Machine learning-based proximity detection using impedance

Abstract

Certain aspects of this disclosure provide techniques and apparatus for improved machine learning. Impedance information relating to a wireless transmitter of a device is determined, and impedance change information is generated based on the difference between the impedance information relating to the wireless transmitter and the previous impedance information. An off-body characteristic is generated based on processing the impedance change information using a trained machine learning model, and this off-body characteristic indicates the probability that the device was off-body when the impedance information was determined.

Claims

1. A processor-driven method for proximity detection using machine learning, Determining the impedance information of the device's wireless transmitter, The process involves generating impedance change information based on the difference between the impedance information and previous impedance information relating to the wireless transmitter, A method performed by a processor, comprising: generating an off-body characteristic that indicates the probability that the device was off-body when the impedance information was determined, based on processing the impedance change information using a trained machine learning model.

2. A method carried out by the processor according to claim 1, further comprising performing a power backoff operation on the wireless transmitter based on the off-body characteristics.

3. A method carried out by the processor according to claim 2, wherein performing the power backoff operation reduces the transmit power of the wireless transmitter based on the determination that the off-body characteristics satisfy one or more criteria indicating that the device was not off-body when the impedance information was determined.

4. The generation of the aforementioned impedance change information is To generate a first value that represents the magnitude of the difference between the aforementioned impedance information and the previous impedance information, A method carried out by the processor according to claim 1, comprising generating a second value representing the direction of the difference between the impedance information and the previous impedance information.

5. The second value indicates the direction of the difference between the real component of the impedance information and the real component of the previous impedance information. The generation of the impedance change information further includes generating a third value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the previous impedance information. A method carried out by the processor described in claim 4.

6. The generation of the aforementioned impedance change information is A third value is generated that represents the magnitude of the difference between the impedance information and the off-body reference point. A method carried out by the processor according to claim 4, further comprising generating a fourth value representing the direction of the difference between the impedance information and the off-body reference point.

7. The fourth value indicates the direction of the difference between the real component of the impedance information and the real component of the off-body reference point. The generation of the impedance change information further includes generating a fifth value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the off-body reference point. A method carried out by the processor described in claim 6.

8. A method carried out by the processor according to claim 1, further comprising generating a charging characteristic based on processing the impedance change information using the trained machine learning model, wherein the charging characteristic indicates the probability that a charging cable was plugged into the device when the impedance information was determined.

9. The aforementioned trained machine learning model is used to provide proximity detection. The device does not include a capacitive sensor for proximity detection. A method carried out by the processor described in claim 1.

10. The aforementioned trained machine learning model is trained on a set of impedance characterization records, and the set is A first subset of impedance characterization records, selected for training the trained machine learning model, wherein each corresponding impedance characterization record within the first subset is determined to have at least threshold similarity to an off-body reference point. A method carried out by the processor according to claim 1, comprising: a second subset of impedance characterization records, wherein each corresponding impedance characterization record in the second subset is selected for training the trained machine learning model on the basis that the corresponding impedance characterization has at least a threshold difference with respect to the off-body reference point.

11. A processor-driven method for training a machine learning model to perform proximity detection, Determining the impedance information of the device's wireless transmitter, The process involves generating impedance change information based on the difference between the impedance information and previous impedance information relating to the wireless transmitter, Based on processing the impedance change information using a machine learning model, an off-body characteristic is generated that indicates the probability that the device was off-body when the impedance information was determined. A method performed by a processor, comprising updating one or more parameters of the machine learning model based on comparing the off-body characteristics with a ground truth label associated with the impedance information.

12. The generation of the aforementioned impedance change information is To generate a first value that represents the magnitude of the difference between the aforementioned impedance information and the previous impedance information, A method carried out by the processor according to claim 11, comprising generating a second value representing the direction of the difference between the impedance information and the previous impedance information.

13. The second value indicates the direction of the difference between the real component of the impedance information and the real component of the previous impedance information. The generation of the impedance change information further includes generating a third value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the previous impedance information. A method carried out by the processor according to claim 12.

14. The generation of the aforementioned impedance change information is A third value is generated that represents the magnitude of the difference between the impedance information and the off-body reference point. A method carried out by the processor according to claim 12, further comprising generating a fourth value representing the direction of the difference between the impedance information and the off-body reference point.

