The Asynchronous Transfer Mode (ATM) Principle

 The Asynchronous Transfer Mode (ATM) principle has dominated the telecom network for a long time as a mainstream datacom technology. Nowadays, the IP technology prevails over the telecom network. It is time for us to unveil the ATM principle thoroughly. 

ATM (Asynchronous Transfer Mode) is an asynchronous transfer mode with cell as the unit. “Transfer mode” refers to the manner of information transmission, multiplex and switching in telecommunication network. Asynchronism means that the information cell flow from any user need not be periodical.

ATM, a kind of special packet transmission technology, provides the QOS guaranty, and meets the requirements for the real time service and non-real time service.

In 1987, ITU-T chose the asynchronous transfer mode (ATM) as the transfer mode in BISDN. The outstanding characteristic of ATM is that it can flexibly support all kinds of services that are existing and likely to exist in the future, reach the high network resource utilization ratio, and support the high-speed switching.

Differences Between ATM, Circuit Switching and Packet Switching

The circuit switching is connection-oriented. It implements switching based on the time slot, and monopolizes the bandwidth. Before making a call, dial the phone number to set up a connection of timeslot between two parties, then speak and hang up. This communication way is known as the connection-oriented switching. Moreover, a circuit exists between two parties of the communication, so it is also known as the circuit switching. In the circuit switching, the network resource may be allocated by the signaling system during setting up the connection. If the network resource is allocated to a connection, the resource cannot be occupied by any other connection, and it can be released only when the connection breaks. The PSTN that is based on the circuit switching also occupies 64k time slots at the tone silence period, and monopolizes the bandwidth.

The packet switching is connectionless-oriented. Connectionless means that no exclusive and independent link exists between two parties of the communication. The packet switching uses packet-switching technology. This technology binds the data to be transmitted into the packets that are actually bytes with special labels. Each packet has its destination address, source address, error control message, and so on. Such packets can search the destination itself. In the non-exclusive transmission, many applications can share a narrow bandwidth, and the utilization of the bandwidth is high.

ATM integrates the advantages of the circuit switching and the packet switching, and it is a connection-oriented packet switching.

Connectionless Network and Connection-Oriented ATM Network

The connectionless-oriented packet switching encapsulates users' information into packets for switching in the packet-switching mode. Each packet has a header that is used for selecting routes and controlling errors and flow. Switching devices check the address carried by each header, and choose a route based on the current network status to send the packets to the lower-level devices. Therefore, different packets of the same service follow different paths. The length and time interval of each packet can be changed. Hence, the packet switching can provide multi-rate switching. The packet switching network cannot guarantee the reliable QoS by such a segment-by-segment forwarding.

ATM is connection-oriented. On the one hand, each ATM terminal user needs to set up the connection when he communicates with another user. On the other hand, the ATM transmission uses the cells that have a fixed length of 53 bytes, so it has the features of the packet switching that enables each connection to share the bandwidth. ATM can carry multiple information media in a single backbone network, bear multiple communication services, and guarantee QoS.

                

Connectionless Network

 Connection-oriented ATM network

Organizations for Establishing the ATM Standards

ITU-T

Absorbs itself in standardizing ATM on the public network. Its representative standards include the Q.2931/Q.2971, BISUP, and I.610.

ATM Forum

Absorbs itself in standardizing ATM on the private network and promoting the ATM products and usage. Its representative standards include the UNI3.1, UNI4.0, ILMI, PNNI, MPOA, VTOA, LANE, and TM4.0.

IETF

Absorbs itself in standardizing the IP over ATM and solving the bottleneck of traditional routers.  Its representative standards include the RFC1483, RFC1577, IP Switch, and MPLS

ATM Cell Structure

ATM is a switching and multiplexing technology based on cells. The ATM cells are the basic carriers for ATM to transmit information, and the cells resemble the packets in the packet switching but have their own features. An ATM cell is of fixed length of 53 bytes that is small. Such a cell consists of two parts: the header and the payload. The header has 5 bytes and the payload has 48 bytes. 

GFC (generic flow control): It is of 4 bits and only used on the UNI interface. The four bits by far are configured as 0000; however, GFC will be probably used to control the flow or identify the access mode on the network where media are shared in the future.

