SIGNAL ANALYSIS METHODS

• US 20070244952A1
• Filed: 03/22/2007
• Published: 10/18/2007
• Est. Priority Date: 03/23/2006
• Status: Abandoned Application

• Current Assignee: RDA Pty., Ltd.
• Original Assignee: RDA Pty., Ltd.
• Inventors: Kennedy, Hugh
• Application Number:
  US11/689,830
• Publication Number:
  US 20070244952A1
• Time in Patent Office: Days
• US Class Current: 708/422
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/16 by matching signal segments

Abstract

The invention relates to a method of comparing two or more series of signal analysis data samples, to produce a measure of their similarity. Corresponding data samples of each series are compared, and measures of the magnitude of the signal and the variation between corresponding data samples are calculated. The magnitude measure may simply be the average power of corresponding data samples. The produced measure of the similarity of the series is a test statistic having Snedecor'"'"'s F distribution. When compared to typical cross correlation methods, a test statistic of the present invention provides a greater distinction between series displaying a high degree of similarity and series with a low degree of similarity. Accordingly, the present invention has applicability to the detection, classification and angular localisation of signals from acoustic or electromagnetic sources.

Claims

1. A method of assessing the similarity of two or more series of signal data samples comprising:

forming a test statistic which is inversely proportional to a variation measure, the variation measure indicating the variation between the series,

whereby the value of the test statistic therefore provides a measure of the similarity of the series.

2. A method according to claim 1, wherein the test statistic is proportional to a magnitude measure, the magnitude measure indicating the magnitude of the signal.

3. A method according to claim 1, further comprising:

calculating, for each data sample, a sample variation measure indicating the variation between corresponding data samples of each series, and

calculating the variation measure from the sample variation measures.

4. A method according to claim 1, wherein the variation measure is

$\sum _{}^{}$

n = 0 N - 1 

σ

^ n 2 ,

where N is the number of samples in each series, and

where {circumflex over (σ

)}_n is the observed standard deviation of the set of the nth samples of each series, at a specific lag.

5. A method according to claim 4, wherein the test statistic is proportional to

$\frac{\sum _{}^{}}{}$

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where {circumflex over (μ

)}_n is the observed mean of the nth samples of each series, at a specific lag.

6. A method according to claim 5, wherein the test statistic is equal to

$\left(M-1\right)$

∑

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where M is the number of series of data samples.

7. A method according to claim 1, wherein the number of series of data samples is more than two.

8. A method according to claim 1, wherein the test statistic has a known distribution.

9. A method according to claim 8, wherein the known distribution is Snedecor'"'"'s F distribution.

10. A method according to claim 1, further comprising:

obtaining at least one of the series of signal data samples from a sensor.

12. A test statistic according to claim 1, wherein the test statistic is proportional to a magnitude measure, the magnitude measure indicating the magnitude of the signal.

31. A computer readable medium encoded with data representing computer programs, that can be used to direct a programmable device to perform a method as claimed in claim 1.

32. A computer processing device, configured to perform the method of claim 1.

11. A test statistic for assessing the similarity of two or more series of signal data samples, wherein the test statistic is inversely proportional to a variation measure, the variation measure indicating the variation between the series.

13. A test statistic according to claim 11, wherein the variation measure is

$\sum _{}^{}$

n = 0 N - 1 

σ

^ n 2 ,

where N is the number of samples in each series, and

where {circumflex over (σ

)}_n is the observed standard deviation of the set of the nth samples of each series, at a specific lag.

14. A test statistic according to claim 13, wherein the test statistic is proportional to

$\frac{\sum _{}^{}}{}$

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where {circumflex over (μ

)}_n is the observed mean of the nth samples of each series, at a specific lag.

15. A method according to claim 14, wherein the test statistic is equal to

$\left(M-1\right)$

∑

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where M is the number of series of data samples.

16. A method of aligning two or more series of signal data samples comprising:

calculating a test statistic at a first lag combination, wherein the test statistic is inversely proportional to a variation measure, the variation measure indicating the variation between the series;

calculating the test statistic at one or more second lag combinations; and

selecting the lag combination for which the test statistic is greatest, thereby aligning the series.

17. A method according to claim 16, wherein there are three or more series of signal data samples.

18. A method according to claim 17, wherein each series is obtained from a different sensor, and the method is used for determining the angle of arrival of a signal relative to the three or more sensors, and wherein each lag combination corresponds to a possible angle of arrival of the signal.

19. A method according to claim 16, wherein the test statistic is proportional to a magnitude measure, the magnitude measure indicating the magnitude of the signal.

20. A method according to claim 16, wherein the calculation of the test statistic comprises:

calculating, for each data sample, a sample variation measure indicating the variation between corresponding data samples of each series, and

forming the variation measure from the sample variation measures.

21. A method according to claim 20, wherein the test statistic is proportional to

$\frac{\sum _{}^{}}{}$

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where {circumflex over (μ

)}_n is the observed mean of the nth samples of each series, at a specific lag.

22. A method according to claim 21, wherein the test statistic is equal to

$\left(M-1\right)$

∑

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where M is the number of series of data samples.

33. A computer readable medium encoded with data representing computer programs, that can be used to direct a programmable device to perform a method as claimed in claim 16.

35. A computer processing device, configured to perform the method of claim 16.

23. A method of detecting a signal of interest by:

obtaining a first series of signal data samples from a first sensor;

obtaining a second series of signal data samples from a second sensor;

calculating, from the first and second series, a test statistic which provides a measure of the similarity between the first and second series at a first lag of zero or more data samples;

calculating the test statistic at one or more further lags of zero or more data samples; and

declaring a detection if the largest calculated test statistic is more extreme than a threshold value.

24. A method according to claim 23, wherein the test statistic is inversely proportional to a variation measure, the variation measure indicating the variation between the series.

25. A method according to claim 23, wherein the test statistic is proportional to a magnitude measure, the magnitude measure indicating the magnitude of the signal.

26. A method according to claim 23, wherein the variation measure is

$\sum _{}^{}$

n = 0 N - 1 

σ

^ n 2 ,

where N is the number of samples in each series, and

where {circumflex over (σ

)}_n is the observed standard deviation of the set of the nth samples of each series, at a specific lag.

27. A method according to claim 26, wherein the test statistic is proportional to

$\frac{\sum _{}^{}}{}$

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where {circumflex over (μ

)}_n is the observed mean of the nth samples of each series, at a specific lag.

28. A method according to claim 27, wherein the test statistic is equal to

$\left(M-1\right)$

∑

n = 0 N - 1 

μ

^ n 2 ∑

n = 0 N - 1 

σ

^ n 2 ,

where M is the number of series of data samples.

29. A method according to claim 23, further comprising:

obtaining at least one further series of signal data samples from at least one further sensor,

wherein the test statistic provides a measure of the similarity between the first, second and further series, and the test statistic is calculated at a first combination of lags of zero or more data samples, and at one or more further combinations of lags of zero or more data samples, and

and wherein each combination of lags relates to an angle of arrival for the signal relative to the first, second and further sensors.

30. A method according to claim 29, wherein the angle of arrival of the signal of interest is determined by identifying the combination of lags at which the test statistic is highest.

34. A computer readable medium encoded with data representing computer programs, that can be used to direct a programmable device to perform a method as claimed in claim 23.

36. A computer processing device, configured to perform the method of claim 23.