15. The fourth value indicates the direction of the difference between the real component of the impedance information and the real component of the off-body reference point. A method carried out by the processor according to claim 14, wherein generating the impedance change information further comprises generating a fifth value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the off-body reference point.

16. Further includes updating one or more parameters of the machine learning model based on a set of impedance characterization records, the set of which A first subset of impedance characterization records, selected for training the machine learning model, wherein each corresponding impedance characterization record within the first subset is determined to have at least threshold similarity to an off-body reference point. A method carried out by the processor according to claim 11, comprising: a second subset of impedance characterization records, the second subset being selected for training the machine learning model on the basis that each corresponding impedance characterization record in the second subset has at least a threshold difference with respect to the off-body reference point.

17. A processing system, Memory containing computer executable instructions, One or more processors configured to execute the aforementioned computer executable instructions and cause the processing system to perform an operation, wherein the operation is Determining impedance information relating to the wireless transmitter of the processing system, The process involves generating impedance change information based on the difference between the impedance information and previous impedance information relating to the wireless transmitter, A processing system comprising one or more processors, which include generating an off-body characteristic that indicates the probability that the processing system was off-body when the impedance information was determined, based on processing the impedance change information using a trained machine learning model.

18. The processing system according to claim 17, further comprising performing a power back-off operation on the wireless transmitter based on the off-body characteristics.

19. The processing system according to claim 18, wherein performing the power back-off operation includes reducing the transmit power of the wireless transmitter based on determining that the off-body characteristics satisfy one or more criteria indicating that the processing system was not off-body when the impedance information was determined.

20. The generation of the aforementioned impedance change information is To generate a first value that represents the magnitude of the difference between the aforementioned impedance information and the previous impedance information, The processing system according to claim 17, comprising generating a second value that represents the direction of the difference between the impedance information and the previous impedance information.

21. The second value indicates the direction of the difference between the real component of the impedance information and the real component of the previous impedance information. The generation of the impedance change information further includes generating a third value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the previous impedance information. The processing system according to claim 20.

22. The generation of the aforementioned impedance change information is A third value is generated that represents the magnitude of the difference between the impedance information and the off-body reference point. The processing system according to claim 20, further comprising generating a fourth value representing the direction of the difference between the impedance information and the off-body reference point.

23. The fourth value indicates the direction of the difference between the real component of the impedance information and the real component of the off-body reference point. The generation of the impedance change information further includes generating a fifth value that represents the direction of the difference between the imaginary component of the impedance information and the imaginary component of the off-body reference point. The processing system according to claim 22.

24. The processing system according to claim 17, wherein the operation further comprises generating a charging characteristic based on processing the impedance change information using the trained machine learning model, the charging characteristic indicating the probability that a charging cable was plugged into the processing system when the impedance information was determined.

25. The aforementioned trained machine learning model is used to provide proximity detection. The processing system does not include a capacitive sensor for proximity detection. The processing system according to claim 17.

26. The aforementioned trained machine learning model is trained on a set of impedance characterization records, and the set is A first subset of impedance characterization records, selected for training the trained machine learning model, wherein each corresponding impedance characterization record within the first subset is determined to have at least threshold similarity to an off-body reference point. The processing system according to claim 17, comprising: a second subset of impedance characterization records, wherein each corresponding impedance characterization record in the second subset is selected for training the trained machine learning model on the basis that the corresponding impedance characterization has at least a threshold difference with respect to the off-body reference point.

27. A processing system, A means for determining impedance information relating to the wireless transmitter of the processing system, A means for generating impedance change information based on the difference between the impedance information and previous impedance information relating to the wireless transmitter, A processing system comprising means for generating an off-body characteristic, which indicates the probability that the processing system was off-body when the impedance information was determined, based on processing the impedance change information using a trained machine learning model.

28. The processing system according to claim 27, further comprising means for performing a power backoff operation on the wireless transmitter based on the off-body characteristics.

29. The generation of the aforementioned impedance change information is To generate a first value that represents the magnitude of the difference between the aforementioned impedance information and the previous impedance information, The processing system according to claim 27, comprising generating a second value that represents the direction of the difference between the impedance information and the previous impedance information.

30. The aforementioned trained machine learning model is used to provide proximity detection. The processing system does not include a capacitive sensor for proximity detection. The processing system according to claim 27.

Images