Machine learning-based antenna array validation, prototyping and optimization

Patent metadata

Field	Value
Publication number	US-11303041-B2
Application number	US-201916584383-A
Country	US
Kind code	B2
Filing date	Sep 26, 2019
Priority date	Jul 24, 2019
Publication date	Apr 12, 2022
Grant date	Apr 12, 2022

Abstract

An example method of estimating the angular resolution of antenna array comprises: receiving a plurality of values of magnitude and phase of a radio frequency (RF) signal for each antenna element of a plurality of antenna elements comprised by an antenna array; performing, by a machine learning model, a feature extraction operation to transform the plurality of values of magnitude and phase into a plurality of data points in a reduced-dimension space; clustering, by the machine learning model, the plurality of data points into a plurality of clusters; and computing, based on the clustered data points, an angular resolution value for the antenna array.

First claim

What is claimed is:

1. A method, comprising:
receiving, by a processing device, a plurality of values of magnitude and phase of a radio frequency (RF) signal for each antenna element of a plurality of antenna elements comprised by an antenna array;
performing, by the processing device applying a machine learning model, a feature extraction operation to transform the plurality of values of magnitude and phase into a plurality of data points in a reduced-dimension space;
clustering, by the processing device applying the machine learning model, the plurality of data points into a plurality of clusters;
computing, by the processing device, based on the plurality of clusters, an angular resolution value of the antenna array, wherein the angular resolution value reflects a difference between a first angle-of-arrival (AoA) value associated with a first cluster centroid of a first cluster of the plurality of clusters and a second AoA value associated with a second cluster centroid of a second cluster of the plurality of clusters; and
outputting, by the processing device, the angular resolution value.

2. The method of claim 1 , further comprising:
receiving a plurality of values of design parameters of the antenna array;
computing, by applying a simulation model to the plurality of values of design parameters, the plurality of values of magnitude and phase of the RF signal for each antenna element of the plurality of antenna elements.

3. The method of claim 1 , wherein computing the angular resolution value further comprises:
performing, by applying the machine learning model, a regression operation to infer the angular resolution value from the plurality of data points.

4. The method of claim 1 , further comprising:
performing, by applying the machine learning model, a regression operation to compute, for each cluster of the plurality of clusters, a corresponding value of an angle-of-arrival (AoA).

5. The method of claim 1 , further comprising:
identifying a subset of overlapping clusters of the plurality of clusters; and
identifying, based on the subset of overlapping clusters, a physical parameter of the antenna array that has caused the subset of overlapping clusters to overlap.

6. The method of claim 1 , wherein the machine learning model is represented by a neural network.

7. The method of claim 6 , further comprising:
perform an unsupervised training procedure for training the neural network for performing the feature extraction operation.

8. The method of claim 6 , further comprising:
perform an unsupervised training procedure for training the neural network for performing the clustering operation.

9. The method of claim 6 , further comprising:
perform a supervised training procedure for training the neural network for performing a regression operation to infer the angular resolution value from the plurality of data points.

10. A method, comprising:
receiving, by a processing device, a plurality of values of design parameters of an antenna array comprising a plurality of antenna elements;
computing, by the processing device applying a simulation model to the plurality of values of design parameters, a plurality of values of magnitude and phase of the RF signal for each antenna element of the plurality of antenna elements;
computing, by the processing device applying a machine learning model, an angular resolution value of the antenna array; and
responsive to determining that the angular resolution value exceeds a predetermined threshold, identifying a design parameter of the antenna array that has caused the angular resolution value to exceed the predetermined threshold.

11. The method of claim 10 , further comprising:
responsive to determining that the angular resolution value is less than or equal to a predetermined threshold, outputting the angular resolution value of the antenna array.

12. The method of claim 10 , wherein computing the angular resolution value further comprises:
performing, by a machine learning model, a feature extraction operation to transform the plurality of values of magnitude and phase into a plurality of data points in a reduced-dimension space;
clustering, by the machine learning model, the plurality of data points into a plurality of clusters; and
performing, by the machine learning model, a regression operation to infer the angular resolution value from the plurality of clusters.

13. The method of claim 12 , further comprising:
perform an unsupervised training procedure for training the machine learning model for performing the feature extraction operation.

14. The method of claim 12 , further comprising:
perform an unsupervised training procedure for training the machine learning model for performing the clustering operation.

15. The method of claim 12 , further comprising:
perform a supervised training procedure for training the machine learning model for performing a regression operation to infer the angular resolution value from the plurality of clusters.

16. The method of claim 12 , further comprising:
computing, for each cluster of the plurality of clusters, a corresponding value of an angle-of-arrival (AoA).

17. The method of claim 10 , wherein the machine learning model is represented by a neural network.

18. A device comprising:
a transceiver configured to couple to an antenna comprising a plurality of antenna elements, the transceiver to receive a plurality of values of magnitude and phase of a radio frequency (RF) signal for each antenna element of the plurality of antenna elements; and
a processor coupled to the transceiver, the processor to:
perform, by a machine learning model, a feature extraction operation to transform the plurality of values of magnitude and phase into a plurality of data points in a reduced-dimension space;
classify, by the processing device applying the machine learning model, the plurality of data points into a plurality of clusters; and
compute, based on the plurality of clusters, an angular resolution value of the antenna, wherein the angular resolution value reflects a difference between a first angle-of-arrival (AoA) value associated with a first cluster centroid of a first cluster of the plurality of clusters and a second AoA value associated with a second cluster centroid of a second cluster of the plurality of clusters.

19. The device of claim 18 , wherein the machine learning model is represented by a neural network.

20. The device of claim 18 , wherein the processor is further to:
compute, for each cluster of the plurality of clusters, a corresponding value of an angle-of-arrival (AoA).

Assignees

• Cypress Semiconductor Corp

Inventors

• Smyth Aidan
• Uln Kiran
• Simileysky Victor
• Zhang Zhuohui

Classifications

• H01Q21/24 Primary

  Combinations of antenna units polarised in different directions for transmitting or receiving circularly and elliptically polarised waves or waves linearly polarised in any direction {(circularly polarised patch antennas H01Q9/0428; circularly polarised horns H01Q13/0241; cross-polarised horns H01Q13/0258; polarisation converters H01Q15/242; cross-polarised rear feeds H01Q19/136; crossed polarisation dual antenna H01Q25/001)} · CPC title

• H04B17/3912 Primary

  Simulation models, e.g. distribution of spectral power density or received signal strength indicator [RSSI] for a given geographic region · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• H01Q3/2605

  Array of radiating elements provided with a feedback control over the element weights, e.g. adaptive arrays · CPC title