| dc.description.abstract | With the current trend towards user mobility and ubiquitous computing, cellular networks are becoming
an evermore important feature of day-to-day life, albeit an unseen one. Mobile networks are
characterised by a scarcity of resources, particularly bandwidth and frequency spectrum. However,
for new multimedia applications, such as video telephony, a large amount of these resources is required
and, moreover, these applications demand that specific quality of service guarantees are met
by the network at all times.
In a cellular network, the traffic is highly variable both temporally and spatially. Therefore, dimensioning
a network so that it can meet peak-time demand is both uneconomic and inefficient, as most
of the time the network will be under-utilised. This leads to frequent and significant congestion in
mobile networks, so that, at a certain time and place, users may find it impossible to start a phone
call, or an ongoing phone call may be interrupted. Some solutions have been proposed to alleviate
the problem of congestion without installing new infrastructure. However, these schemes only improve
the network performance for some incoming traffic rates, but cannot meet QoS guarantees at
peak-times.
Another possible solution to this problem is to attempt to modify the user demand to fit the available
resource. This leads to dynamic pricing: charging users according to the current traffic conditions,
hence providing negative or positive incentives to regulate the traffic entering the network. As they
know the price they will be charged, users can decide whether to make the phone call or not, and
the importance of the call will influence their choice. Hence dynamic pricing leads to a natural
prioritisation of calls, ensuring that only low priority calls are blocked.
Dynamic pricing has been applied successfully in several domains, but its application to cellular
networks is an emerging research area and is particularly challenging due to the mobility of users.
This project investigates dynamic pricing in cellular networks from a technical perspective. For this
purpose, detailed network and traffic models are defined based on an investigation of current research
in related areas. These models are implemented in a simulator which is then used to test, refine and
improve existing dynamic pricing schemes for both GSM and GSM/GPRS networks. | en |
