Scalable layered multiple description video transmission systems based on H.264

Chien Min Ou*, Wen Jyi Hwang, Chu Ting Chou

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


This paper presents a layered multiple description coding (LMDC) scheme for a scalable video delivery system which operates over heterogeneous networks and distributes real-time streaming video to diverse types of clients. The LMDC is a hybrid combination of discrete wavelet trans-form (DWT) and H.264. In the algorithm, an input video sequence is first decomposed into a fundamental sequence and a number of orthogonal supplemental sequences using DWT. Each sequence is encoded by H.264 for effective exploitation of spatial and temporal correlations. As compared with its layered coding counterparts for streaming over networks providing transportation prioritization, the LMDC offers a wider range of bit rates and a smoother transition in video quality for variations on the network capacity. When the transportation prioritization is not available, the same LMDC-encoded bitstreams can still be reused to form a pure multiple description bitstream for scalable and robust transmission. The existing H.264 codec software and hardware can be reused for the implementation of the proposed algorithm. The LMDC therefore is a low cost solution to the realization of the scalable video streaming. The LMDC has superior performance over motion JPEG2000 and MPEG4. It also outperforms the H.264-based simulcast systems subject to the same rate for information delivery.

Original languageEnglish
Pages (from-to)365-377
Number of pages13
JournalInternational Journal of Electrical Engineering
Issue number4
Publication statusPublished - 2006 Nov


  • Layered coding
  • Layered multiple description coding
  • Multiple description coding
  • Scalable coding

ASJC Scopus subject areas

  • Electrical and Electronic Engineering


Dive into the research topics of 'Scalable layered multiple description video transmission systems based on H.264'. Together they form a unique fingerprint.

Cite this