Systems, methods, and devices for automatic signal detection with temporal feature extraction within a spectrum
Systems, methods and apparatus are disclosed for automatic signal detection in an RF environment. An apparatus comprises at least one receiver and at least one processor coupled with at least one memory. The apparatus is at the edge of a communication network. The apparatus sweeps and learns the RF environment in a predetermined period based on statistical learning techniques, thereby creating learning data. The apparatus forms a knowledge map based on the learning data, scrubs a real-time spectral sweep against the knowledge map, and creates impressions on the RF environment based on a machine learning algorithm. The apparatus is operable to detect at least one signal in the RF environment.
1. An apparatus for signal detection in an electromagnetic environment, comprising:
at least one receiver, at least one processor, and at least one memory;
wherein the apparatus is operable to create power level measurements of the electromagnetic environment;
wherein the apparatus is operable to determine a baseline;
wherein the apparatus is operable to reveal at least one signal based on the baseline to create signal data;
wherein the apparatus is operable to use a calibration vector to de-bias raw signal data;
wherein the apparatus is operable to determine if the at least one signal is moving using a frequency-locked loop by determining if there is a Doppler change in the at least one signal;
wherein the apparatus is operable to calculate a first derivative of the power level measurements and a second derivative of the power level measurements;
wherein the apparatus is operable to select most prominent derivatives of the first derivative and the second derivative;
wherein the apparatus is operable to perform a squaring function on the most prominent derivatives;
wherein the apparatus is operable to create impressions of the electromagnetic environment based on a machine learning algorithm;
wherein the apparatus is operable to process the signal data using compressed data for deltas;
wherein the apparatus is operable to reconstruct the at least one signal using the deltas and the baseline;
wherein the apparatus is operable to fill gaps during reconstruction of the at least one signal where data of the at least one signal is absent; and
wherein the apparatus is operable to produce a topographic map displaying propagation of spectral power per frequency band.
2. The apparatus of claim 1 , further comprising a multiplicity of receivers, wherein the multiplicity of receivers is operable to monitor multiple bandwidths and perform time-frequency analyses at the same time.
3. The apparatus of claim 1 , wherein the apparatus is operable to reveal the at least one signal by subtracting the baseline from a spectral sweep to reveal an average power above the baseline.
4. The apparatus of claim 1 , wherein the apparatus is operable to determine a baseline by averaging a spectral sweep, removing areas identified by matched positive and negative gradients, and connecting points between removed areas.
5. The apparatus of claim 1 , wherein the apparatus is mobile and linked to a Global Positioning System (GPS) system for time and location.
6. The apparatus of claim 1 , wherein the apparatus is operable to estimate a location of a signal emitting device from which the at least one signal is emitted based on in-phase and quadrature (I/Q) data generated from a spectral sweep.
7. The apparatus of claim 1 , wherein the at least one signal is interference.
8. The apparatus of claim 1 , wherein the apparatus is operable to form a knowledge map of the electromagnetic environment based on the power level measurements of the electromagnetic environment.
9. The apparatus of claim 8 , wherein the apparatus is operable to scrub a spectral sweep against the knowledge map.
10. The apparatus of claim 1 , wherein the apparatus is operable to detect the at least one signal in the electromagnetic environment based on matched positive and negative gradients.
11. The apparatus of claim 1 , wherein the machine learning algorithm is an artificial neural network (ANN) algorithm.
12. A method for signal detection in an electromagnetic environment, comprising:
creating power level measurements of the electromagnetic environment;
forming a knowledge map of the electromagnetic environment based on the power level measurements of the electromagnetic environment;
determining a baseline;
subtracting the baseline from a spectral sweep to reveal at least one signal to create signal data;
using a calibration vector to de-bias raw signal data;
determining if the at least one signal is moving using a frequency-locked loop by determining if there is a Doppler change in the at least one signal;
calculating a first derivative of the power level measurements and a second derivative of the power level measurements;
selecting most prominent derivatives of the first derivative and the second derivative;
performing a squaring function on the most prominent derivatives;
creating impressions of the RF environment based on a machine learning algorithm;
processing the signal data using compressed data for deltas;
reconstructing the at least one signal using the deltas and the baseline; and
filling gaps during reconstruction of the at least one signal where data of the at least one signal is absent.
13. The method of claim 12 , further comprising scrubbing the spectral sweep against the knowledge map.
14. A method for signal detection in an electromagnetic environment, comprising:
scrubbing a spectral sweep against a knowledge map of the electromagnetic environment;
smoothing the spectral sweep with a correction vector;
detecting at least one signal in the electromagnetic environment;
determining a baseline;
subtracting the baseline from the spectral sweep to reveal the at least one signal to create signal data;
using a calibration vector to de-bias raw signal data;
determining if the at least one signal is moving using a frequency-locked loop by determining if there is a Doppler change in the at least one signal;
calculating a first derivative of the power level measurements and a second derivative of the power level measurements;
selecting most prominent derivatives of the first derivative and the second derivative;
performing a squaring function on the most prominent derivatives;
creating impressions of the RF environment based on a machine learning algorithm;
processing the signal data using compressed data for deltas;
reconstructing the at least one signal using the deltas and the baseline; and
filling gaps during reconstruction of the at least one signal where data of the at least one signal is absent.
15. The method of claim 14 , further comprising extracting at least one time-frequency feature from the knowledge map.
16. The method of claim 14 , wherein the at least one signal is a narrowband signal hidden in a wideband signal, and wherein the narrowband signal having a bandwidth ranging from 1 kHz to 60 kHz is inside the wideband signal with a bandwidth up to 100 MHz.
17. The method of claim 14 , wherein the knowledge map comprises an array of normal distributions, wherein each normal distribution corresponds to how often a power level at each frequency has been at a particular level.
18. The method of claim 14 , further comprising creating a profile of the electromagnetic environment based on the knowledge map, wherein the profile comprises a highest power level at each frequency during a predetermined period.
19. The method of claim 14 , further comprising sending a notification and/or an alarm to at least one remote device after detecting the at least one signal.
20. The method of claim 14 , wherein detecting the at least one signal in the electromagnetic environment is based on matched positive and negative gradients.