Constant false alarm rate robust adaptive detection using the fast fourier transform

• US 8,194,766 B2
• Filed: 05/22/2009
• Issued: 06/05/2012
• Est. Priority Date: 05/22/2009
• Status: Active Grant

• Current Assignee: Aerospace Corporation
• Original Assignee: Aerospace Corporation
• Inventors: Kun, David S., Morgan, Neil
• Primary Examiner(s): Payne, David C.
• Assistant Examiner(s): Cadeau, Wednel
• Application Number:
  US12/471,191
• Publication Number:
  US 20100296565A1
• Time in Patent Office: 1,110 Days
• Field of Search: 375/145, 375/227, 375/350, 375/130, 375/340, 375/346, 375/317, 375/354, 375/316
• US Class Current: 375/260
• CPC Class Codes: G01S 7/021 Auxiliary means for detecting or identifying radar signals or the like, e.g. radar jamming signals
• CPC Class Codes: G01S 7/52001 Auxiliary means for detecting or identifying sonar signals or the like, e.g. sonar jamming signals

Abstract

A signal detection apparatus includes electronics configured to receive an RF input and to process the RF input applying a decision rule that is a normalized quantity to determine if a man-made signal is present or if only noise is present while achieving a constant false alarm rate.

Claims

1. A signal detection apparatus, comprising:

electronics configured to receive an RF input and to process the RF input applying a decision rule that utilizes Fast Fourier Transforms (FFTs) of non-overlapping subsequences for a window function to generate K decision statistics whose cumulative distribution function (cdf) yield, independent of the noise variance and the FFT window type, a numerical value for the threshold, τ

, of each decision statistic to determine if a man-made signal is present or to achieve a constant false alarm rate when only noise is present;

wherein each of the decision statistics, denoted as C_M(l) for l ε

κ

, has a cumulative distribution function (cdf), which is equal to F_CM (τ

), as follows

2. The signal detection apparatus of claim 1, wherein the decision rule computes a normalized power ratio.

3. The signal detection apparatus of claim 1, wherein the decision rule immediately adjusts to power fluctuations due to power normalization.

4. The signal detection apparatus of claim 1, wherein the decision rule represents examining the ratio of the power of the RF input within a single frequency bin to the total power of the RF input over the frequency band of interest.

5. The signal detection apparatus of claim 1, wherein the decision rule determines an exact value of a decision statistic threshold, rather than upper and lower bounds, by analytically determining the probability distribution of the decision statistics.

6. The signal detection apparatus of claim 1, wherein the decision statistic threshold ranges from zero to one.

7. The signal detection apparatus of claim 1, wherein the electronics are configured to employ the Fast Fourier Transform (FFT) to detect the presence of man-made signals within a frequency band of interest without the need for implementing any calibration or noise power estimation techniques.

8. The signal detection apparatus of claim 1, wherein the decision statistic threshold is independent of noise variance, FFT window type, and the statistics of the environment.

9. A signal detection apparatus, comprising:

a tunable bandpass filter that filters an RF input to generate a bandpass filter output signal;

an IF double balanced mixer that downconverts the bandpass filter output signal to complex baseband;

lowpass filters that receive outputs of the IF double balanced mixer, the lowpass filters being configured to act as anti-alias filters for I and Q channels and to output an IF waveform;

analog-to-digital converters (ADCs) configured to sample the IF waveform into digital samples; and

electronics configured to receive the digital samples and implement a decision rule that utilizes Fast Fourier Transforms (FFTs) of non-overlapping subsequences for a window function to generate K decision statistics whose cumulative distribution function (cdf) yield, independent of the noise variance and the FFT window type, a numerical value for the threshold, τ

, of each decision statistic to determine if a man-made signal is present or to achieve a constant false alarm rate when only noise is present;

wherein each of the decision statistics, denoted as C_M (l) for l ε

κ

, has a cumulative distribution function (cdf), which is equal to F_CM (τ

), as follows

10. The signal detection apparatus of claim 9, wherein the lowpass filters are configured such that input power is filtered outside an anti-aliasing filter cutoff frequency.

11. The signal detection apparatus of claim 9, wherein the electronics include a field-programmable gate array (FPGA).

12. The signal detection apparatus of claim 9, wherein the decision rule computes a normalized power ratio.

13. The signal detection apparatus of claim 9, wherein the decision rule immediately adjusts to power fluctuations due to power normalization.

14. The signal detection apparatus of claim 9, wherein the decision rule represents examining the ratio of the power of the RF input within a single frequency bin to the total power of the RF input over the frequency band of interest.

15. The signal detection apparatus of claim 9, wherein the decision rule determines an exact value of a decision statistic threshold, rather than upper and lower bounds, by analytically determining the probability distribution of the decision statistics.

16. The signal detection apparatus of claim 9, wherein the decision statistic threshold ranges from zero to one.

17. The signal detection apparatus of claim 9, wherein the electronics are configured to employ the Fast Fourier Transform (FFT) to detect the presence of man-made signals within a frequency band of interest without the need for implementing any calibration or noise power estimation techniques.

18. The signal detection apparatus of claim 9, wherein the decision statistic threshold that is independent of noise variance, FFT window type, and the statistics of the environment.