Managing prerecorded traffic in packet-switched networks

We study how the emerging packet-switched networks can efficiently carry traffic from prerecorded multimedia sources. Toward this end we first develop call admission policies for statistically multiplexing prerecorded sources over a bufferless transmission link. We develop a stochastic model which captures the random phases of the sources. We then apply large deviation theory to our model to develop admission rules. The accuracy of the large deviation approximation is verified with simulation experiments employing importance sampling techniques. We next present a high-performance prefetching protocol for the delivery of prerecorded VBR video from a server across a packet-switched network to a large number of clients. The protocol requires (1) that each client have a small amount of memory dedicated to the application; (2) that there is one bottleneck shared link between the server and the clients. The server chooses prefetched frames according to a join-the-shortest-queue (JSQ) policy. We also develop a variation of the the JSQ prefetching protocol that allows for multiple servers that are distributed deeper into the network. We finally study call admission control and traffic management of regulated multimedia traffic. We develop a refined call admission control procedure for buffered multiplexers. This procedure involves searching over the buffered multiplexers\u27 buffer-bandwidth tradeoff curve and finding the most adversarial traffic pattern of the regulated sources. Importantly, we show that on-off is not the most adversarial traffic pattern of leaky bucket constraint sources. We next propose a traffic management scheme for multimedia traffic which consists of smoothers at the network ingresses and bufferless statistical multiplexing within the network. We explicitly characterize the smoothers that maximize the connection carrying capacity of the network. We develop call admission rules for our smoothing/bufferless multiplexing scheme that are based on the assumption of adversarial traffic at the smoother outputs