Quality of service control for real-time variable bit rate video transport in broadband networks

The introduction of asynchronous transfer mode (ATM) based Broadband Integrated Services Digital Network (B-ISDN) has made it possible to support a new range of bandwidth-demanding variable bit rate (VBR) services more efficiently. Of these new services, VBR video services are expected to play a major role in the evolution of these networks. The goal of this work is to study an important issue related to VBR video transport: the efficient and robust provision of quality of service (QoS) guarantees for real-time VBR video services. In order to prevent cell loss incurred by congestion, ATM networks implement traffic control based on call admission control (CAC) and usage parameter control (UPC). The effect of leaky bucket (LB) based traffic control on VBR video transport is first investigated and it is shown that to achieve the desired high network resources utilization, a substantial amount of video traffic is subject to losses, this reduces the quality and can greatly disrupt the video service. To combat such losses, two measures that can be implemented by the end-user are investigated: error concealment to recover from losses and source rate control to prevent losses. Experimental results revealed that when combined with the LB based traffic control inside the network, neither can provide satisfactory video quality. The main reason is the LB algorithm\u27s inability to capture the significant burstiness in multiple time scales exhibited by VBR video. The questionable suitability of the LB scheme for VBR video leads us to examine other useful techniques for admission control and bandwidth allocation for VBR video services. One solution would be to allow renegotiation of traffic parameters over the duration of a call. A receiver-initiated resource renegotiation scheme is proposed and evaluated. Traffic control schemes to support and provide QoS for renegotiated VBR service are also investigated. Finally, effective means to reduce the renegotiation overhead are presented which make resource renegotiation feasible even in large-scale networks