VPI (virtual path identifier): It is of 12 bits on the NNI and of 8 bits on UNI.

VCI (virtual channel identifier): It is of 16 bits to identify the virtual channel in the virtual path. And VPI/VCI together identifies a virtual connection.

PT (payload type): It is of 3 bits to indicate whether the payload is the data cell or the management cell.

CLP (cell loss priority): It is of 1 byte to be used to control the congestion.

The cell has two priority levels when transmitted on the ATM network. Their priority level is identified by the CLP in the cell header. If the CLP equals 0, it indicates the higher priority level; if the CLP equals 1, it indicates the lower priority level. When the network is congested, the cells with the lower priority are discarded first to alleviate the congestion, and to guarantee the transmission quality of the cells with higher priority.

HEC (header error control): It is of 8 bits to detect error headers. It can correct 1-bit error in the header. HEC can delimit the cell by using the HEC field and the relativity of the 4 bytes before HEC to locate the header. Because the VPI/VCI has different values in different links, HEC must be recalculated in each link.

Compared with the packet header in the packet-switching, the ATM cell header has simplified functions. the functions of the ATM cell header are greatly simplified. For example, it does not detect and correct errors of links section by section. Due to the link quality improvement, the end-to-end error control is only processed by the terminal if it is necessary. HEC only takes charge of controlling the error in the cell header. In addition, the header only uses VPI/VCI to identify a connection and does not need the source address, destination address, and the serial number of packets. The cell sequence is guaranteed by each Network Element (NE).

ATM Cell Structure

ATM Connection: connection-oriented switching

ATM is the connection-oriented switching. Such connections are identified by VPI and VCI. The VPI and VCI values are locally valid. That is, VPI and VCI are only valid between two interfaces that are directly connected through physical media. The same value can be reused in other interfaces. Each VPI/VCI is processed at the corresponding VP/VC switching node. The same VPI/VCI value at the different VP/VC links does not represent the same virtual connection.

Viewed from routes, VPI and VCI are the routing addresses along which the cell is transmitted on the ATM network. Multiple routing addresses identify a connection. When receiving a cell, the switching network checks the mapping table based on VPI and VCI carried by the cell header and defines the output VPI and VCI. As shown in the figure, the cell VPI/VCI=1/40 sent by User A is switched to VPI/VCI=2/44 after passing Switch B, to VPI/VCI=3/44 after passing Switch C, and to VPI/VCI=4/50 after passing Switch D. It at last is sent to User E. Here, (1, 40) (2, 44) (3, 44) (4, 50) identify a connection between A and E.

The VPI and the  VCI are only locally valid.

VP and VC

The most important parts in the header are VPI and VCI, namely, virtual path identifier and virtual channel identifier.

The ATM network operation resembles the connecting calls. Before communication, the connection must be set up between the source end and the destination end. This is a virtual connection, that is, the VCC that can be identified by VPI/VCI.

The most important features of ATM are the processes of multiplexing, switching and transmitting cells. Such processes are carried out in VCs that are identified by VCI. VC is a kind of logic connection between the two ends of a link on the ATM network, and it is the communication channel for transmitting the ATM cells between two ends or more. VC can be used to transmit information from a user to another user, from a user to the network, and from a network to anther network.

Virtual paths (VP) are a group of VCs that share the same VPI on a given reference point. In transmission, VCs combine together to form a VP. Therefore, the cells of different users on the ATM network are transmitted in different VPs and VCs. And different VPs/VCs are identified by VPI and VCI.

The virtual identifier VPI/VCI

VP Switching and VC Switching

The ATM switching is classified into the VP switching and VC switching.

In the VP switching, only the VPI value is changed, and the VCI value is transparently transmitted. Therefore, in application, VP serves as a big pipe and VC as a small pipe.

              

                    VP Switching and VC Switching

ATM  Connection Modes

·      PVC (permanent virtual channel): The NMS sets up PVCs. Whether any service passes or any terminal device is accessed, the PVCs are always enabled till they are released by NMS. PVCs are just like lease lines in the telephony network.

·  

       SVC (switching virtual channel)：SVCs are  set up and released by signaling. They are just like subscriber lines in the telephony network

Note: artikel ini sengaja disajikan dalam bahasa Inggris atas permintaan salah seorang pembaca

·