End-to-end Quality of Service (QoS) management for continuous media delivery over distributed network systems

Due to the phenomenal growth of multimedia systems and their applications e.g. IP Telephony, Video-on-Demand (VOD), there have been numerous research efforts directed at providing good Quality of Service (QoS) for continuous media (CM) services over various types of networks. Many network systems including the Internet provide a "single class best effort" service, and do not provide any sort of guarantees. Thus provisioning network resources for good QoS is inherently a complex issue and for the past several years there is considerable ongoing research within this field of QoS support for distributed multimedia systems.; Meeting QoS guarantees in distributed multimedia systems is fundamentally an end-to-end issue i.e. from an end host system like a server to another end host system like a client. It is critical for continuous media flows that QoS is configurable, predictable and maintainable system wide, including end systems, communication subsystems and network systems, and all of them working in conjunction and unison to achieve desirable application level behavior. Most of the research and development till date can be broadly classified into the following one of the two areas: end-host systems and network systems but never both. There is need for an overall framework, whose overarching philosophy is to build upon and manage QoS at both the end-host and network systems level as also bridge/translate the disparity existing between them in a meaningful manner. The following dissertation addresses this very issue.; This dissertation decomposes the overall managing and provisioning of end-to-end QoS into three important and independent sub problems and presents solutions for each of them respectively. First, we try to figure out "where to place" geographically distributed "proxy servers" on the network which manage content on behalf of the server for the clients. Second, we address the problem of admission decision and control through QoS expansion and contraction. Finally we consider a QoS based delivery adaptation mechanism which tries to ensure that the application maintains the desired QoS due to unanticipated variations in the resource availability of the network.