Moment estimation in delay tomography with spatial dependence

  • Authors:
  • Ian H. Dinwoodie;Eric A. Vance

  • Affiliations:
  • ISDS Box 90251, Duke University, Durham, NC 27708, United States;ISDS Box 90251, Duke University, Durham, NC 27708, United States

  • Venue:
  • Performance Evaluation
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Parametric models for estimating network link delays with incomplete data that incorporate spatial correlation are formulated. Fast numerical methods for estimation of parameters and confidence intervals are derived. It is shown that including correlation in the model gives better model fit on sample data. Finally, certain results from graphical models are described, which can be applied in model search.