Extracting spectral features from a signal in a multiplicative and additive noise environment

Patent metadata

Field	Value
Publication number	US-9239372-B2
Application number	US-201213613870-A
Country	US
Kind code	B2
Filing date	Sep 13, 2012
Priority date	Sep 13, 2012
Publication date	Jan 19, 2016
Grant date	Jan 19, 2016

Abstract

A spectral feature extracting engine, chirp detecting engine, and target classifying engine with corresponding method, system, and computer product are provided. Given a “noisy” signal having multiplicative and additive noise, the spectral feature extracting engine extracts spectral features from the noisy signal in the form of a spectral detection density. The extraction includes identifying an initial detection by comparing the spectral content of a subject time-frequency slice with an initial detection threshold calculated from a set of time-frequency slices, and determining the spectral detection density as a function of a number of identified initial detections and number of compared time-frequency slices. Based on the spectral detection density of the noisy signal, the chirp detecting engine detects multiple channel-induced and target-induced chirps present in the noisy signal and the target classifying engine classifies a target from the noisy signal.

First claim

What is claimed is:

1. An improved method of operating a RADAR or wireless communications signal processing apparatus implemented as either one or more programmable processors or as special purpose logic circuitry for detecting channel-induced and target-induced chirps in a signal or for classifying a target from a signal using spectral feature extraction, the method comprising:
operating a spectral feature extracting engine of the apparatus to:
remove direct current bias from a RADAR or wireless communication signal captured by a receiver, the signal having impairments including multiplicative and additive noise, in order to produce a bias-free signal;
transform the bias-free signal into a time-frequency spectrum;
slice the time-frequency spectrum across time into time-frequency slices, each of the time-frequency slices having spectral content representing a measure of power spectral density of the bias-free signal at a given point in time;
compute an initial detection threshold associated with the given signal from the time-frequency slices;
compare, for at least some of the time-frequency slices the spectral content of a subject time-frequency slice with the computed initial detection threshold to identify an initial detection;
determine a spectral detection density as a function of a number of initial detections being identified and number of time-frequency slices being compared; and
operating a chirp detecting engine of the apparatus to detect channel-induced and target-induced chirp(s) present in the given signal based on the determined spectral detection density or, a target classifying engine of the apparatus to classify a target from the given signal based on the determined spectral detection density, thereby improving an ability to detect the channel-induced and target-induced chirps in the RADAR or the wireless communication signal impaired with multiplicative and additive noise.

2. The method of claim 1 wherein transforming the bias-free signal includes using any one of Short-Time Fourier Transform, and Wignerville Transform, to transform the given signal.

3. The method of claim 1 wherein computing the initial detection threshold includes computing any order moment of the spectral content of the time-frequency slices measured in power, log-power, magnitude, or combinations thereof.

4. The method of claim 1 wherein comparing includes identifying the initial detection when the spectral content of the subject time-frequency slice exceeds the computed initial detection threshold.

5. The method of claim 1 wherein determining the spectral detection density includes scaling the number of initial detections being identified by a time-bandwidth product of a given number of time-frequency slices.

6. The method of claim 1 wherein detecting channel-induced and target-induced chirp(s) includes comparing the spectral detection density to a threshold indicative of one or more chirps being present in the given signal.

7. The method of claim 1 wherein classifying the target includes comparing the determined spectral detection density with spectral signatures obtained from training data.

8. The method of claim 7 wherein classifying the target includes comparing the determined spectral detection density over time with the spectral signatures.

9. The method of claim 1 further comprising:
given one or more chirps detected based on the spectral detection density of the given signal, associating initial detections occurring over a given span of time together; and
curve fitting the associated initial detections to determine one or more chirp estimates.

10. The method of claim 9 further comprising demodulating the one or more detected chirps from the given signal using the chirp estimates.

11. A RADAR or wireless communications signal processing apparatus for detecting channel-induced and target-induced chirps in a signal or for classifying a target from a signal using spectral feature extraction, the apparatus comprising:
a hardware memory having computer executable instructions stored thereupon; and
one or more programmable processors or special purpose logic circuitry communicatively coupled to the hardware memory and configured to execute the instructions to:
remove direct current bias from a RADAR or wireless communication signal captured by a receiver, the signal having impairments including multiplicative and additive noise, in order to produce a bias-free signal;
transform the bias-free signal into a time-frequency spectrum;
slice the time-frequency spectrum across time into time-frequency slices, each of the time-frequency slices having spectral content representing a measure of power spectral density of the bias-free signal at a given point in time;
compute an initial detection threshold associated with the given signal from the time-frequency slices;
compare, for at least some of the time-frequency slices; the spectral content of a subject time-frequency slice with the computed initial detection threshold to identify an initial detection;
determine a spectral detection density as a function of a number of initial detections being identified and number of time-frequency slices being compared; and
either detect channel-induced and target-induced chirp(s) present in the given signal based on the determined spectral detection density, or classify a target from the given signal based on the determined spectral detection density, thereby improving an ability to detect the channel-induced and target-induced chirps in the RADAR or the wireless communication signal impaired with multiplicative and additive noise.

12. A non-transitory computer-readable storage medium encoded with instructions that when executed by one ore more programmable processors of a RADAR or wireless communications signal processing apparatus cause the apparatus to:
remove direct current bias from a RADAR or wireless communication signal captured by a receiver, the signal having impairments including multiplicative and additive noise, in order to produce a bias-free signal;
transform the bias-free signal into a time-frequency spectrum;
slice the time-frequency spectrum across time into time-frequency slices, each of the time-frequency slices having spectral content representing a measure of power spectral density of the bias-free signal at a given point in time;
compute an initial detection threshold associated with the given signal from the time-frequency slices;
compare, for at least some of the time-frequency slices, the spectral content of a subject time-frequency slice with the computed initial detection threshold to identify an initial detection;
determine a spectral detection density as a function of a number of initial detections being identified and number of time-frequency slices being compared; and
either detect channel-induced and target-induced chirp(s) present in the given signal based on the determined spectral detection density, or classify a target from the given signal based on the determined spectral detection density, thereby improving an ability to detect the channel-induced and target-induced chirps in the RADAR or the wireless communication signal impaired with multiplicative and additive noise.

Assignees

• Maalouli Ghassan C
• Young Brett J
• Raytheon Co

Inventors

• Maalouli Ghassan C
• Young Brett J

Classifications

• G01S7/41

  using analysis of echo signal for target characterisation; Target signature; Target cross-section · CPC title

• G01S7/4004 Primary

  of parts of a radar system · CPC title

• G01S7/414 Primary

  Discriminating targets with respect to background clutter · CPC title

• G01S13/02

  Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems · CPC title