Signal classification methods for scanning receiver and other applications

• US 7,035,593 B2
• Filed: 04/22/2004
• Issued: 04/25/2006
• Est. Priority Date: 07/28/2003
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cognio, Inc. (Cisco Systems, Inc.)
• Inventors: Miller, Karl A., Mason, David F.
• Primary Examiner(s): Nguyen, Simon
• Application Number:
  US10/830,414
• Publication Number:
  US 20050032479A1
• Time in Patent Office: 733 Days
• Field of Search: 455/67.11, 455/226.1, 375/340, 375/343
• US Class Current: 455/67.11
• CPC Class Codes: G01R 29/02 Measuring characteristics of individual pulses, e.g. deviation from pulse flatness, rise time or duration
• CPC Class Codes: H04W 16/14 Spectrum sharing arrangements between different networks

Abstract

Signal classification techniques for classifying signals occurring in a frequency band using a plurality of classifier procedures each dedicated to identify a particular signal or signal type. The classification procedures operate on spectrum activity data that may include pulse event data describing particular types of signal pulses occurring in the frequency band, power versus frequency data for sampling intervals of activity in a frequency band and/or raw analog-to-digital converter samples take of a received signal. The signal classification techniques are useful to identify wireless radio signals or other radio emissions occurring in an unlicensed radio frequency band. One type of classification procedure is a pulse timing template that is compared against accumulated signal pulse data to determine occurrence of a particular signal type. To enhance the pulse timing template signal classifier, techniques are provided to examine the center frequency distribution of signal pulses that match the timing template in order to confirm that certain signal pulses are associated with a particular frequency hopping signal type. This is particularly useful when the accumulated signal pulse data is obtained from a scanning radio receiver that is tuned to different portions/channels of the frequency band in which the frequency hopping signal is expected to hop.

Claims

1. A method for classifying signals occurring in a frequency band, comprising:

a. generating signal pulse data by computing spectral information representing the radio frequency energy and comparing the spectral information with a set of signal pulse characteristics that comprises ranges for center frequency, duration and bandwidth, to produce signal pulse data for pulses that meet the set of signal pulse characteristics;

b. accumulating over a time interval signal pulse data for signal pulses detected in the frequency band, wherein the accumulated signal pulse data comprises the start time, center frequency, bandwidth and duration for each detected signal pulse; and

c. executing against the signal pulse data a plurality of classification procedures to classify signals occurring in the frequency band, wherein each classification procedure is associated with a particular signal type.

2. The method of claim 1, wherein the step of executing comprises executing each of the plurality of the classification procedures against the accumulated signal pulse data.

3. The method of claim 2, and further comprising the step of designating for one or more signal pulses in the accumulated signal pulse data an identifier corresponding to the classification procedure that achieves a match to the one or more signal pulses.

4. The method of claim 2, wherein the step of executing each of the plurality of classification procedures comprises determining multiple instances of the same signal type.

5. The method of claim 1, wherein the step of accumulating comprises accumulating signal pulse data for a plurality of time intervals, and wherein the step of executing comprises executing each of the plurality of classification procedures against the accumulated signal pulse data for each time interval, one time interval at a time.

6. The method of claim 5, wherein the step of executing comprises executing a classification procedure that compares a timing template describing at least one time period between signal pulses that satisfy defined ranges of bandwidth, duration and center frequency against the accumulated data to classify a particular signal occurring in the frequency band.

7. The method of claim 6, and further comprising the step of tracking the number of matches and misses of signal pulses with respect to the timing template.

8. The method of claim 7, and further comprising the step of declaring that a particular signal is classified when the number of matches of signal pulses with respect to a timing template achieves a first threshold and subsequently determining that the number of matches of signal pulses with respect to the timing template achieves a second threshold, wherein the second threshold is less rigorous than the first threshold.

9. The method of claim 8, wherein the step of generating data comprises generating data for received radio frequency energy as a receiver is scanned across the frequency band and dwells at a particular frequency channel for a period of time.

10. The method of claim 6, and further comprising the step of tracking the center frequency of each signal pulse that satisfies the timing template.

11. The method of claim 10, and further comprising the step of analyzing a distribution of center frequency for signal pulses that satisfy the timing template in order for the classification procedure to declare a match.

12. The method of claim 11, wherein one of the plurality of classification procedures comprises a timing template that analyzes signal pulse data to determine occurrence of Bluetooth™

ACL signals.

13. The method of claim 6, and further comprising the step of adjusting the timing template to account for drift over time between the template and the signal pulses that have been determined to match the template.

14. The method of claim 13, and further comprising the step of estimating the drift between the signal pulses and the template.

15. The method of claim 1, and further comprising the step of accumulating one or more histograms from the accumulated signal pulse data that represent trends over time for the center frequency, bandwidth, duration and time between signal pulses detected in the frequency band.

