Asynchronous transfer mode (ATM)

ATM is a member of the fast packet−switching family called cell relay. As part of its heritage, it is an evolution from many other sets of protocols. In fact, ATM is a statistical time−division multiplexed (TDMed) form of traffic that is designed to carry any form of traffic and enables the traffic to be delivered asynchronously to the network.


How an ATM sends data?

Speed comparison between ATM over other technology and protocols.

                With this information we can derive is that the ATM technique is a combination of TDM, with cells using pre-assigned slots, and Statistical TDM, with cells using whatever slots are available or needed to handle a particular traffic flow. It is also a connection−oriented protocol much the same as dialup voice communications services, but it uses virtual circuits, such as permanent virtual circuits (PVCs) and switched virtual circuits (SVC), to handle the connection.

Advantage of ATM over other technology and protocols

                ATM was designed from the ground up to work across the various places where we communicate: the Local Area Network (LAN), the Campus Area Network (CAN), and the Metropolitan

Area Network (MAN), also the Wide Area Network (WAN).

Speed comparison between ATM over other technology and protocols.

ATM features and functions

ATM Protocols

It takes many protocols to support an ATM network, which is one of the issues that continually Comes up as a negative from the supporters of the gigabit Ethernet crowd. The actual protocols needed depend on where the traffic originates, what transport mechanisms must be traversed, and where the traffic will terminate.

To understand more about this protocol here is an image that will show you the graphic representation of the ATM protocol interfaces and a table that shows where protocols are used for ATM

The problem with ATM is that in order to support the older legacy systems, many protocol points and interfaces are necessary. To get around the problem of "forklift" changes, the necessary protocols have been developed.

Mapping Circuits through an ATM Network

ATM uses one of two connection types, namely PVC and SVC. There is actually no permanency to the circuits. They are logically mapped through the network and are used when Needed for PVC or dial−connected when using the SVC, the carriers promise only to make a best attempt to serve the needs of the end user when the time is appropriate.

The concept is that the network provider will provide a committed bandwidth available to the user on demand whenever the user wants to use it. This forms the basis of what ATM networks are all about: on−demand, high−speed communications networks.

The connection is built into a routing table in each of the switches involved with the connection from end to end. As such, the switches only need to look up a table for the incoming port and channel and then determine the mapping (in the same table) for the output port and channel. Using virtual path identifiers (VPI) and virtual channel identifiers (VCI), the carrier maps the table.

This figure shows a full virtual connection is mapped through the various switches across the network. Here the end−user device is connected across an ATM access link through a switch. The switches provide the cross connect ion and link to the next downstream node. Note that the connection from the end user to the network may be on a T1, T3, or OC−n. From the first switch out, the network will use Synchronous Optical Network (SONET) or synchronous Digital Hierarchy (SDH) capabilities possibly mapped onto a Dense Wave Division Multiplexer (DWDM). The network carrier will use whatever services and bandwidth is available at the connection points.

The network switches handle the mapping on the basis of VPI switching.  VPI switching means that the switches use the virtual path for mapping through the network and will remap from one virtual path to another, while the virtual channel number is held consistent through the entire network.

A second alternative is to use VPI/VCI Switching, In this case, the ATM switches along the route will switch and remap both on a virtual path and a virtual channel, the virtual path and virtual channel switches process and remap both elements.

The OSI and ATM Layered Architecture

This is the upper-layer services of ATM.

The types of ALL and services offered.

ATM Traffic Management

                Some requirements are needed for the ATM to manage the traffic.

• ATM must be flexible. It must meet the constantly changing demands of the user population. These goals mean that the demands for traffic will rise or fall as necessary, and therefore managing this traffic is of paramount importance.
• ATM must meet the diverse needs of the end−user population. Many users will have varying demands for both high− and low−speed traffic across the network. Using a QoS capability throughout the ATM network, a user can determine the performance and the capabilities of how the ATM network will meet their demands.
• Cost efficiency is a must, If ATM is truly to succeed, and traffic management must also include the effective usage of all of the circuitry available. ATM is designed to reduce the inefficient circuit usage by efficiently mapping cells into dead spaces, particularly when data is involved.
• Robustness in the event of failure or in the event of excess demand is a requirement of the traffic management goals. s. If the network is to be readily available for all users to be able to transmit information on demand, then the network must be very robust to accommodate failures,   link downtime, and so on.

Traffic in the form of asynchronous bursts of information (cells) enters the network at random times. This randomness is what causes the confusion and the unpredictability of the data. To manage the traffic flow, buffers are used to enable the flow and ebb of traffic volumes. Because data tends to be very bursty, it is extremely difficult to predict the demands of the network and the capacity needed at a given time.

The use of "leaky buckets" in the buffering of the traffic helps to manage and control the flow of           traffic onto and through the network. The leaky bucket, as the name implies, is a buffer that is constantly flowing.

Traffic enters into the buffers and is tagged, based on the amount of cells enabled by the carrier. If the user exceeds the amount of cell flow per increment (per second, and so on), then the buffer is filled and begins to empty out the bottom side. If more cells enter the buffer than are allowed, the cells are flagged for discard.

Shaping the Traffic

The ITU−TSS defines four possible situations when a cell enters an ATM network:

• Successfully delivered the cell arrives at the destination with less than time T−cell delay.
• An error cell occurs a cell arrives with at least one detected bit error in the information field in the cell. Another possibility is the severely error cell with information bits errors equal to n or n>1
• Lost cells a cell either never arrives or arrives after the time T−cell delay, in which case it is discarded at the destination.
•       An inserted cell a cell contains an undetected error or is misdirected by an ATM node and therefore shows up at the wrong destination.

This figure shows that if the user exceeds the network rate, the cells will be discarded, and the user will, therefore, have to retransmit at a later time.

And if the user does not exceeds the network rate the network should deliver all available cells delivered to the network. As shown in the figure bellow.

Source

Broadband telecommunication handbook 2nd edition by Regis J. "Bud" Bates