. 89
( 132 .)


come to fruition, but the concept is now being implemented through the use of IP.
As mentioned, the size of the ATM cell is small and ¬xed because one design consider-
ation was to make it suitable for transmission of voice traf¬c, which is extremely sensitive
to timing and delays. There is a tradeoff when considering a small packet. On the one
hand, the smaller the cell, the lower the number of voice samples that can be sent in the
cell, and hence the shorter the delay between speaking and hearing. On the other hand,
as has been noted, the smaller the cell size, the more inef¬cient the transfer, since more
header overhead is introduced. When a standard cell size was being decided by the ITU-T,
it had to consider the views of two lobbies. The phone companies in the US wanted a cell
size of 64 bytes, since their large area of operation meant that they had already installed
much echo cancellation equipment to overcome delays. However, the European phone
companies had not, so wanted a 32-byte cell. The compromise decided was 48 bytes.

7.3.1 Virtual circuits and virtual paths
ATM implements a virtual circuit type packet switched network. The basic unit of an
ATM system is the virtual circuit, or virtual channel (VC). This is a connection from one

Virtual Path

Virtual Circuits

Figure 7.4 Virtual circuits and virtual paths

source to one destination. However, there is a provision in ATM allowing for multicasting.
Virtual circuits are considered to carry information in one direction only, but for duplex
operation two circuits are established at the same time. These two circuits are addressed
as one, but the QoS properties of each, most commonly the data rate, may be different.
Between a source and a destination, a group of virtual circuits can be grouped together
into a virtual path (VP). A virtual path provides the advantage that if a re-routing of circuits
is required, then a re-routing of the virtual path automatically and transparently re-routes
all the virtual circuits which it encapsulates. The concept is outlined in Figure 7.4.
The virtual circuits and paths can be one of three different types: permanent virtual
circuits (PVC), soft PVC or switched virtual circuits (SVC). A PVC is established in
advance, either by the network administrator or by arrangement with a carrier. Both the
end points of the connection and the route through the network are prede¬ned. This
presents the problem that intermediate device failure results in failure of the entire PVC,
unless the underlying infrastructure (e.g. SDH) can re-route below the ATM layer. It
is similar to a leased line, and requires no setup phase. A soft PVC also prede¬nes
the end points of the connection; however, the route is not ¬xed and can be altered to
deal with failure. In the early days of ATM, practical implementation of a soft PVC
required establishing a network with equipment from a single vendor since the re-routing
was proprietary. However, the ATM Forum has de¬ned a routing protocol, the private
network-to-network interface (PNNI), to deal with this in an open standard. PNNI is
discussed in more detail later. An SVC is established when required, immediately prior
to data transfer, and is set up by signalling. If there is failure somewhere in the network,
then the SVC is broken and must be re-established.

To explore the application of ATM to a UMTS network, an understanding of the structure
of the ATM protocol is required. In this section, the ATM reference model is discussed.
Like most protocols, ATM can be split into a layered model. Each layer performs particular
functions, but is self-contained, communicating with layers above and below through
primitives. The point at which layers communicate and exchange primitives is referred to
as a service access point (SAP). The block of data exchanged across a SAP, the contents
of which are not altered, is known as a service data unit (SDU). This distinguishes it from
a protocol data unit (PDU), which includes all the header and/or trailer information that
may be added at that layer, i.e. the data plus protocol control information for the layer.
A general model of this is shown in Figure 7.5.

Layer x
PDU for layer x
SDU sent to layer y
header data trailer

Service Access Point

Layer y

Figure 7.5 Layered model

Management Plane

Control Plane User Plane

Upper Layers Upper Layers

ATM Adaptation Layer
Segmentation and
Reassembly Sublayer

ATM Layer

Transmission Convergence
Physical Layer
Physical Medium Dependent

Figure 7.6 The ATM reference model

The reference model for ATM consists of three layers, plus the user layers operating on
top of these. The three layers are the physical, ATM and AAL, as shown in Figure 7.6.
The ATM reference model is a three-dimensional one, with control, user and man-
agement planes. The role of each can be summarized thus. The user plane handles data
transport, ¬‚ow control, error correction and other user functions. The control plane handles

