Multiple Description Video Coding Over Multiple Path Routing Networks
Shaoshuai Gao, Hamid Gharavi
This paper presents a multiple description (MD) video-coding scheme, which uses interlaced high signal to noise ratio (H-SNR) and low signal to noise ratio (L-SNR) coded frames to produce two bit-streams. At the decoder, when both bit-streams are received a high quality video will be reconstructed. If either one is received, a poorer but acceptable quality video can be reconstructed. In this approach, we also considered the mismatch between the encoding and decoding loops due to motion compensation if only one bit-stream is received. We first give the results of the rate distortion performance of our scheme assuming that only one description is received. We then present simulation results under packet loss circumstances, which are simulated by real-world IP networks and ad-hoc networks. It is shown that the proposed scheme has a superior performance when compared with the single description MPEG-4 video coder and MD coding using video redundancy coding.