Predicting and measuring quality of service for mobile multimedia

Telecommunications is continually evolving from its historical remit of offering speech over fixed lines to an increasingly feature-rich and mobile environment. The next generation of mobile networks, such as the General Packet Radio Service (GPRS) and the Universal Mobile Telecommunications Service (UMTS) will be able to offer multimedia (combined audio and video) using the packet-based, connectionless technology of the Internet. An important factor in the delivery of such services is the quality of service (QoS) as perceived by the customer. This paper shows how an understanding of human perception and the simulation of a mobile IP network may be used to tackle the relevant issues of providing acceptable QoS for mobile multimedia. The paper starts with an overview of how studies into human perception and cognition have led to work on predictive computer models of the human perception of quality. It also discusses the Napoleon system for testing applications across a degraded IP channel and gives results from some preliminary studies. Finally, it shows how both sets of work may be used to educate applications developers to ensure that the finite resources available for mobile multimedia are used efficiently to provide a QoS to the customer that is both acceptable and robust.