A dynamic resource reservation and pricing policy for scalable video delivery

Multimedia computer applications allow users to retrieve and display multimedia data in real time from distributed sources. An important component of an end-to-end multimedia delivery system is an architecture for reliable and timely delivery of real-time data with Quality of Service (QoS) guarantees. Existing network protocols were designed to transfer data without real-time constraints, whereas traffic generated by video applications requires strict performance guarantees from the network. We propose mechanisms for the delivery of continuous video over packet-switched, virtual-circuit-oriented computer networks. Using video scalability as a basis, We formulate a pricing policy and admission control algorithm, and integrate these mechanisms into a dynamic renegotiation protocol.; Video scaling is permissible due to human tolerance to degradation in the quality of presentation. This facilitates the specification of a range of resource requirements, which we call the "admissible region." With this scheme, a connection can be admitted by the network if it can meet the constraints of the admissible region. During congestion, the network can scale down the new or existing connections within this range while maintaining the specified QoS requirements. We study the admissible region for a number of video sequences and propose mechanisms for its computation in the general case.; While video scaling improves network utilization and connectivity, users will not tolerate the corresponding quality degradation unless it is offset by monetary or availability incentives. We propose a pricing policy and admission control scheme for this purpose. The policy trades-off performance degradation with monetary incentives to improve user benefit and network revenue. We demonstrate by simulation that this policy encourages users to specify application scalability to the network.; Finally, we propose a dynamic resource reservation protocol which provides a framework for the implementation of the aforementioned scaling and pricing mechanisms. The protocol provides control and signaling for resource reservation during connection set-up and subsequent renegotiation when connections are scaled. The protocol addresses performance issues from both the network and user's perspective and is shown to improve network revenue and user benefit.