中文摘要 因為現今行動裝置的普及以及無線網路的進步，網路頻寬的提升，影音串流服務在行動裝置上的應用越來越多，人們可以在任何時間、任何地點使用行動裝置流暢地觀看自己喜愛的影音串流服務。影音串流服務在應用上分為影音即時串流及隨選影音串流，影音即時串流是將當下的影音內容以最即時的傳輸速度送到使用者的行動裝置上，此類型常見的應用包括視訊對話、即時監控系統及網路直播串流。 影音即時串流在有其他封包流量的網路環境下會遇到網路延遲與抖動（Jitter）的問題，資料封包傳送到目的地會有不固定時間的延遲。這問題會降低使用者的使用經驗。在聲音串流中，大多會用資料流量控制與抖動緩衝區來解決網路延遲與抖動的問題。影像串流龐大的資料量容易被網路延遲與抖動造成影像資料流量不穩定且無法預測，另外，因為編碼與應用上的關係，並不能隨意遺失或丟棄封包，所以影像串流對於網路延遲與抖動的問題與聲音串流的處理方式所有不同。 自適應多媒體播放機制是解決影像串流抖動問題的方法之一，它依據緩衝區的狀態去控制影像的播放速率。因為這個機制存在著即時性的問題，所以它並不適合使用在及時影像串流中。 本論文提出了一個專門應用於即時影像串流的自適應多媒體播放系統，系統在自適應多媒體播放機制中使用了以抖動情形為基礎的播放控制演算法來增進及時影像串流的表現。這個機制同時考量了流暢與即時性兩個要素。另外，我們將這套系統實作於行動平台上來驗證我們的系統在現實環境中的能力。實驗結果表示出該系統有著更穩定、更流暢且不失即時性的能力。 ;ABSTRACT Multimedia streaming applications have been increasing because wireless network such as Wi-Fi, 3G, and LTE have been rapidly growing. More and more people use wireless networks to watch streaming programs on their mobile platform. They can choose to watch their favorite streaming programs anytime. Multimedia streaming has two main services: live multimedia streaming and on-demand multimedia streaming. Live multimedia streaming will deliver the multimedia content as soon as possible in order to let users watch the real-time content. Nowadays using live multimedia streaming such as video call, surveillance system and live broadcast are common. Network delay and jitter constrained smoothing for live multimedia streaming. These problems reduce the quality of the multimedia experience. Most of audio streaming systems use rate control mechanisms to control data rate or use jitter buffer to solve jitter problems. The video frame arrival rate is unstable and unpredictable. Besides, the video packets dropping are restricted by video codec and application. Thus, the solutions of video streaming system and audio streaming system for network delay and jitter problems are different. Adaptive media playout (AMP) mechanism, which is one of the solutions for the video streaming jitter problem, controls the video playout rate depends on the status of its jitter buffer. The mechanism has the timeliness problem, so the mechanism is not suitable for live video streaming systems. In this thesis, we proposed an adaptive media playout system adapted specially for live video streaming. The system puts jitter-based playout control algorithm in our AMP mechanism for improving the performance of live video streaming systems. The mechanism considers both smoothness and timeliness. In addition, we implement the system on mobile platforms in order to demonstrate the performance of our system in reality. Experimental results show that our system is more stable, smoother, and keeps the timeliness.
