IP Library Granted Patent US 11,901,916
Granted Patent B2
US 11,901,916 · App. 17/641,488 · Granted Feb 13, 2024

Soft-decision decoding

Inventor: Terence Dodgson (Stevenage, GB)
Assignee: Airbus Defence and Space Limited
H03M13/3927G06N20/00H03M13/1111H03M13/458H03M13/3994H03M13/4138H04L1/005
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,901,916
App. No.
17/641,488
Granted
Feb 13, 2024
Kind
B2
Abstract

A method of soft-decision decoding including training a machine learning agent with communication signal training data; providing to the trained machine learning agent a signal that has been received via a communications channel; operating the machine learning agent to determine respective probabilities that the received signal corresponds to each of a plurality of symbols; and, based on the determined probabilities, performing soft decision decoding on the received signal.

Claims (40)

1. A method of soft-decision decoding, the method comprising:

training a machine learning agent with communication signal training data;

providing to the trained machine learning agent a signal that has been received via a communications channel;

operating the machine learning agent to determine respective probabilities that the received signal corresponds to each of a plurality of symbols; and

based on the determined respective probabilities, performing soft decision decoding on the received signal,

wherein operating the machine learning agent comprises determining a normalized probability pair for each symbol bit, a normalized probability pair comprising a first probability that a respective symbol bit is a zero and a second probability that the respective symbol bit is a one, and

wherein the first probability and the second probability of the normalized probability pair sum to 1.

2. The method according to claim 1 , comprising generating the training data by applying a noise function to a set of known symbols.

3. The method according to claim 2 , wherein a magnitude of the applied noise function is varied across the training data.

4. The method according to claim 3 , wherein varying the magnitude of the applied noise function comprises varying a signal to noise ratio between 0 dB and 9 dB.

5. The method according to claim 2 , wherein a magnitude of the applied noise function is varied between each symbol of the training data.

6. The method according to claim 1 , wherein the training data comprises one or more pilot signals.

7. The method according to claim 1 , wherein the training data comprises one or more previously transmitted communications signals.

8. The method according to claim 7 , wherein the previously transmitted communications signals were transmitted via the communications channel.

9. The method according claim 1 , comprising, subsequent to the performing, further training the machine learning agent with further communication signal training data.

10. The method according to claim 9 , wherein the further communication signal training data comprises one or more pilot signals.

11. The method according to claim 9 , wherein the further training is performed periodically.

12. The method according to claim 1 , comprising determining one or more signal characteristics for each of the plurality of symbols,

wherein the operating is carried out based on the determined signal characteristics.

13. The method according to claim 12 , wherein the one or more signal characteristics comprise received power values for each of the plurality of symbols.

14. The method according to claim 13 , comprising performing a Fourier transform on the received signal to determine the received power values.

15. The method according to claim 12 , wherein the one or more signal characteristics comprise correlation values for each of the plurality of symbols.

16. The method according to claim 12 , comprising:

converting the one or more signal characteristics into a vector; and

providing the vector to the machine learning agent as an input.

17. The method according to claim 1 , wherein the received signal comprises a frequency shift keying modulated signal.

18. An apparatus for use in soft-decision decoding, the apparatus being configured to:

train a machine learning agent with communication signal training data;

provide to the trained machine learning agent a signal that has been received via a communications channel;

operate the machine learning agent to determine respective probabilities that the received signal corresponds to each of a plurality of symbols; and

based on the determined respective probabilities, perform soft decision decoding on the received signal,

wherein operating the machine learning agent comprises determining a normalized probability pair for each symbol bit, a normalized probability pair comprising a first probability that a respective symbol bit is a zero and a second probability that the respective symbol bit is a one, and

wherein the first probability and the second probability of the normalized probability pair sum to 1.

19. A non-transitory computer-readable storage media comprising a set of instructions which, when executed by a computerized device, cause the computerized device to perform a method of soft-decision decoding, the method comprising:

training a machine learning agent with communication signal training data;

providing to the trained machine learning agent a signal that has been received via a communications channel;

operating the machine learning agent to determine respective probabilities that the received signal corresponds to each of a plurality of symbols; and

based on the determined respective probabilities, performing soft decision decoding on the received signal,

wherein operating the machine learning agent comprises determining a normalized probability pair for each symbol bit, a normalized probability pair comprising a first probability that a respective symbol bit is a zero and a second probability that the respective symbol bit is a one, and

wherein the first probability and the second probability of the normalized probability pair sum to 1.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded May 3, 2022
From: DODGSON, TERENCE
To: AIRBUS DEFENCE AND SPACE LIMITED
Reel/Frame 059847/0080 →
Priority Claims (1)
GB 1913178 · Sep 12, 2019 · national
Continuity (1)
Related Publication 20220321149A1 · Oct 6, 2022