A novel long-span traffic predictor for real-time VBR videos via ρ-domain rate model

Chu Chuan Lee, Pao Chi Chang

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

To predict the traffic of frames that may be far from the current frame, this letter extends the use of the ρ-domain rate model from macroblock-based source rate control to frame-based long-span traffic prediction. Moreover, this work enhances the linearity and convergence speed of ρ-domain frame-based rate function by adding a parameter that is the number of non-zero motion vectors. Simulation results reveal that the proposed predictor can significantly lower the prediction error compared with two conventional LMS methods. More importantly, the process of the proposed predictor is unique but simple for different video contents and prediction spans.

Original languageEnglish
Pages (from-to)279-281
Number of pages3
JournalIEEE Communications Letters
Volume9
Issue number3
DOIs
StatePublished - Mar 2005

Keywords

  • Real-time videos
  • Source rate control
  • Traffic prediction

Fingerprint

Dive into the research topics of 'A novel long-span traffic predictor for real-time VBR videos via ρ-domain rate model'. Together they form a unique fingerprint.

Cite this