Supporting rate adaptive multimedia streams on the Internet
MetadataShow full item record
This thesis investigates the feasibility of using rate adaptation, i.e., selective service degradation, as a mechanism for achieving various system level Quality of Service (QoS) targets on communication networks. In particular, we investigate how to optimally support rate adaptive multimedia streams on best-effort networks like the Internet. Optimal and practical mechanisms to maximize the client average QoS, de- fined in terms of a normalized time average received rate, are established. By scaling the arrival rate and link capacity, we obtain closed form expressions for asymptotic client average QoS. The optimal adaptation policy is identified as the solution to an integer programming problem which has an intuitive “sort by volume” inter- pretation. Our asymptotic analysis shows the optimal adaptation policy may yield performance improvements of up to 42% over baseline policies. We demonstrate that a static multi–class admission control policy can achieve the same asymptotic QoS as that of the optimal adaptation policy. This implies that dynamic adaptation may be unnecessary for large capacity networks with appropri- ate call admission. The multi–class admission policy, however, requires the stream load char- acteristics be both stationary and known a priori. To address this drawback we investigate a class of distributed algorithms where the frequency of rate adapta- tions depends on the stream “volume.” We show that these algorithms are able to achieve a QoS comparable to that achieved under the optimal adaptation policy, but without requiring knowledge of system wide parameters. Our simulations indi- cate our algorithm may yield performance improvements of up to 28% over baseline algorithms. Finally, we investigate using optimal adaptation in a networking environment supporting multiple service classes with distinct QoS guarantees. Our results confirm that rate adaptation, i.e., selective service degradation, is a viable means of achieving several different types of system level quality of service targets.