Bandwidth Smoothing Algorithms
In this chapter, we address the issues involved with smoothing the bandwidth requirements for a single stream of video data. Video applications, such as video-on-demand services, rely on both high-speed networking and data compression. Data compression can introduce burstiness into video data streams, which complicates the problem of network resource management. For live-video applications, the problem of video delivery is constrained by the requirement that decisions must be made on-line and that the delay between sender and receiver must be limited. As a result, live-video applications may have to settle for weakened guarantees of service or for some degradation in quality of service. Work on problems raised by the requirements of live video includes work on statistical multiplexing [COHE94, REIN93], smoothing in exchange for delay [WALL91], jitter control [PARE92, JEFF92], and adjusting the quality of service to fit the resources available [PANC94, PANC92]. For stored video applications, the system can take a flexible approach to the latency of data delivery. In particular, it can make use of buffering to smooth the burstiness introduced by data compression. Because the entire video stream is known a priori, it is possible to calculate a complete plan for the delivery of the video data that avoids both the loss of picture quality and the wasting of network bandwidth through overstatement of bandwidth requirements.
KeywordsBuffer Size Bandwidth Allocation Bandwidth Requirement Compressed Video Smoothing Algorithm
Unable to display preview. Download preview PDF.