Leveraging spectral diversity for machine learning-based estimation of radio frequency signal parameters

Patent metadata

Field	Value
Publication number	US-11531080-B2
Application number	US-201916584545-A
Country	US
Kind code	B2
Filing date	Sep 26, 2019
Priority date	Jul 24, 2019
Publication date	Dec 20, 2022
Grant date	Dec 20, 2022

Abstract

An example method for estimating the angle-of-arrival (AoA) and other parameters of radio frequency (RF) signals that are received by an antenna array comprises: receiving a plurality of radio frequency (RF) signal power measurements by a plurality of antenna elements at a plurality of RF channels; computing, by applying a machine learning model to the plurality of RF signal power measurements, an estimated RF signal parameter value; and outputting the RF signal parameter value.

First claim

What is claimed is:

1. A method, comprising:
receiving, by a processing device, a plurality of radio frequency (RF) signal power measurements by a plurality of antenna elements at a plurality of RF channels;
computing, by the processing device applying a first machine learning model to the plurality of RF signal power measurements, an estimated RF signal parameter value;
responsive to determining that an estimated error value exceeds a predetermined threshold, re-computing the estimated RF signal parameter value by applying a second machine learning model, wherein the second machine learning model corresponds to a larger number of RF signal multi-paths than the first machine learning model; and
outputting, by the processing device, the estimated RF signal parameter value.

2. The method of claim 1 , wherein the plurality of RF signal power measurements includes a plurality of RF signal magnitude values and a plurality of RF signal phase values.

3. The method of claim 1 , wherein the RF signal parameter value is one of: a line-of-sight (LoS) angle-of-arrival (AoA), a reflection AoA, an attenuation of reflection, a relative delay of reflection, or a relative phase.

4. The method of claim 1 , wherein the first machine learning model performs a feature extraction to transform the plurality of RF signal power measurements to an intermediate representation in a reduced-dimension space.

5. The method of claim 4 , wherein the first machine learning model performs a regression task to infer the estimated RF signal parameter value from the intermediate representation in the reduced-dimension space.

6. The method of claim 1 , further comprising:
training the first machine learning model by processing a training data set comprising a second plurality of RF signal power measurements by a second plurality of antenna elements at a second plurality of RF channels, wherein the RF signal power measurements are labeled with corresponding RF signal parameter values.

7. The method of claim 1 , wherein the first machine learning model is a neural network, and wherein receiving the plurality of RF signal power measurements and applying the first machine learning model is performed iteratively until the estimated error value falls below the predetermined threshold.

8. A system comprising:
a transceiver configured to couple to an antenna comprising a plurality of antenna elements to receive a plurality of radio frequency (RF) signal power measurements at a plurality of frequency channels; and
a processor coupled to the transceiver, the processor to:
select a machine learning model from a sequence of machine learning models, wherein a first machine learning model of the sequence has been trained on a first training data set produced in a first signal propagation environment that includes a first number of signal paths, wherein a second machine learning model of the sequence has been trained on a second training data set produced in a second signal propagation environment that includes a second number of signal paths, and wherein the second number of signal paths exceeds the first number of signal paths;
compute, by applying the selected machine learning model to the plurality of RF signal measurements, an estimated RF signal parameter value; and
responsive to determining that a parameter estimation error is less than or equal a predetermined error threshold value, output the estimated RF signal parameter value.

9. The system of claim 8 , wherein the processor is further to:
responsive to determining that the parameter estimation error exceeds the predetermined error threshold value, iteratively repeat the selecting and computing operations until the estimated RF signal parameter value falls below the predetermined error threshold value.

10. The system of claim 8 , wherein the plurality of RF signal power measurements includes a plurality of RF signal magnitude values and a plurality of RF signal phase values.

11. The system of claim 8 , wherein the RF signal parameter value is one of: a line-of-sight (LoS) angle-of-arrival (AoA), a reflection AoA, an attenuation of reflection, a relative delay of reflection, or a relative phase.

12. The system of claim 8 , wherein each machine learning model of the sequence of machine learning models performs a feature extraction to transform the plurality of RF signal power measurements to an intermediate representation in a reduced-dimension space.

13. The system of claim 12 , wherein the machine learning model performs a regression task to infer the estimated RF signal parameter value from the intermediate representation in the reduced-dimension space.

14. A device comprising:
a transceiver configured to couple to an antenna comprising a plurality of antenna elements to receive a plurality of radio frequency (RF) signal power measurements at a plurality of frequency channels at a plurality of RF channels; and
a processor coupled to the transceiver, the processor to:
compute, by applying a first machine learning model to the plurality of RF signal measurements at the plurality of RF channels, an estimated RF signal parameter value;
responsive to determining that an estimated error value exceeds a predetermined threshold, re-compute the estimated RF signal parameter value by applying a second machine learning model, wherein the second machine learning model corresponds to a larger number of RF signal multi-paths than the first machine learning model; and
responsive to determining that a parameter estimation error is less than or equal a predetermined error threshold value, output the estimated RF signal parameter value.

15. The device of claim 14 , wherein the processor is further to:
responsive to determining that the parameter estimation error exceeds the predetermined error threshold value, repeat the computing operation for the next plurality of RF signal measurements received at a different set of RF channels.

16. The device of claim 14 , wherein the plurality of RF signal power measurements includes a plurality of RF signal magnitude values and a plurality of RF signal phase values.

17. The device of claim 14 , wherein the RF signal parameter value is one of: a line-of-sight (LoS) angle-of-arrival (AoA), a reflection AoA, an attenuation of reflection, a relative delay of reflection, or a relative phase.

18. The device of claim 14 , wherein the first machine learning model performs a feature extraction to transform the plurality of RF signal power measurements to an intermediate representation in a reduced-dimension space.

19. The device of claim 18 , wherein the first machine learning model performs a regression task to infer the estimated RF signal parameter value from the intermediate representation in the reduced-dimension space.

Assignees

• Cypress Semiconductor Corp

Inventors

• Smyth Aidan
• Simileysky Victor
• Uln Kiran

Classifications

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• H04B7/0697

  using spatial multiplexing · CPC title

• G06N20/00

  Machine learning · CPC title

• G01S11/06

  using intensity measurements · CPC title