Cognitive radio methodology, physical layer policies and machine learning

• US 20080293353A1
• Filed: 03/06/2008
• Published: 11/27/2008
• Est. Priority Date: 03/08/2007
• Status: Active Grant

• Current Assignee: BAE Systems Information and Electronic Systems Integration Incorporated (BAE Systems Plc)
• Original Assignee: BAE Systems Information and Electronic Systems Integration Incorporated (BAE Systems Plc)
• Inventors: Mody, Apurva N., Blatt, Stephen R., Mills, Diane G., McElwain, Thomas P., Thammakhoune, Ned B.
• Granted Patent: US 8,515,473 B2
• Time in Patent Office: Days
• US Class Current: 455/1
• CPC Class Codes: H04K 3/226 Selection of non-jammed channel for communication spectrum sharing arrangements H04W16/14; selection of wireless resources by user or terminal H04W72/02
• CPC Class Codes: H04K 3/44 characterized by the control of the jamming waveform or modulation type
• CPC Class Codes: H04K 3/45 characterized by including monitoring of the target or target signal, e.g. in reactive jammers or follower jammers for example by means of an alternation of jamming phases and monitoring phases, called "look-through mode"
• CPC Class Codes: H04K 3/94 related to allowing or preventing testing or assessing
• CPC Class Codes: H04W 48/02 Access restriction performed under specific conditions
• CPC Class Codes: H04W 72/541 using the level of interference
• CPC Class Codes: Y02D 30/70 in wireless communication networks

Abstract

In a method of cognitive communication for non-interfering transmission, wherein the improvement comprises the step of conducting radio scene analysis to find not just the spectrum holes or White spaces; but also to use the signal classification, machine learning and prediction information to learn more things about the existing signals and its underlying protocols, to find the Gray space, hence utilizing the signal space, consisting of space, time, frequency (spectrum), code and location more efficiently.

In a method of cognitive jamming where smart and energy efficient jamming techniques are suggested based on sensing, classification and machine learning of the existing signals.

Claims

35. A method of smart or cognitive jamming using a jammer, comprising the step of using smart and energy efficient jamming techniques based on radio scene analysis including sensing, classification and machine learning of existing signals.

1. A method for permitting maximum utilization of a frequency spectrum utilizing a cognitive communication system, comprising of steps:

conducting a radio scene analysis to find spectrum holes and signal spaces for non-interfering signal transmissions; and

,

responsive to the radio scene analysis providing a signal transmission either in the spectrum holes or the signal space for non-interfering signal transmission.

2. The method of claim 1, wherein the signal space for non-interfering signal transmission is Gray Space.

3. The method of claim 1, wherein the spectrum holes include White Space, where White space is a subset of the Gray space.

4. The method of claim 1, wherein the radio scene analysis includes clutter suppression.

5. The method of claim 1, wherein the radio scene analysis includes signal detection including spectrum sensing.

6. The method of claim 1, wherein the radio scene analysis includes feature extraction.

7. The method of claim 6, wherein feature extraction takes place on detected signals and wherein the feature extraction includes outputting a feature vector.

8. The method of claim 1, wherein the radio scene analysis includes signal classification.

9. The method of claim 8, and further including the step of providing feature extraction that outputs a feature vector, and wherein the signal classification utilizes the feature vector in the classification.

10. The method of claim 1, wherein the step of conducting radio scene analysis includes signal detection, followed by feature extraction, followed by signal classification, and wherein the output from the feature extraction step and the signal classification step are coupled to a learning and prediction process.

11. The method of claim 10, and further including the process of generating decision metrics and policies based on the output of the learning and prediction process.

12. The method of claim 11, and further including the step of generating communications in White Space as well as Gray Space as determined from the outputs of the signal classification process and the decision metrics and policies process.

13. The method of claim 12, and further including the step of measuring the performance in generating communications in White and Gray Spaces and updating the learning and prediction process if performance falls below a predetermined threshold.

14. The method of claim 13, wherein if the performance falls below the predetermined threshold, causing the feature extraction process to extract more features.

15. The method of claim 14, wherein the performance measuring process outputs a signal to the learning and prediction process if the measured performance is below a predetermined value.

16. The method of claim 1, wherein the step of conducting radio scene analysis includes feature extraction, clustering and signal classification including the classifications of a signal being of at least one of broadband, narrowband, frequency hopping, non-frequency hopping, single carrier, multi-carrier, broad pulse, narrow pulse, single signal or multiple signals, time division multiple access, code division multiple access, spatial multiple access, beam forming or multi-input multi-output.

17. The method of claim 1, wherein the step of conducting radio scene analysis includes analyzing the signal space in terms of space, time, frequency, code and location.

