|dc.description.abstract||The opening up of telecommunications markets has forced a differentiation between service
and basic connectivity provision; many providers have accepted that new services will
become their main source of income in the face of regulations that ensure that the market
incumbent, and other providers with significant market power, provide unrestricted network
access in the common-carrier model.
One basic requirement for a new service is that it is capable of, at least, recouping its costs.
This is usually brought about by charging for service use, normally with a complementary
need to account for this use. Thus far, standardisation for accounting has concentrated on
mechanisms based on data collected from the resources used to support the communications
channel at the network level. In the common model this data is then correlated and aggregated
to produce a service transaction record which is in turn used as a basis for charges. The
applicability of this to multi-service networks in an age of abundant bandwidth is
questionable; the most successful of such multi-service networks, the Internet, depends on a
packet based transport mechanism whose network level usage data are not easily related to
specific users, or even specific services. Even with an agreed basis for charging, as is the case
with telephony, the interpretation of network level usage data to produce charges has largely
been declared out of scope of standardisation, and most usage data, although produced in a
standardised way, is interpreted to produce charges in a service specific way.
This thesis argues that network level accounting measures are not appropriate when
accounting for services provided on multi-service networks. It proposes a value, rather than
cost, based pricing mechanism founded on economic pricing models for network industries. It
then suggests an architecture to support this mechanism. This architecture promotes a clear
separation between a service?s operation and its accounting, enabling faster service