Research On Key Techniques Of Real-time Streaming System

Recent developments in computing technology, video compression technology, high bandwidth storage devices, and high-speed networks have made it feasible to provide real-time streaming media applications over the Internet. However, the current Internet is a best-effort network where loss, variable delays and bandwidth fluctuation occur during data delivery in a streaming session. Traditional non-scalable streaming cannot adapte well to variation of the available network bandwith. Scalable/layered encoding has been believed to be promising to provide network adaptive streaming service. FGS (Fine Granular Scalable) coding schemes is accepted as scalable video coding by the MPEG-4 standard. FGS video streaming is to code a video sequence into a base layer and multiple enhancement layers. The base layer uses non-salable coding to reach the lower bound of the bit-rate range and must be transmitted. The enhancement layer may be truncated into any number of bits or not transmitted. Therefore, the FGS streaming can adapt a wide range of data rate variability by distributing enhancement layers over a wide range of bit rates. In this thesis, we investigate how to schedule and send the packets by real-time scheduling scheme and how to provide an interactive real-time streaming service over network with packet losses and variable delay. The main contributions of this thesis are as follows:1. A real-time network-adaptive architecture for end-to-end streaming system is proposed. The important concept"real-time"is introduced in the framework. All the packets must be tramsitted before their deadline, otherwise they can not be decoded on time. This is the foundation of this thesis.2. This thesis analyses the real-time characteristic of scalable streaming system and proposes a layer-based real-time scheduling algorithm. And then, a real-time imprecise computation workload model and an imprecise computation scheduling algorithm on scalable media stream delivery are proposed. The scheduling task of each stream is divided into two parts: a mandatory subtask and an optional subtask. The mandatory task is for the base layer substream and the optional task is for the enhancement layer substreams. Different subtask adopts different priority. The imprecise computation scheduling algorithm enables the use of imprecise computation workload model as a means to provide scheduling flexibility in scalable streaming systems and enhance their fault tolerance and improve utility of the system resource and the playback quality in client. The low complexity of the proposed algorithm also enables them to be applied in real-time applications.3. This thesis proposes an online smoothing algorithm for a single of scalable media with the variable network bandwidth. Without smoothing, the playback quality would be variation...