16. The method of claim 15, wherein the step of accumulating one or more histograms comprises building one or more histograms selected from the group consisting of:

a center frequency histogram that tracks a percentage of time a given center frequency was observed for detected signal pulses, a bandwidth histogram that tracks the percentage of time a given bandwidth was observed for detected signal pulses, a pulse duration histogram that tracks a percentage of time a given duration or durations was observed for detected signal pulses, a time between pulses histogram that tracks a percentage of time that a given time duration or durations was observed between signal pulses and a number of active transmissions histogram that tracks when several different signal pulses simultaneously occur.

17. The method of claim 15, and further comprising the step of selecting a subset of the plurality of classification procedures that correspond to signals that are likely occurring in the frequency band based on the histograms, and wherein the step of executing comprises executing the subset of plurality of classification procedures against the data.

18. The method of claim 1, wherein the step of generating data comprises capturing baseband modulated digital samples associated with received radio frequency energy.

19. The method of claim 18, wherein the step of executing comprises executing a classification procedure that compares a template of baseband modulated digital samples for a portion of a signal pulse for a known signal against the captured baseband modulated digital samples.

20. A method for classifying signals occurring in a frequency band, comprising:

a. generating data for one or more attributes of radio frequency energy received in the frequency band over time;

b. executing against the data a plurality of classification procedures to classify signals occurring in the frequency band, wherein each classification procedure is associated with a particular signal type; and

c. wherein the step of executing comprises executing a classification procedure that compares a power template describing power versus frequency characteristics against power versus frequency data derived from the received radio frequency energy to classify a signal occurring in the frequency band.

21. The method of claim 20, wherein the step of executing comprises executing a classification procedure that compares each of a plurality of power templates describing power versus frequency characteristics against the power versus frequency data, wherein a particular signal is declared to be occurring when the power versus frequency data matches any one of the plurality of power templates.

22. The method of claim 20, wherein the step of executing comprises executing a classification procedure that compares each of the plurality of power templates against power versus frequency data for multiple sampling intervals, and wherein the particular signal is determined to be occurring when the power versus frequency data matches at least one of the plurality of power templates for each of a predetermined number of time intervals of power versus frequency data.

23. The method of claim 20, wherein the step of executing comprises executing a classification procedure that compares each of a plurality of power templates against power versus frequency data for multiple time intervals, and wherein the particular signal is determined to be occurring when the power versus frequency data matches all of the power templates over the multiple time intervals.

24. The method of claim 23, wherein the step of executing comprises executing a classification procedure that requires a match to a different one of the plurality of power templates during each of the multiple time intervals.

25. The method of claim 20, wherein the step of executing a classification procedure comprises comparing a power template describing average power versus frequency characteristics against average power versus frequency data for a time interval derived from the radio frequency energy occurring in the frequency band.

26. The method of claim 20, wherein the step of executing comprises comparing a power template describing maximum power versus frequency characteristics against maximum power versus frequency data for a time interval derived from the radio frequency energy occurring in the frequency band.

27. The method of claim 20, wherein the step of executing comprises executing a classification procedure that compares a template of Fast Fourier Transform (FFT) data for a portion of a known signal pulse against spectral information for a portion of a signal pulse determined to occur in the frequency band.

28. The method of claim 27, wherein the step of executing comprises comparing a template describing spectral information for a beginning portion of the known signal against FFT data for the beginning portion of a detected signal pulse.

29. A processor readable medium encoded with instructions that, when executed by a processor, cause the processor to classify signals occurring in a frequency band, by:

a. generating signal pulse data by computing spectral information representing the radio frequency energy and comparing the spectral information with a set of signal pulse characteristics that comprises ranges for center frequency, duration and bandwidth to produce signal pulse data for pulses that meet the set of signal pulse characteristics;

b. accumulating over a time interval signal pulse data for signal pulses detected in the frequency band, wherein the accumulated signal pulse data comprises the start time, center frequency, bandwidth and duration for each detected signal pulse; and

c. executing against the signal pulse data a plurality of classification procedures to classify signals occurring in the frequency band, wherein each classification procedure is associated with a particular signal type.

30. The processor readable medium of claim 29, and further comprising instructions encoded on the medium, that when executed by the processor, cause the processor to designate for one or more signal pulses in the accumulated signal pulse data an identifier corresponding to the classification procedure that achieves a match to the one or more signal pulses.

31. The processor readable medium of claim 29, and further comprising instructions encoded on the medium, that when executed by the processor, cause the processor to executing each of the plurality of classification procedures comprises detect multiple instances of the same signal type.

32. A method for classifying signals occurring in a frequency band, comprising:

a. comparing a pulse timing template defining a period between at least two signal pulses against data representing signal pulses detected in the frequency band and declaring that a particular signal is occurring when a percentage of matches to the pulse timing template occur over a first time interval exceed a first threshold;

b. subsequent to said declaring, continuing to compare the pulse timing template against the data over a second time interval; and

c. declaring termination of the particular signal when a percentage of matches to the pulse signature template over the second time interval do not exceed a second threshold, and otherwise continuing to maintain that the particular signal is occurring;

d. wherein the first threshold is greater than the second threshold.

33. The method of claim 32, and further comprising the step of adjusting the pulse timing template to account for drift over time between the template and the signal that has been determined to match the template.

