We are currently witnessing a rapid growth of the wireless multimedia market. This in turns results in an increasing traffic volume over wireless network segments, as well as the need to multiplex heterogeneous services (typically IP-based) over a same radio access technology.
A first approach to cope with this emerging traffic growth is overprovisioning, i.e. increase resources to avoid the occurrence of congestion and thus provide toll-quality service levels on an end-to-end basis. Most of the datacom/IP industry argues that there is little need for sophisticated QoS mechanisms [1] (at least in the field of wired networks). Indeed, although the Internet Engineering Task Force (IETF) has carried out two large QoS related projects (Differentiated Services [2] and Integrated Services [3]), we have seen very little business drivers that demand such QoS architectures. This is because the vast majority of IP networks is lightly loaded, and there are research data available that indicate that it will remain that way for the foreseeable future [4]. In addition, overprovisioning helps to eliminate the need to maintain state information in the network, which helps in keeping the network architecture as simple as possible [5].
As IP meets the wireless world, the debate whether overprovisioning is a viable solution for QoS is still open. On one side, the proponents of open spectrum argue that spectrum itself is not such a scarce resource as many believed so far [6]. Also, the tremendous success of wireless local area networks, basically without any support for QoS differentiation mechanisms seems to reinforce the argument that providing application level QoS is technically possible over best effort networking technologies. On the other side, claims about the abundance of spectrum resources remain questionable (just think about the tight regulatory policies, the narrowness of the ISM bands, and the price GSM operators in Europe had to pay for licenses). Moreover, a strong standardization effort is being carried out in the frame of the 802.11 task group e, to develop service differentiation mechanisms. This appears to be a clear signal that the wireless LAN community ultimately suffers from the lack of effective QoS mechanisms. Also, 3G networks and services are currently being deployed all over the world. These networks employ sophisticated QoS management and make very efficient use of spectrum resources, be they based on the cdma2000 or the UMTS standards suite. 3G network operators are interested in QoS mechanisms, because they help reduce operational and capital expenditures and facilitate the timely introduction of new services. Finally, the emergence of beyond 3G architectures and the integration of various multiple access technologies over a possibly scarce wireless spectrum opens new exciting issues on the design of QoS architectures, which need to include resource management mechanisms at various levels (admission control, scheduling, routing when multi-hop wireless networks are considered, differentiated channel access mechanisms and priority support, etc) to either Copyright # 2004 John Wiley & Sons, Ltd.