System and method for waveform classification and characterization using multidimensional higher-order statistics
View Patent ↗A receiver exploits unique higher order statistics of temporally dependent waveforms to classify, characterize, identify (fingerprint) and intercept a waveform within the field of view of the receiver. The method uses 4 th -order non zero lag auto cumulants of the received waveform and 4 th -order non zero lag auto cumulants of known waveforms to classify and characterize the signal. The receiver includes a multi-element array and does not need a priori knowledge of the transmitted signal source obtain a fingerprint.
1. A method for obtaining the fingerprint of a received waveform comprising the steps of:
(a) sampling the received waveform to thereby obtain a predetermined number of samples of an attribute of the received waveform; and
(b) estimating a multidimensional higher-order nonzero-lag auto-cumulant of the received waveform attribute samples to thereby obtain the fingerprint of the received waveform.
2. The method of claim 1 wherein the attribute is a random process.
3. The method of claim 2 wherein the random process is stationary with statistical properties up to order N+1 that are invariant to shift in the origin over the sample observations for a fingerprint of dimension N.
4. The method of claim 2 wherein the random process is zero-mean.
5. The method of claim 2 wherein the attribute is selected from the group consisting of phase, frequency, amplitude, power spectral density, symbol rate, symbol energy, bandwidth, modulation, phase jitter, timing jitter, signal constellation, pulse shape and frequency offset.
6. The method of claim 1 wherein the auto-cumulant is a fourth order cumulant.
7. The method of claim 1 wherein the auto-cumulant is determined as a function of three lags.
8. The method of claim 1 wherein the fingerprint is used to classify the received waveform.
9. The method of claim 1 wherein the fingerprint is used to characterize the received waveform.
10. The method of claim 1 wherein the fingerprint is used to identify the received waveform.
11. The method of claim 1 wherein the fingerprint is used to intercept the received waveform.
12. The method of claim 1 further comprising the steps of:
(c) comparing the estimated auto-cumulant of the received waveform attribute samples with a multidimensional higher-order nonzero-lag auto-cumulant of a known waveform to thereby obtain the fingerprint of the received waveform.
13. The method of claim 12 wherein the auto-cumulant of the known waveform is obtained from a database.
14. The method of claim 12 wherein the auto-cumulant of the known waveform is obtained from a lookup table.
15. A method for classifying a received waveform comprising the steps of:
(a) sampling the received waveform to thereby obtain a predetermined number of samples of an attribute of the received waveform;
(b) estimating a multidimensional higher-order nonzero-lag auto-cumulant of the received waveform attribute samples; and
(c) comparing the estimated auto-cumulant of the received waveform attribute samples with a multidimensional higher-order nonzero-lag auto-cumulant of a known waveform to thereby classify the received waveform.
16. The method of claim 15 wherein the attribute is a random process.
17. The method of claim 16 wherein the random process is stationary with statistical properties up to order N+1 that are invariant to a shift in the origin over the sample observations for a fingerprint of dimension N.
18. The method of claim 16 wherein the random process is zero-mean.
19. The method of claim 16 wherein the attribute is selected from the group consisting of phase, frequency, amplitude, power spectral density, symbol rate, symbol energy, bandwidth, modulation, phase jitter, timing jitter, signal constellation, pulse shape and frequency offset.
20. The method of claim 15 wherein the auto-cumulant is a fourth order cumulant.
21. The method of claim 15 wherein the auto-cumulant is determined as a function of three lags.
22. The method of claim 15 wherein the received waveform is classified by modulation type.
23. The method of claim 22 wherein the modulation types are selected from the group consisting of PSK, PAM, and QAM.
24. The method of claim 23 wherein the PSK modulation types are selected from the group consisting of BPSK, QPSK, and 8PSK.
25. The method of claim 23 wherein the QAM modulation types are selected from the group consisting of 16-QAM, 64-QAM, and 256-QAM.
26. The method of claim 15 wherein the auto-cumulant of the known waveform is obtained from a database.
27. The method of claim 15 wherein the auto-cumulant of the known waveform is obtained from a lookup table.
28. A method for characterizing a received waveform comprising the steps of:
(a) sampling the received waveform to thereby obtain a predetermined number of samples of an attribute of the received waveform;
(b) estimating a multidimensional higher-order nonzero-lag auto-cumulant of the received waveform attribute samples; and
(c) comparing the estimated auto-cumulant of the received waveform attribute samples with a multidimensional higher-order nonzero-lag auto-cumulant of a known waveform to thereby characterize the received waveform.
