| Summary: | 碩士 === 國立屏東科技大學 === 資訊管理系 === 87 === ABSTRACT
Title of Thesis: Efficient Deterministic Multiplexing Techniques for Transporting Video Streams over Multi-hop Network Paths
Name of Institute: Graduate Institute of Management Information System, National Pingtung University of Science and Technology
Graduate Date: June ,1999 Degree Conferred: Master pages: 78
Name of Student: Wei-Chen Liu Advisor : Sheau-Ru Tong
Contents:
Rapid progress of video compression techniques in recent years has significantly reduced the data size of videos. And it becomes practical to deliver videos through networks. After compression, the sizes of video frames in a video sequence differ considerably. This nature, however, imposes a great challenge of network resource utilization. In the passe, most research have studied the trade-offs among several factors, such as bandwidth, buffer and delay based on the per-stream scenario. In fact, the sizes of video frames present themselves with a regular pattern. Our scheme termed a deterministic multiplexing (or DM), different from the formers, is then to multiplex video streams by exploiting such regularity. In principle, we apply a specific schedule for transmission of video streams based on their frame patterns. By doing so, streams properly compensate their resource requirements with each other (still with a 100% guarantee on their quality of services). The resource utilization is thus improved.
DM is further extended for wide-area networks, where a route usually consists of several hops. We propose an algorithm to perform the best DM in each hop along a path. The summation of the delay incurred by each hop is bounded by the end-to-end delay tolerable to the upper-layered applications.
For validating the correctness and superiority of the proposed scheme, we have simulated the system in Simscript simulation package. The experimental results reveal that when the number of video streams are many, we clearly observe the performance gain brought by DM. In general, a 60~70% utilization on bandwidth can be achieved with moderate buffer space deployed in intermediate hops.
|
|---|
