Distributed Adaptive Multimedia Delivery in Kendra

Currently Metadata is used to locate information on the Internet. The Kendra project’s aim is to exploit metadata further by using it to drive the adaptation of Internet data distribution and delivery architectures. The central hypothesis is that metadata not only can increase retrieval effectiveness but also improve delivery performance and address its temporal characteristics. We are developing a distributed multi-media delivery system which utilises intelligent, distributed caching mechanisms to improve availability, performance and reliability while minimising storage space and network utilisation. In this paper we introduce a metadata reference model for multimedia entertainment data. We show how the model permits specification of metadata describing not only configurations of the delivery system tailored to the data being delivered, thus improving performance, but also adaptations, allowing the delivery system to respond to a degradation in its environment by reconfiguring itself. We illustrate our approach by describing an Internet audio-on-demand service.