Adaptive Video Streaming With Optimized Bitstream Extraction and PID-Based Quality Control

To cope with the challenges brought about by bandwidth fluctuation and improve the experience of watching online videos, an adaptive video streaming system that can adjust video quality according to actual network conditions is proposed based on the scalable video coding (SVC) extension of H.264/AVC. First, a simple and effective linear error model is proposed and verified for quality scalability of SVC. The model exploits the linear feature of pixel value errors and can be used to accurately estimate the distortion caused by discarding any combination of enhancement data packets in an SVC bitstream. On that basis, a greedy-like algorithm is designed to assign each data packet a priority value according to its rate-distortion (R-D) impact, thus enabling R-D optimized bitstream extraction under certain bitrate constraints. Finally, the proportional-integral-derivative (PID) method is utilized to control the video quality adjustment and determine a suitable bitrate for transmission. By monitoring and predicting the past, current, and future bandwidth information, the PID-based quality control algorithm is able to reduce quality fluctuation, while still preserving a high quality level. Experimental results show that compared with the baseline software, the proposed system that integrates the above algorithms can achieve much lower video quality fluctuation, with PSNR variance reduced from 1.24 to 0.69, and at the same time deliver higher video quality, with the PSNR average increased by 0.83 dB.