Categorizing radio signals while obfuscating characteristics of federal radio signals
Systems and methods are provided for categorizing signals sensed from receivers into one of a plurality of categories using a machine learning component that obfuscates other characteristics of the signals beyond the categorization of the signal. A set of training signals is generated that includes signals representative of each of a plurality of categories of signals. A machine learning component is trained using the set of training signals so that the machine learning technique can categorize new signals into one of the plurality of categories. The machine learning component is transmitted to one or more receivers so that each receiver can categorize sensed signals into one of the plurality of categories using the machine learning component while obfuscating other characteristics of the sensed signals.
1. A computerized method of categorizing signals sensed from receivers into one of a plurality of categories using a machine learning component that obfuscates other characteristics of the signals beyond the categorization of the signal, comprising:
generating, by a computing device, a set of training signals comprising signals representative of each of a plurality of categories of signals, the plurality of categories including categories for incumbent signal types, managed signal types, noise signal types, and clutter signal types;
training, by the computing device, a machine learning component using the set of training signals so that the machine learning technique can categorize new signals into one of the plurality of categories; and
transmitting, by the computing device, the machine learning component to one or more receivers so that each receiver can categorize sensed signals into one of the plurality of categories using the machine learning component while obfuscating other characteristics of the sensed signals.
2. The computerized method of claim 1 further comprising:
receiving data from a receiver of the one or more receivers indicative of a device interfering with an incumbent device; and
transmitting a notification to the device to terminate transmissions to prevent the device from interfering with the incumbent device.
3. The computerized method of claim 1 wherein:
generating the set of training signals comprises determining a set of spectrum artifacts that occur below a given noise floor but that may be integrated above the noise floor; and
the machine learning component comprises training the machine learning component using the set of spectrum artifacts so that the machine learning component can detect and categorize interference between signals from a plurality of devices.
4. The computerized method of claim 1 further comprising:
detecting interference in a given band of spectrum;
detecting signal artifacts in adjacent or distant bands of the spectrum; and
correlating the detected interference with the detected signal artifacts to determine whether a signal categorization is reliable.
5. The computerized method of claim 1 , wherein training the machine learning component comprises training a set of signal structures with respect to time and frequency occupancy to generate a hidden Markov model to integrate a diversity of signal artifacts occurring in one or more bands in the machine learning component.
6. The computerized method of claim 5 , wherein the one or more bands comprises a managed band, a band that is distant in center frequency from a band of operation, a band that is distant in center frequency from a band of interest for future usage, or any combination thereof.
7. The computerized method of claim 1 , further comprising:
receiving data indicative of incumbent interference in a frequency band; and
determining a degree of interference in a band of interest based on the incumbent interference.
8. A computing device for categorizing signals sensed from receivers into one of a plurality of categories using a machine learning component that obfuscates other characteristics of the signals beyond the categorization of the signal, comprising:
a database; and
a processor in communication with the database, and configured to run a module stored in memory that is configured to cause the processor to:
generate a set of training signals comprising signals representative of each of a plurality of categories of signals, the plurality of categories including categories for incumbent signal types, managed signal types, noise signal types, and clutter signal types;
train a machine learning component using the set of training signals so that the machine learning technique can categorize new signals into one of the plurality of categories; and
transmit the machine learning component to one or more receivers so that each receiver can categorize sensed signals into one of the plurality of categories using the machine learning component while obfuscating other characteristics of the sensed signals.
9. The computing device of claim 8 , wherein the module stored in memory is configured to cause the processor to:
receive data from a receiver of the one or more receivers indicative of a device interfering with an incumbent device; and
transmit a notification to the device to terminate transmissions to prevent the device from interfering with the incumbent device.
10. The computing device of claim 8 , wherein the module stored in memory is configured to cause the processor to:
generate the set of training signals comprising determining a set of spectrum artifacts that occur below a given noise floor but that may be integrated above the noise floor; and
train the machine learning component comprising training the machine learning component using the set of spectrum artifacts so that the machine learning component can detect and categorize interference between signals from a plurality of devices.
11. The computing device of claim 8 , wherein the module stored in memory is configured to cause the processor to:
detect interference in a given band of spectrum;
detect signal artifacts in adjacent or distant bands of the spectrum; and
correlate the detected interference with the detected signal artifacts to determine whether a signal categorization is reliable.
12. The computing device of claim 8 , wherein training the machine learning component comprises training a set of signal structures with respect to time and frequency occupancy to generate a hidden Markov model to integrate a diversity of signal artifacts occurring in one or more bands in the machine learning component.
13. The computing device of claim 12 , wherein the one or more bands comprises a managed band, a band that is distant in center frequency from a band of operation, a band that is distant in center frequency from a band of interest for future usage or any combination thereof.
14. The computing device of claim 8 , wherein the module stored in memory is configured to cause the processor to:
receive data indicative of incumbent interference in a frequency band; and
determine a degree of interference in band of interest based on the incumbent interference.
15. A non-transitory computer readable medium having executable instructions operable to cause an apparatus to:
generate a set of training signals comprising signals representative of each of a plurality of categories of signals, the plurality of categories including categories for incumbent signal types, managed signal types, noise signal types, and clutter signal types;
train a machine learning component using the set of training signals so that the machine learning technique can categorize new signals into one of the plurality of categories; and
transmit the machine learning component to one or more receivers so that each receiver can categorize sensed signals into one of the plurality of categories using the machine learning component while obfuscating other characteristics of the sensed signals.
16. The non-transitory computer readable medium of claim 15 , having executable instructions operable to cause an apparatus to:
receive data from a receiver of the one or more receivers indicative of a device interfering with an incumbent device; and
transmit a notification to the device to terminate transmission to prevent the device from interfering with the incumbent device.
17. The non-transitory computer readable medium of claim 15 , having executable instructions operable to cause an apparatus to:
generate the set of training signals comprising determining a set of spectrum artifacts that occur below a given noise floor but that may be integrated above the noise floor; and
train the machine learning component comprising training the machine learning component using the set of spectrum artifacts so that the machine learning component can detect and categorize interference between signals from a plurality of devices.
18. The non-transitory computer readable medium of claim 15 , having executable instructions operable to cause an apparatus to:
detect interference in a given band of spectrum;
detect signal artifacts in adjacent of distant bands of the spectrum; and
correlate the detected interference with the detected signal artifacts to determine whether a signal categorization is reliable.
19. The non-transitory computer readable medium of claim 15 , wherein training the machine learning component comprises training a set of signal structures with respect to time and frequency occupancy generate a hidden Markov model to integrate a diversity of signal artifacts occurring in one or more bands in the machine learning component.
20. The non-transitory computer readable medium of claim 15 , having executable instructions operable to cause an apparatus to:
receive data indicative of incumbent interference in a frequency band; and
determine a degree of interference in a band of interest based on the incumbent interferences.