Forecasting Application-level Quality of Service (QoS) Metrics for Real-time Multimedia in Best-effort Networks

Internet performance is far from being satisfactory while transporting continuous media or multimedia, such as video and audio, requiring extensive use of caching and buffering. The case is worse for real-time multimedia content generated byvideoconferencing and Internet telephony applications, where extensive buffering is not feasible. Persistent and uncertain end-to-end delays, delay variations (jitter), packet losses and throughput result in unsatisfactory end-user experience. Severalalternatives have been proposed for efficient transport of real-time multimedia. Most require network infrastructure modifications to support the proposed multimedia-related protocols. This project will address forecasting application-levelQuality-of-Service (QoS) metrics for real-time multimedia applications in best-effort networks, such as the Internet, using the recently developed Active Content Control (ACC) technology. This adaptive flow control technology embedded in RTSwitch, acontent switch for real-time applications residing at the network edges, uses forecasts of application-level QoS metrics to enhance the performance of best effort networks in transporting real-time multimedia, without any core network infrastructuremodifications. In Phase I of this SBIR project the feasibility of forecasting application-level QoS metrics for real-time multimedia in best-effort networks will be demonstrated through comparisons with network measurements. Based on our recent success inforecasting application-level QoS metrics for non-real-time multimedia applications, the proposed project will extend the ACC technology to real-time multimedia applications. The commercial potential of an application-level QoS forecasting technology forreal-time multimedia applications transported over best-effort networks, such as the Internet, is enormous. Its immediate use will be in application-level QoS control of real-time multimedia applications. Every segment of the Nation's economy, includingthe defense establishment, can greatly and immediately benefit from such a technological development. From videoconferencing and telephony applications for distance learning and training, to remote surveillance for national security purposes, to remotelyoperated medical equipment and instruments, to remotely controller factory automation equipment, networked real-time multimedia applications are countless. The worldwide videoconferencing market alone is currently estimated at $3 billion in product andservice revenues. Besides its commercial potential, the proposed technology will have a significant societal impact for offering to the public-at-large a cost-effective medium for real-time multimedia communications.