2002Unpublished venueRequires access

New properties for discrete, bilinear time-frequency distributions

J.C. O'Neill, William J. Williams

Open publisher page 11 citations

Abstract

The most straightforward discrete implementation of bilinear time-frequency distributions (TFDs) are aliased in frequency, so as a result many different discrete TFDs have been developed. The traditional viewpoint in constructing discrete TFDs is that they should be samples of continuous TFDs. This paper introduces a second viewpoint: that discrete TFDs are inherently different, and thus, while having the same general goal, will not emulate continuous TFDs exactly. This paper develops properties corresponding to the second viewpoint for three types of discrete TFDs. A class of TFDs that satisfies the appropriate properties is also presented for each of the three types of discrete distributions. One of these is the class of alias-free generalized discrete time time-frequency distributions (AF-GDTFDs). It can be shown that the three classes of discrete TFDs presented include every time and frequency shift covariant, bilinear TFD, so can be considered to be Cohen's (1989) classes for discrete signals.

About this research paper

What this paper is about

The most straightforward discrete implementation of bilinear time-frequency distributions (TFDs) are aliased in frequency, so as a result many different discrete TFDs have been developed. The traditional viewpoint in constructing discrete TFDs is that they should be samples of continuous TFDs. This paper introduces a second viewpoint: that discrete TFDs are inherently different, and thus, while having the same general goal, will not emulate continuous TFDs exactly. This paper develops properties corresponding to the second viewpoint for three types of discrete TFDs. A class of TFDs that satisfies the appropriate properties is also presented for each of the three types of discrete distributions. One of these is the class of alias-free generalized discrete time time-frequency distributions (AF-GDTFDs). It can be shown that the three classes of discrete TFDs presented include every time and frequency shift covariant, bilinear TFD, so can be considered to be Cohen's (1989) classes for discrete signals.

Why it matters

OpenAlex reports 11 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

The most straightforward discrete implementation of bilinear time-frequency distributions (TFDs) are aliased in frequency, so as a result many different discrete TFDs have been developed. The traditional viewpoint in constructing discrete TFDs is that they should be samples of continuous TFDs. This paper introduces a second viewpoint: that discrete TFDs are inherently different, and thus, while having the same general goal, will not emulate continuous TFDs exactly. This paper develops properties corresponding to the second viewpoint for three types of discrete TFDs. A class of TFDs that satisfies the appropriate properties is also presented for each of the three types of discrete distributions. One of these is the class of alias-free generalized discrete time time-frequency distributions (AF-GDTFDs). It can be shown that the three classes of discrete TFDs presented include every time and frequency shift covariant, bilinear TFD, so can be considered to be Cohen's (1989) classes for discrete signals.

Key concepts: Bilinear interpolation, Discrete time and continuous time, Time–frequency analysis, Computer science, Class (philosophy), Algorithm, Mathematics, Applied mathematics

Related papers

Back to paper searchBrowse research topicsOriginal source
New properties for discrete, bilinear time-frequency distributions — Research Paper | ScholarLens