Secure Media Streaming & Secure Adaptation for Non-scalable Video (TP-S1)

Author(s) : John Apostolopoulos (HP Labs, USA)

Abstract : This paper examines three topics related to source and network heterogeneity and how they can be used to improve video streaming performance. First, we examine the difference in importance of individual coded P-frames in a sequence of P-frames, where some P-frames may be worth more than a order of magnitude more than the typical P-frame in the sequence. Second, we extend our prior work on examining the effect of packet loss pattern by quantifying the performance change that results for trading off average packet loss rate and average burst loss length. Specifically, to minimize the total MSE distortion, it can be beneficial to operate in a manner that leads to a higher average packet loss rate, if it also leads to fewer burst losses. Third, we extend prior work to assess the potential benefits of using multiple heterogeneous paths for streaming, when both the path characteristics and importance of each media packet are known. These results may be incorporated with other R-D optimization algorithms to improve video streaming performance.

Menu