Hyperbolic kernel for time-frequency power spectrum

Khoa N. Le, Kishor P. Dabke, Gregory K. Egan

Research output: Contribution to journalArticlepeer-review

19 Citations (Scopus)

Abstract

We propose a new family of hyperbolic kernels Φhyperbolic(θ, τ) = [sech(βθτ)]n, where n=1,3,5,⋯, for a joint time-frequency distribution. The first-order hyperbolic kernel sech(βθτ) is mainly considered. Theoretical aspects of the new hyperbolic kernel are examined in detail. The effectiveness of a kernel is determined by three factors: cross-term suppression, auto-term resolution, and noise robustness. The effectiveness of the new kernel is compared with other kernels including Choi-Williams, Wigner-Ville, and multiform tiltable exponential using two different signals: complex-exponential and chirp.
Original languageEnglish
Pages (from-to)2400-2415
Number of pages16
JournalOptical Engineering
Volume42
Issue number8
Publication statusPublished - Aug 2003

Keywords

  • kernel functions
  • noise
  • signal processing

Fingerprint

Dive into the research topics of 'Hyperbolic kernel for time-frequency power spectrum'. Together they form a unique fingerprint.

Cite this