34. The method of claim 32, and further comprising the step of estimating the drift between the signal and the template.

35. The method of claim 32, and further comprising the step of generating the data derived from radio frequency energy simultaneously received across the entire frequency band for intervals of time.

36. The method of claim 32, and further comprising the step of generating the data derived from radio frequency energy received by a receiver scanned across portions of the frequency band, and dwelling on each portion for a time interval.

37. A method for classifying signals occurring in a radio frequency band comprising:

a. accumulating over time signal pulse data for signal pulses detected in the frequency band, wherein the accumulated signal pulse data comprises the start time, center frequency, bandwidth and duration for each detected signal pulse;

b. executing a classification procedure against the accumulated signal pulse data, wherein the classification procedure compares against the accumulated signal pulse data a timing template describing at least one time period between signal pulses that satisfy defined ranges for bandwidth, duration and center frequency, to determine whether a particular signal is occurring in the frequency band; and

c. designating for one or more signal pulses in the accumulated signal pulse data an identifier associated with the classification procedure when a match is determined with respect to the one or more signal pulses, thereby tracking which signal pulses in the accumulated signal pulse data are associated with a match to the particular signal.

38. The method of claim 37, and further comprising the step of determining multiple instances of the same signal type occurring in the frequency band.

39. A method for classifying signals occurring in a radio frequency band, comprising steps of:

a. generating data derived from received radio frequency energy in a frequency band when a radio receiver is scanned across the radio frequency band and dwells for a period of time at portions of the frequency band;

b. examining the data derived from received radio frequency energy;

c. preliminarily determining occurrence of a particular signal type in the frequency band based on the data; and

d. applying a classification procedure for the particular signal against data obtained when a radio receiver is tuned to only those portions of the radio frequency band in which the particular signal type is known to occur.

40. The method of claim 39, wherein the step of generating data comprises generating spectral data derived from the received radio frequency energy, and the step of examining comprises examining the spectral data.

41. The method of claim 39, wherein the step of generating data comprises generating signal pulse data for pulses of radio frequency energy detected in the frequency band over time.

42. The method of claim 41, wherein the step of generating signal pulse data comprises computing spectral information representing received radio frequency energy and comparing the spectral information with a set of signal pulse characteristics that comprises ranges for center frequency, duration and bandwidth, and outputting signal pulse data for pulses that meet the set of signal pulse characteristics.

43. A method for classifying signals occurring in a radio frequency band, comprising steps of:

a. generating power versus frequency data derived from received radio frequency energy in the radio frequency band; and

b. comparing a power template describing power versus frequency characteristics for a particular signal against the power versus frequency data derived from the received radio frequency energy to determine whether the particular signal is occurring in the radio frequency band.

44. The method of claim 43, wherein the step of comparing comprises comparing each of a plurality of power templates describing power versus frequency characteristics associated with the particular signal against the power versus frequency data derived from the received radio frequency energy, wherein the particular signal is declared to be occurring when the power versus frequency data matches any one of the plurality of power templates.

45. The method of claim 43, wherein the step of comparing comprises comparing each of a plurality of power templates against power versus frequency data obtained during multiple time intervals, and wherein the particular signal is determined to be occurring when the power versus frequency data matches at least one of the plurality of power templates for each of a predetermined number of time intervals of power versus frequency data.

46. The method of claim 43, wherein the step of comparing comprises comparing each of a plurality of power templates against power versus frequency data for multiple time intervals, and wherein the particular signal is determined to be occurring when the power versus frequency data matches all of the power templates over the multiple time intervals.

47. The method of claim 46, wherein the step of executing comprises executing a classification procedure that requires a match to a different one of the plurality of power templates during each of the multiple time intervals.

48. The method of claim 43, wherein the power template describes average power versus frequency characteristics for the particular signal, and wherein the step of comparing comprises comparing the power template with average power versus frequency data derived from the received radio frequency energy for a time interval.

49. The method of claim 43, wherein the power template describes maximum power versus frequency characteristics for the particular signal, and wherein the step of comparing comprises comparing the power template with maximum power versus frequency data derived from the received radio frequency energy for a time interval.

50. A method for classifying signals occurring in a radio frequency band, comprising steps of:

a. accumulating over time signal pulse data for signal pulses detected in the frequency band, wherein the accumulated signal pulse data comprises the start time, center frequency, bandwidth and duration for each detected signal pulse; and

b. executing a classification procedure against the accumulated signal pulse data, wherein the classification procedure compares against the accumulated signal pulse data a timing template describing at least one time period between signal pulses that satisfy defined ranges of at least one or bandwidth, duration and center frequency, to determine whether multiple instances of the same signal type are occurring in the frequency band.

51. The method of claim 50, and further comprising the step of comparing the timing template against the accumulated signal pulse data multiple iterations, each iteration at a different time offset with respect to the accumulated signal pulse data in order to detect multiple occurrences of the same signal type.

52. The method of claim 50, and further comprising the step of designating for one or more signal pulses in the accumulated signal pulse data an identifier associated with the classification procedure when a match is determined with respect to the one or more signal pulses, thereby tracking which signal pulses in the accumulated signal pulse data are associated with a match to the particular signal.