haps even more in the terminal equipment. For
the latter, there are factors affecting the user per-
ceived quality, such as (see [Reyn01]):
- Speech coding applied (e.g. G.711, G.726,
G.729, G.723.1, GSM); - Packetisation efficiency, including how many
samples are put into the same packet; - Silence suppression;
- Error-concealment methods;
- Codec-tandem performance.
Examples of these, referring to the E-model, are
given in [Reyn01] and [Vlee01].
One may say that the bit rate per voice channel
is a measure for efficiency. Basically, this effi-
ciency can be increased by:
- Using low bit rate codings;
- Increasing the packet lengths (less overhead
compared to the payload); - Multiplexing speech samples from several
conversations into the same set of packets; - Compressing headers, e.g. for the combination
of IP/UDP/RTP; - Suppressing silence periods.
8 Concluding Remarks
Facing the dynamic environment, a network
operator looks for a general-purpose network.
Most often these days, this network is IP-based.
However, is should also be “future proof” in the
sense that future services are supported. This
asks for mechanisms additional to pure IP for-
warding, inviting for Traffic Engineering mecha-
nisms. A part of the argumentation, at least to an
incumbent network operator, is the need to con-
solidate his current portfolio of networks, being
PSTN/ISDN and others, such as Frame Relay,
ATM and X.25. During this, more simple and
efficient solutions are sought, including all
aspects, like infrastructure, service network,
service control, management, etc.
Before the complete network integration has
been achieved, interworking between different
networks is needed. In particular, the telephony
service will likely require gateways between IP-
based network and PSTN.
When the access networks are upgraded, for
instance by introducing xDSL, the total traffic
loads into IP-based networks are expected to
increase drastically. This places further pressure
on the router networks. The requirements are
expected not only to refer to higher throughput
measures (e.g. capable of handling several Gbit/s
and Tbit/s), but also to offer differentiated and
ensured service levels. This allows for support-
ing a wider range of applications and accompa-
nying ranges of traffic characteristics. However,
a lot of challenges remain to be solved for mak-
ing complete solutions, including customer
equipment, service control and management.
A suggestion is to logically divide the network
into a number of virtual networks, e.g. each of
the networks supporting certain types of traffic
flow characteristics.
Considering the range of customers connected,
some more advanced than others, there will also
be a need for tailoring the service portfolios.
Some customers may well be more or less self-
provided (self-service), while others may want
more “customer nursing” packages. This again
asks for differentiation and adaptation. A particu-
lar challenge is to arrive at fast, accurate and effi-
cient ways of implementing such mechanisms in
the operator’s organisation and systems.
Further supporting mobility implies the need to
decouple the static home address of each termi-
nal/user from its current whereabouts. This will
place performance/capacity requirements on
mobility-like servers. It also means that interoper-
ability/interconnection arrangements between
several providers/operators must be solved.
Exporting/importing relevant information, e.g. for
roaming users, must be done in a secure and effi-
cient way. Other interactions should also be auto-
mated, for instance applying e-commerce solu-
tions. This would also open for having more
dynamic arrangements between the different
actors; customers, network operators and service
providers, making the challenges of adequate traf-
fic engineering functions even tougher to fulfil.
In this article the more basic topics have been
addressed. These are directed towards the IP-
based network, the main protocols, as well as
relations with underlying media (fibre) and cer-
tain applications of the IP-networks (VPN, tele-
phony, mobility, multicast). By no means is the
Figure 28 Interconnecting
gateways with examples of
MGC MGC protocols
MG MG
SIP, H.245/H.450
BICC
RTP
Megaco/
H.248