SVC-based scalable multiple description video coding and optimization of encoding configuration

Abstract

It is well known that multiple multicast trees can avoid single point of failures in peer-to-peer (P2P) streaming originating from a single source by providing path diversity. Multiple description video coding is well suited for serving video over multiple multicast trees, such that each description can be streamed over a different tree. In scalable multiple description coding (SMDC), each description is scalable coded (e.g., derived from a single scalable video coding (SVC)-compliant bitstream) so that descriptions can be efficiently adapted to the available rate of each link for effective congestion control. This paper proposes (i) new SMDC generation methods based on various combinations of video segments (GOPs or fames) coded at high and low rates (one or more layers) and (ii) a multiple-objective optimization (MOO) framework for selection of the best encoding configuration of the proposed SMDC from a set of candidates in order to achieve the best tradeoff between redundancy and reliability (measured by end-to-end rate-distortion performance over a range of extraction points for a set of packet loss probabilities), while maximizing the range of extraction points of each scalable description. The optimization is performed over selected SVC encoding parameters and MD generation alternatives that feature various levels of redundancy at a fixed total rate for all descriptions. The framework is generic to allow optimization over other SVC encoding variables and MD generation methods, or just over the SVC encoding variables without MD, if desired. Results of Monte-Carlo simulation of SMDC streaming with two descriptions over two paths with packet losses are provided to demonstrate performance of the proposed methods.