Adding definition to active probing

  • Authors:
  • Sridhar Machiraju;Darryl Veitch;François Baccelli;Jean C. Bolot

  • Affiliations:
  • Sprint ATL, California;University of Melbourne, Australia;INRIAENS, Ecole Normale Supérieure, France;Sprint ATL, California

  • Venue:
  • ACM SIGCOMM Computer Communication Review
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Active probing techniques have overwhelmingly been based on a few key heuristics. To progress to the next level a more powerful approach is needed, which is capable of filtering noise effectively, designing (and defining) optimal probing strategies, and understanding fundamental limitations. We provide a probabilistic, queueing-theoretic treatment that contributes to this program in the single hop case. We provide an exact inversion method for cross traffic distributions, rigorous system identifiability results to help determine what active probing can and canâ聙聶t achieve, a new approach for treating queueing theoretic â聙聵noiseâ聙聶 based on conditioning, and cross traffic estimators with enhanced properties