18. The method of claim 17, wherein the analyzing step includes feature extraction and clustering that results in signal classification of least one of broadband, narrowband, frequency hopping, non-frequency hopping, single carrier, multi-carrier, broad pulse, narrow pulse, single signal or multiple signals, time division multiple access, code division multiple access, orthogonal frequency division multiple access, spatial multiple access, beam forming or multi-input multi-output.

19. The method of claim 1, wherein the step of conducting radio scene analysis includes a signal classification process including a feature measurement phase and a classification phase.

20. The method of claim 19, wherein the weightings or coefficients in the classification process are modified during a training phase based on the classification results against known data.

21. The method of claim 20, wherein the classification process includes at least one algorithm that uses at least one of definite metrics to separate classes that class estimation based on models having a priori assumptions, and connectionist classifications based processes including neural networks with variably weighted inputs.

22. The method of claim 21, wherein the classification process includes at least one of associative and incremental learning.

23. The method of claim 1, wherein the step of conducting radio scene analysis includes a process of feature extraction with features including at least one of higher order statistics, singular values using singular value decomposition, eigen values using eigen decomposition, time width of the clusters, bandwidth of the clusters, center frequencies of the clusters, time of arrival, time difference of arrival, mean, variance, standard deviation, probability of event occurrence of the various features, raw features, processed features, and time frequency detection ratios.

24. The method of claim 1, wherein a step of conducting radio scene analysis includes machine learning and prediction.

25. The method of claim 1, wherein a step of conducting radio scene analysis includes a process for predicting future behavior of detected signals.

26. The method of claim 25, wherein the step of generating non-interfering signal transmission takes into account the predicted signal behavior.

27. The method of claim 1, wherein the radio scene analysis includes at least one of feature extraction, signal classification and learning and a prediction process, and wherein information from the feature extraction, the signal classification and learning and the prediction process is used to make a decision on signal transmission in a signal space using a Scenario and Action Table consisting of Policy Sets.

28. The method of claim 27, wherein the policy sets are one of competitive or non-competitive.

29. The method of claim 27, wherein the policy sets are location dependent.

30. The method of claim 27, wherein the policy sets are adaptive, machine learning and game theoretic formulations being utilized to adapt the policy sets.

31. The method of claim 27, wherein transmitted signals in the Gray space or the White space are of at least one of broadband, narrowband, frequency hopping, non-frequency hopping, single carrier, multi-carrier, broad pulse, narrow pulse, single signal or multiple signals, time division multiple access, code division multiple access, orthogonal frequency division multiple access, spatial multiple access, beam forming and multi-input multi-output.

32. The method of claim 27, wherein, an orthogonal frequency division multiple access signal is used to generate at least one of broadband, narrow band, broad pulse, narrow pulse, frequency hopping, non frequency hopping, single carrier, multi carrier, space division multiple access, multi-carrier code division multiple access, multiple input multiple output signals by selectively switching its sub-channels on or off over a period of time.

33. The method of claim 1, wherein the radio scene analysis includes knowledge of the existing signals derived from external sources as well as signal classification and machine learning, the knowledge being combined to enable transmitting of a new signal that maximizes the signal space utilization without causing significant harmful interference to the existing signals.

34. The method of claim 1, wherein, White Space, Gray Space, or a combination of both is used for signal transmission.

36. A method of claim 35, wherein or cognitive jamming causes maximum harmful interference to disrupt information exchange, at a minimum cost to itself.

37. The method of claim 35, wherein the jamming includes injecting an interfering signal in a signal space that is in the Gray Space.

38. The method of claim 35, wherein the radio scene analysis includes signal detection including spectrum sensing.

39. The method of claim 35, wherein the radio scene analysis includes feature extraction.

40. The method of claim 39, wherein feature extraction takes place on detected signals and wherein the feature extraction includes outputting a feature vector.

41. The method of claim 35, wherein the radio scene analysis includes signal classification.

42. The method of claim 39, wherein the step of providing feature extraction includes outputting a feature vector, and further including the step of signal classification utilizing the feature vector in the classification.

43. The method of claim 35, wherein the radio scene analysis includes signal detection, followed by feature extraction, followed by signal classification, and wherein the output from the feature extraction step and the signal classification step are coupled to a learning and prediction process.

44. The method of claim 35, and further including the step of generating decision metrics and policies based on the output of the learning and prediction process.

45. The method of claim 35, wherein information from signal sensing, classification, and machine learning is used to jam the existing signals at a minimum cost to the jammer.

44-1. The method of claim 35, wherein cost to the jammer is measured in terms of energy spent to disrupt communication.