still need to be addressed before the introduction of actual
BISDN systems, as ATM technology and its many advantages become the perfect mode of transmitting the mixture of voice,
data, and video information.
 

C. Conclusion

The advantages of ATM technology have contributed a lot to the progress of transfering data over networks. The need for faster communication forces the change in technology, and the need to simplify the interconnecting of networks, then ATM came out as the solution.  Nowadays, the information systems and telecommunications industries are focusing and standardizing on ATM It has been designed to be scalable and flexible in geographic distance, number of users access and trunk bandwidths (As of today, the speeds range from Megabits to Gigabits). This flexibility and scalability assures that ATM will be around for a long time.