| Abstract |
With the emerging of high-definition video, a mobile station (MS) using a single radio interface may suffer insufficient bandwidth, leading to serious mosaic or even frame-frozen phenomenon. To solve the above problem, in this thesis, we propose an SVC Stream-Splitting Mechanism (SSSM) using multi-radio access technologies (Multi-RAT). In our proposed scheme, an SVC (Scalable Video Coding) streaming server divides a video stream into two sub-streams, and each sub-stream is delivered, respectively, via WiFi and LTE to an MS. When WiFi network encounters congestion, packet loss rate (PLR) will largely increase. A threshold of PLR is therefore set to divert the transmission of a certain upper SVC layers to LTE. One of the major contributions in this thesis is that we modify the DESCRIBE message of Real-time Transport Streaming Protocol (RTSP) such that the bit rate of each SVC layer can be delivered to an MS. Thus, an MS can compute the number of SVC layers to be diverted to LTE by comparing the received bit rates with the measured WiFi throughput. Additionally, since a video stream after the split into two sub-streams may produce out-of-order packets, we design a pre-buffer scheme to re-sequence the out-of-order packets by adding a common sequence number (CSN) in the Real-time Transport Protocol (RTP) header. Finally, we implement the SSSM on the Linux platform. From the experimental results, we demonstrate that the proposed SSSM with stream splitting can achieve higher throughput than that without using stream splitting. Also, the video stream received at MS can exhibit much better quality. |
|---|
