This is where navigation should be.


GSP_JTV_ESTIMATE_PSD - Estimate the PSD of a time vertex process

Usage

psd = gsp_jtv_estimate_psd(G, X)
psd = gsp_jtv_estimate_psd(G, X, param)

Input parameters

G Graph
X Signal(s) (matrix of NxTxR)
param Optional parameters

Output parameters

psd PSD matrix
ft Filter type

Description

This function estimates the PSD for vertex time processes. It is a generalization of the Bartlett method.

Additional parameters

  • param.estimator : Method used for the estimation. By default 'TA' - 'TA': Average over the time domain. - 'FC': Fourier convolution a trick used to reduce the computation. - 'VA': Average over the vertex domain - 'TVA': Average over the time and the vertex domain - 'FA' : Average on multiple realization in the joint spectral domain
  • param.L : Length of the window for the 'GSTFT' method
  • param.kernel : Kernel for the convolution for 'FC' method.

References:

N. Perraudin and P. Vandergheynst. Stationary signal processing on graphs. arXiv preprint arXiv:1601.02522, 2016.