CSI - An Adaptive Data Pulling Framework for Supporting Time-Critical Streaming Media Applications

点击翻译按钮获取中文摘要

Timely delivery of data streams is important to time-criticalapplications relying on streams of dynamic data. Drastic reductionon the sending rates of data streams typically has negative impact onthe operations of such applications. Effective rate adaptation isneeded to make data streams adapt to network congestion through slowlyvarying their sending rate. This research project is motivated bythe need of streaming dynamic radar data at the required rates fromradars to a control center in the surveillance application using gangedground-based radars. This application is crucial to the operationof the unmanned aerial vehicles.In this research project, an adaptive data pulling framework isstudied to support stable data streaming. This framework consistsof a transport-layer congestion control scheme and anapplication-layer data pulling control mechanism. Theapplication-layer data pulling control mechanism allows theapplication actively fetch data from sources. The transport-layercongestion control scheme adopts a new rate control method whichprovides weighted fairness to data streams. Weighted fairnessamong data streams is the desired feature for controlling the sendingrates of streams carrying layered encodings. Weighted fairnessmeans that the throughput of a stream is degraded reverselyproportional to its priority when the network becomes congested.Weighted fairness brings new vision on rate control for datastreaming. Stability and responsiveness to congestion indicationsare the two key research questions of the new rate control scheme.