視訊串流於盡力傳送式網路
上之調適性平順化研究

摘要

   隨著寬頻網路的普及與多媒體壓縮技術的成熟，多媒體資料透過網路傳輸的相關應用日趨豐富。而不用等待多媒體資料的完整下載便可以即時播放的串流技術是目前視訊傳輸的發展趨勢。但由於目前的網際網路大都是盡力傳送式的環境，其變動的可用頻寬和網路的延遲變異等問題，均對具有叢集特性的視訊訊務其接收品質造成嚴重的影響。此外，在多媒體網路應用的蓬勃發展下，非TCP的資料流在網路中將佔有極大的比例；然而非TCP資料流大都不具有壅塞控制的特性，勢必造成TCP資料流的可用頻寬枯竭，進而癱瘓整個網路。

   在本論文中，根據上述問題提出一套在盡力傳送式網路下，具有調適性平順化傳輸機制的端點對端點視訊串流傳輸系統。本系統分為兩部分：視訊串流傳輸平順化機制與對TCP友善的傳輸率控制機制。本論文針對此兩個機制進行整合，首先根據初始的網路狀況、可用的緩衝區資源以及視訊訊務的特性，規劃視訊流量平順化的傳輸排程，並在一定的視訊接收品質保障下，利用對TCP友善的傳輸率控制機制，適時地反應網路壅塞情況予視訊傳輸平順化機制，進一步動態地調整平順化的傳輸排程以符合目前的網路現況，提高視訊接收品質。模擬結果顯示，在盡力傳送式的網路環境下，當平均可用頻寬約等於視訊所需的平均傳輸率時，可將原本僅使用TFRC的線上視訊串流其5.67%的封包丟棄率降至0.91%。
　

Adaptive Smoothing for Streaming Videos
over Best-Effort Network

Abstract

    With the technology advances in multimedia compression and Internet, multimedia applications delivered over Internet are dramatically boosted. Currently, real-time streaming without waiting for complete downloads is the trend of video delivery. However, it is very difficult to allocate resources effectively due to the bursty nature of video traffic. The impact of variable available bandwidth and network delay, which results from a best-effort network such as today’s Internet, may degrade the received video quality drastically. Besides, these multimedia streaming applications generally utilize the UDP protocol that does not provide the congestion control. This may lead to congestion collapse and starvation of TCP traffic in the Internet. Therefore, this thesis proposes an end-to-end adaptive video streaming system by integrating the TCP-Friendly Rate Control (TFRC) with the video traffic smoothing algorithm. The proposed framework can adaptively adjust the smoothing transmission schedule to ensure the smooth delivery quality for different real-time and pre-stored video streams with low quality degradation. The adaptation is based on the current network condition, the available resources and the characteristics of video traffic. Simulation results show that, when the average available bandwidth is close to the video target encoding rate, the proposed system can effectively reduce the total packet loss rate of online video streaming from 5.67% to 0.91% compared with traditional TFRC solution.