Open Access

SMART: An Efficient, Scalable, and Robust Streaming Video System

  • Feng Wu1Email author,
  • Honghui Sun1,
  • Guobin Shen1,
  • Shipeng Li1,
  • Ya-Qin Zhang1,
  • Bruce Lin2 and
  • Ming-Chieh Lee2
EURASIP Journal on Advances in Signal Processing20042004:421079

https://doi.org/10.1155/S1110865704310218

Received: 9 December 2002

Published: 4 March 2004

Abstract

SMART, the acronym of scalable media adaptation and robust transport, is a suite of compression and transmission technologies for efficient, scalable, adaptive, and robust video streaming over the best-effort Internet. It consists of two indispensable parts: SMART video coding and SMART video streaming. The SMART video coding part is an efficient DCT-based universal fine granularity scalable coding scheme. Since the SMART video coding scheme adopts multiple-loop prediction and drifting reduction techniques at the macroblock level, it can achieve high coding efficiency at a wide range of bit rates. More importantly, it provides all sorts of scalabilities, that is, quality, temporal, spatial, and complexity scalabilities, in order to accommodate heterogeneous time-variant networks and different devices. The SMART video streaming part is a transport scheme that fully takes advantages of the special features of the scalable bit streams. An accurate bandwidth estimation method is first discussed as the prerequisite of network adaptation. Then, flexible error resilience technique and unequal error protection strategy are investigated to enhance the robustness of streaming SMART bit stream. The SMART system shows excellent performances with regard to high coding efficiency, flexible channel bandwidth adaptation, smooth playback, and superior error robustness in static and dynamic experiments.

Keywords and phrases

video streaming scalable video coding fine granularity scalability video transmission bandwidth estimation error resilience unequal error protection

Authors’ Affiliations

(1)
Microsoft Research Asia, 3F Sigma Center
(2)
Microsoft Corporation

Copyright

© Wu et al. 2004