Signal recognizer for communications signals

• US 6,690,746 B1
• Filed: 06/09/2000
• Issued: 02/10/2004
• Est. Priority Date: 06/11/1999
• Status: Expired due to Term

• Current Assignee: Southwest Research Institute
• Original Assignee: Southwest Research Institute
• Inventors: Sills, James A., Black, Q. Robert
• Primary Examiner(s): Tran, Khai
• Application Number:
  US09/591,014
• Time in Patent Office: 1,341 Days
• Field of Search: 375/316, 375/350, 375/323, 375/279, 375/261, 375/272
• US Class Current: 375/316
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: H04L 27/0012 arrangements for identifying the type of modulation

Abstract

A signal recognizer capable of classifying any of all commonly used communications signals. The recognizer has a number of modules that may operate in parallel, each module associated with a different signal type. Each module determines signal parameters by first estimating one or more parameters of a detected signal of interest. The estimated parameter(s) is then used as the basis for demodulating the input signal. The demodulated symbols are used for hypothesis testing, during which the module decides on a candidate signal type for that module. Each module subjects its candidate signal type to a “false alarm” test, which evaluates the likelihood that the signal is not the signal type associated with the module. The resulting confidence data is collected and analyzed to determine a best candidate signal type from among all the modules, together with its signal parameters.

Claims

1. A computer-implemented signal recognizer for classifying noncooperative signals, comprising:

an up/down detector for detecting the presence of a noncooperative signal of interest;

a signal classifier having a number of classifier modules, each module associated with a different signal modulation type and each module operable to perform the following tasks;

to receive said signal of interest in digital form, to estimate at least one parameter of said signal of interest, to demodulate said signal of interest based on the estimated parameter, to determine a candidate signal type having the estimated parameter, and to calculate confidence data representing the extent to which said signal of interest is likely of said candidate signal type versus not of that signal type; and

a confidence analyzer for receiving said confidence data from each said module and for determining a best signal type from the candidate signal types determined by said modules.

2. The recognizer of claim 1, wherein said candidate signal type has a parameter having a finite number of possibilities, each associated with one of a number of hypotheses, and wherein said signal classifier determines said candidate signal type by reducing said hypotheses to a single hypothesis on the basis of at least one estimated parameter.

3. The recognizer of claim 2, wherein said signal of interest is a PSK/QAM signal and said hypotheses represent PSK/QAM configurations.

4. The recognizer of claim 2, wherein said signal of interest is an FSK signal and said hypotheses represent signal levels.

5. The recognizer of claim 2, wherein said signal of interest is an analog signal and said hypothesis represent analog signal types.

6. The recognizer of claim 1, wherein said signal of interest is an MSK signal and wherein an estimated parameter is baud rate.

7. The recognizer of claim 1, wherein said signal of interest is an MSK signal and wherein an estimated parameter is frequency.

8. The recognizer of claim 1, wherein said signal of interest is an OOK signal and wherein an estimated parameter is words per minute.

9. The recognizer of claim 1, wherein said signal of interest is an OOK signal and wherein an estimated parameter is frequency.

10. The recognizer of claim 1, wherein said signal of interest is an OOK signal and wherein an estimated parameter is bandwidth.

11. The recognizer of claim 1, wherein said signal of interest is an FSK signal and wherein an estimated parameter is tone frequency.

12. The recognizer of claim 1, wherein said signal of interest is a PSK/QAM signal and wherein an estimated parameter is signal to noise ratio.

13. The recognizer of claim 1, wherein said signal of interest is an analog signal and wherein an estimated parameter is center frequency.

14. The recognizer of claim 2, wherein said signal classifier determines said candidate signal type by calculating probability density functions for each hypothesis.

15. The recognizer of claim 1, wherein said signal classifier demodulates said signal of interest using an adaptive equalization process.

16. The recognizer of claim 1, wherein said signal classifier demodulates said signal of interest using matched filtering techniques.

17. The recognizer of claim 1, wherein said signal classifier demodulates said signal of interest using a symbol timing recovery process based on squaring said signal of interest and filtering with a narrow bandpass filter.

18. The recognizer of claim 1, wherein said signal classifier demodulates said signal of interest using an interpolation process operable to mitigate interference from noncoherent sampling.

19. The recognizer of claim 1, wherein said signal classifier demodulates said signal of interest using a carrier recovery process based on principle component analysis.

20. The recognizer of claim 1, further comprising the step of using said candidate signal type as the basis for additional parameter estimation calculations.

37. The method of claim 20, wherein each said module demodulates said signal of interest using a symbol timing recovery process based on squaring said signal of interest and filtering with a narrow bandpass filter.

38. The method of claim 20, wherein each said module demodulates said signal of interest using an interpolation process operable to mitigate interference from noncoherent sampling.

39. The method of claim 20, further comprising the step of using said candidate signal type as the basis for additional parameter estimation calculations.

21. A method of classifying noncooperative signals, comprising:

detecting the presence of a noncooperative signal of interest;

processing said signal of interest with a number of signal classifier modules, each module associated with a different signal modulation type and each module operable to perform the following tasks;

to receive said signal of interest in digital form, to estimate at least one parameter of said signal of interest, to demodulate said signal of interest based on the estimated parameter, to determine a candidate signal type having the estimated parameter, and to calculate confidence data representing the extent to which said signal of interest is likely of said candidate signal type versus not of that signal type; and

receiving said confidence data from each said module and determining a best signal type from the candidate signal types determined by said modules.

22. The method of claim 21, wherein said candidate signal type has a parameter having a finite number of possibilities, each associated with one of a number of hypotheses, and wherein said signal classifier determines said candidate signal type by reducing said hypotheses to a single hypothesis on the basis of at least one estimated parameter.

23. The method of claim 22, wherein said signal of interest is a PSK/QAM signal and said hypotheses represent PSK/QAM configurations.

24. The method of claim 22, wherein said signal of interest is an FSK signal and said hypotheses represent signal levels.

25. The method of claim 22, wherein said signal of interest is an analog signal and said hypothesis represent analog signal types.

26. The method of claim 21, wherein said signal of interest is an MSK signal and wherein an estimated parameter is baud rate.

27. The method of claim 21, wherein said signal of interest is an MSK signal and wherein an estimated parameter is frequency.

28. The method of claim 21, wherein said signal of interest is an OOK signal and wherein an estimated parameter is words per minute.

29. The method of claim 21, wherein said signal of interest is an OOK signal and wherein an estimated parameter is frequency.

30. The method of claim 21, wherein said signal of interest is an OOK signal and wherein an estimated parameter is bandwidth.

31. The method of claim 21, wherein said signal of interest is an FSK signal and wherein an estimated parameter is tone frequency.

32. The method of claim 21, wherein said signal of interest is a PSK/QAM signal and wherein an estimated parameter is signal to noise ratio.

33. The method of claim 21, wherein said signal of interest is an analog signal and wherein an estimated parameter is center frequency.

34. The method of claim 22, wherein said each said module determines said candidate signal type by calculating probability density functions for each hypothesis.

35. The method of claim 21, wherein each said module demodulates said signal of interest using an adaptive equalization process.

36. The method of claim 21, wherein each said module demodulates said signal of interest using matched filtering techniques.