Techniques for Improved Rate-Distortion Optimized Video Streaming

We present techniques recently developed by our research group at Stanford University to improve the performance of existing algorithms for rate-distortion optimized video streaming. In rate-distortion optimized streaming, media packet transmissions are controlled in a way that gives an optimal tradeoff between the expected bit-rate that will be required for the transmissions and the expected media reconstruction distortion that will result. In one of the techniques that we present here, performance is improved by incorporating richer, more reliable feedback information. In another, multiple arrival deadlines are considered for media packets. A third technique utilizes an improved statistical model characterizing packet delay and loss. Another seeks to minimize congestion when media is streamed over abottleneck link.