29. The method of claim 28 wherein the attribute is a random process.
30. The method of claim 29 wherein the random process is stationary with statistical properties up to order N+1 that are invariant to a shift in the origin over the sample observations for a fingerprint of dimension N.
31. The method of claim 29 wherein the random process is zero-mean.
32. The method of claim 29 wherein the attribute is selected from the group consisting of phase, frequency, amplitude, power spectral density, symbol rate, symbol energy, bandwidth, modulation, phase jitter, timing jitter, pulse shape, signal constellation, and frequency offset.
33. The method of claim 28 wherein the auto-cumulant is a fourth order cumulant.
34. The method of claim 28 wherein the auto-cumulant is determined as a function of three lags.
35. The method of claim 28 wherein the known waveform is idealized so that the samples of the attribute do not contain anomalous variations.
36. The method of claim 35 wherein the received waveform is characterized by anomalous variations of the attribute in the samples as determined by the difference between the estimated auto-cumulant of the received waveform and the auto-cumulant of the idealized known waveform.
37. The method of claim 28 wherein the auto-cumulant of the known waveform is obtained from a database.
38. The method of claim 28 wherein the auto-cumulant of the known waveform is obtained from a lookup table.
39. A method for identifying a received waveform comprising the steps of:
(a) sampling the received waveform to thereby obtain a predetermined number of samples of an attribute of the received waveform;
(b) estimating a multidimensional higher-order nonzero-lag auto-cumulant of the received waveform attribute samples; and
(c) comparing the estimated auto-cumulant of the received waveform attribute samples with a multidimensional higher-order nonzero-lag auto-cumulant of a known waveform to thereby identify the received waveform.
40. The method of claim 39 wherein the attribute is a random process.
41. The method of claim 40 wherein the random process is stationary with statistical properties up to order N+1 that are invariant to a shift in the origin over the sample observations for a fingerprint of dimension N.
42. The method of claim 40 wherein the random process is zero-mean.
43. The method of claim 40 wherein the attribute is selected from the group consisting of phase, frequency, amplitude, power spectral density, symbol rate, symbol energy, bandwidth, modulation, phase jitter, timing jitter, pulse shape, signal constellation, and frequency offset.
44. The method of claim 40 wherein the auto-cumulant is a fourth order cumulant.
45. The method of claim 40 wherein the auto-cumulant is determined as a function of three lags.
46. The method of claim 40 wherein the auto-cumulant of the known waveform is obtained from a database.
47. The method of claim 40 wherein the auto-cumulant of the known waveform is obtained from a lookup table.
48. A method for intercepting a received waveform comprising the steps of:
(a) sampling the received waveform to thereby obtain a predetermined number of samples of an attribute of the received waveform;
(b) estimating a multidimensional higher-order nonzero-lag auto-cumulant of the received waveform attribute samples;
(c) comparing the estimated auto-cumulant of the received waveform attribute samples with a multidimensional higher-order nonzero-lag auto-cumulant of a known waveform to thereby determine the modulation of the received waveform; and
(d) processing the received waveform using the determined modulation to thereby intercept the received waveform.
49. The method of claim 48 wherein the attribute is a random process.
50. The method of claim 49 wherein the random process is stationary with statistical properties up to order N+1 that are invariant to a shift in the origin over the sample observations for a fingerprint of dimension N.
51. The method of claim 49 wherein the random process is zero-mean.
52. The method of claim 49 wherein the attribute is selected from the group consisting of phase, frequency, amplitude, power spectral density, symbol rate, symbol energy, bandwidth, modulation, phase jitter, timing jitter, pulse shape, signal constellation, and frequency offset.
53. The method of claim 48 wherein the auto-cumulant is a fourth order cumulant.
54. The method of claim 48 wherein the auto-cumulant is determined as a function of three lags.
55. The method of claim 48 wherein the auto-cumulant of the known waveform is obtained from a database.
56. The method of claim 48 wherein the auto-cumulant of the known waveform is obtained from a lookup table.
57. In a method for classifying a received signal from an unknown emitter using characteristics of the waveform of the received signal, the improvement comprising the steps of:
(a) determining for the received signal variations of 4 th -order cumulant multidimensional features over a plurality of lag combinations; and
(b) comparing the 4 th -order cumulant multidimensional features of the received signal with the variations of similar 4 th -order cumulant multidimensional features of a known reference signal to thereby classify the received signal.