Learn R Programming

An R package for network tomography

networkTomography implements the methods developed and evaluated in Blocker and Airoldi (2011) and Airoldi and Blocker (2012). These include the authors' own dynamic multilevel model with calibration based upon a Gaussian state-space model in addition to implementations of the methods of Tebaldi and West (1998; Poisson-Gamma model with MCMC sampling), Zhang et al. (2002; tomogravity), Cao et al. (2000; Gaussian model with mean-variance relation), and Vardi (1996; method of moments). Data from the 1router network of Cao et al. (2000), the Abilene network of Fang et al. (2007), and the CMU network of Blocker and Airoldi (2011) are included for testing and reproducibility.

Copy Link

Version

Install

install.packages('networkTomography')

Monthly Downloads

32

Version

0.3

License

LGPL-2

Issues

Pull Requests

Stars

Forks

Last Published

January 10th, 2014

Functions in networkTomography (0.3)

cmu

CMU data from Blocker & Airoldi (2011)
llCalibration

Evaluate marginal log-likelihood for calibration SSM
dobj.dxt.tomogravity

Analytic gradient of objective function of Zhang et al. 2003
locally_iid_EM

Run EM algorithm to obtain MLE for locally IID model of Cao et al. (2000)
ipfp

Function to run basic IPFP (iterative proportional fitting procedure)
grad_smoothed

Compute analytic gradient of Q-function for smoothed EM algorithm of Cao et al. (2000)
vardi.algorithm

Run algorithm of Vardi (1996) given B and S matrices
bayesianDynamicFilter

Function for inference with multilevel state-space model
smoothed_EM

Run EM algorithm to obtain MLE (single time) for smoothed model of Cao et al. (2000)
Q_smoothed

Q function for smoothed EM algorithm of Cao et al. (2000)
agg

Function to aggregate results from matrix to matrix
obj.tomogravity

Objective function of Zhang et al. 2003
tomogravity

Run tomogravity estimation on complete time series of aggregate flows
calibration_ssm

Estimation for the linear SSM calibration model of Blocker & Airoldi (2011)
buildRoutingMatrix

Build routing matrix from table of link relationships
getActive

Check for deterministically-known OD flows at single time
buildStarMat

Build routing matrix for star network topology
buildPrior

Construct prior from calibration model estimates
calcN

Compute total traffic from a particular time.
move_step

Move step of sample-resample-move algorithm for multilevel state-space model
buildRoutingMat

Build routing matrices for linked star topologies; that is, a set of star-topology networks with links between a subset of routers
tomogravity.fit

Tomogravity estimation for a single time point using L-BFGS-B
vardi.iteration

Execute single iteration for algorithm of Vardi (1996)
getSrcDstIndices

Find indices of source and destination for each point-to-point flow
strphour

Convert time string to decimal hour
Q_iid

Q function for locally IID EM algorithm of Cao et al. (2000)
abilene

Abilene data from Fang et al. (2007)
m_estep

Compute conditional expectations for EM algorithms of Cao et al. (2000)
diag_mat

Make diagonal matrix from vector
bell.labs

Bell Labs 1router data from Cao et al. (2000)
diag_ind

Make vector of 1-dimensional diagonal indices for square matrix
gravity.fit

Gravity estimation for a single time point
mle_filter

Filtering & smoothing at MLE for calibration SSM
thin

Thinning vector of indices for MCMC
gravity

Run tomogravity estimation on complete time series of aggregate flows
twMCMC

Function to run MCMC sampling for model of Tebaldi & West (1998)
R_estep

Compute conditional covariance matrix for EM algorithms of Cao et al. (2000)
phi_init

Simple initialization for phi in model of Cao et al. (2000)
vardi.compute.BS

Compute B and S matrices in algorithm of Vardi (1996)
decomposeA

Compute pivoted decomposition of routing matrix A into full-rank and remainder, as in Cao et al. 2000, via the QR decomposition.
grad_iid

Compute analytic gradient of Q-function for locally IID EM algorithm of Cao et al. (2000)