Layer Function

Higher Layers Higher Layer Functions

Service specific (SSCS)
sublayer (CS)
Common part (CPCS)

Layer Management
SAR sublayer Segmentation & reassembly

Generic flow control
Cell header generation
Cell header extraction
Cell VCI/VPI translation

Transmission Cell delineation
Convergence Transmission frame
sublayer (TC) generation & recovery
Physical medium Bit timing
dependent (PMD) physical medium

Figure 7.7 ATM layer functions

connection management; for example, this is where the signalling protocols operate. The
management plane handles resource management and interlayer coordination. The key
operation and function of each of the layers is presented in Figure 7.7.
Note that although there are only three layers, both the AAL and PHY layers are further
split into sublayers. The key roles of each is now explained in further detail, commencing
at the physical medium and working up through the protocol stack.

The physical layer deals with interactions with the physical medium. However, ATM
is designed to be independent of transmission medium and ¬‚exible in its use of the
underlying infrastructure. ATM cells can travel by themselves on the medium, which
is utilized for lower data rates on local connections, but more commonly the cells are
packaged inside other carrier systems, for example ATM over plesiochronous digital
hierarchy (PDH).
To deal with this, the physical layer is further divided into two sublayers. The lower
sublayer is the physical medium dependent (PMD) sublayer, which interfaces to the phys-
ical medium. It is concerned with moving bits on and off the cable or physical layer
protocol and handling timing. A different sublayer is used for different media or carriers.
The upper sublayer is the transmission convergence (TC) sublayer. It is responsible for
passing the cells to the PMD as a bitstream, and also for splitting an incoming bitstream
up into cells. These two sublayers are now considered in more detail.

7.5.1 PMD sublayer
ATM speci¬cations allow for travelling over media such as optical ¬bre, coaxial and
twisted-pair cables, at a range of data rates. For example, an ATM to the desktop scheme
travels at a rate of 25.6 Mbps over unshielded twisted-pair (UTP) category 3 cable in
what is known as ˜cell-stream™, i.e. the cells are sent as-is and not framed within another
protocol. Table 7.2, lists some of the principal carrier systems, the data rate and the
medium/media that can be used. Synchronous digital hierarchy (SDH)

Outside of the local area network (LAN), although ATM is independent of the under-
lying medium, it most commonly is implemented to run over SDH or SONET. SDH is
an optical communications standard and was established by the Consultative Commit-
tee for International Telegraphy and Telephony (CCITT), now the ITU-T. It came about
because so many telecommunications companies were running their own proprietary opti-
cal networks that interconnectivity was becoming a serious problem. A second standard,
synchronous optical network (SONET), was developed by Bell Labs, USA, just prior
to SDH. For the purposes of discussion, their differences are so minor that the two are
normally discussed together.
SDH speci¬es the communications mechanism at the physical layer, i.e. on the ¬bre.
Today, most long-distance telephone traf¬c runs over SDH, and because of the availability
of SDH equipment, it is straightforward for companies to plug into the network. The
SDH standard is expected to provide suf¬cient transport infrastructure for worldwide
telecommunications for at least the next two or three decades.
The key aims of SDH may be summarized as follows:

• internetworking between different carriers;
• uni¬cation of world digital networks;
• multiplexing together of digital channels;
• operations, administration and maintenance support.

Table 7.2 ATM carrier schemes
Frame format Data rate (Mbps) Media
Cell-stream 25.6 UTP-3
Cell-stream 155.52 STP, MM ¬bre
STS-1 51.84 UTP-3
FDDI 100 MM ¬bre
STS-3c 155.52 UTP-5, coax pair
OC-3 155.52 SM ¬bre, MM ¬bre
STS-12 622.08 SM ¬bre, MM ¬bre
E1 2.048 UTP-3, coax pair
T1 1.544 UTP-3

9 columns 261 columns


Frame 1
9 rows


Path overhead
Section Data

Frame 2


. 89
( 132 .)