Receiver apparatus, and method for producing parameters for demodulation

Patent metadata

Field	Value
Publication number	US-2022337341-A1
Application number	US-202217673093-A
Country	US
Kind code	A1
Filing date	Feb 16, 2022
Priority date	Apr 14, 2021
Publication date	Oct 20, 2022
Grant date	—

Abstract

A method implemented in a computer system includes training a network, which is obtained by unfolding an iterative algorithm for demodulation or demodulation and decoding, using a machine learning technique with a loss function that takes into account non-Gaussianity of a log-likelihood ratio (LLR) distribution calculated from an output of the network. The method further includes producing a first set of learned parameters of that iterative algorithm.

First claim

1 . A method implemented in a computer system, the method comprising:
training a network, which is obtained by unfolding an iterative algorithm for demodulation or demodulation and decoding, using a machine learning technique with a first loss function that takes into account non-Gaussianity of a log likelihood ratio (LLR) distribution calculated from an output of the network; and producing, by the training, a first set of learned parameters of the iterative algorithm.

2 . The method according to claim 1 , wherein the first loss function is defined to include a term representing negentropy or kurtosis of the LLR distribution to measure the non-Gaussianity of the LLR distribution.

3 . The method according to claim 1 , wherein the first loss function is defined to further take into account a difference between training data and the output of the network.

4 . The method according to claim 3 , wherein the first loss function is defined as a weighted sum of a term representing the negentropy of the LLR distribution and a term representing a mean squared error between the training data and the output of the network.

5 . The method according to claim 3 , wherein the first loss function is defined as a weighted sum of a term representing the negentropy of the LLR distribution and a term representing cross entropy between the training data and the output of the network.

6 . The method according to claim 1 , wherein the iterative algorithm is an iterative Belief Propagation (BP) algorithm, and the first set of learned parameters includes one or any combination of a plurality of scaling factors, a plurality of damping factors, and a plurality of node selection factors.

7 . The method according to claim 1 , further comprising producing a second set of learned parameters of the iterative algorithm by training the network using a machine learning technique with a second loss function, wherein
the second loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function, the first set is used when a code rate is a first value or when a modulation order is a second value, and the second set is used when the code rate is lower than the first value or when the modulation order is higher than the second value.

8 . The method according to claim 1 , further comprising producing a third set of learned parameters of the iterative algorithm by training the network using a machine learning technique with a third loss function, wherein
the third loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function, the first set is used when a signal to noise power ratio (SNR) is a third value, and the third set is used when the SNR is less than the third value.

9 . The method according to claim 1 , further comprising producing a fourth set of learned parameters of the iterative algorithm by training the network using a machine learning technique with a fourth loss function, wherein
the fourth loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function, the first set is used for demodulation process on initial transmission, and the fourth set is used for demodulation process on retransmission.

10 . A non-transitory computer readable medium storing a program including instructions that, when loaded into a computer system, cause the computer system to perform a method comprising:
training a network, which is obtained by unfolding an iterative algorithm for demodulation or demodulation and decoding, using a machine learning technique with a first loss function that takes into account non-Gaussianity of a log likelihood ratio (LLR) distribution calculated from an output of the network; and producing, by the training, a first set of learned parameters of the iterative algorithm.

11 . A receiver apparatus comprising:
a memory storing one or more sets of learned parameters produced by a method as claimed in claim 1 ; and at least one processor configured to:
perform on a plurality of received signals an iterative algorithm that uses any of the one or more sets of learned parameters, and generate a plurality of log likelihood ratio (LLR) vectors corresponding to a plurality of transmitted symbols; and
perform error correction decoding using the plurality of LLR vectors to generate a plurality of decoded bit sequences.

12 . A receiver apparatus comprising:
a memory; and at least one processor coupled to the memory and configured to:
perform on a plurality of received signals an iterative algorithm that uses a first set of learned parameters, and generate a plurality of log likelihood ratio (LLR) vectors corresponding to a plurality of transmitted symbols; and
perform error correction decoding using the plurality of LLR vectors to generate a plurality of decoded bit sequences,
wherein the first set of learned parameters is a parameter set generated by training a network, which is obtained by unfolding the iterative algorithm, using a machine learning technique with a first loss function that takes into account non-Gaussianity of an LLR distribution calculated from an output of the network.

13 . The receiver apparatus according to claim 12 , wherein the at least one processor is configured to select between the first set and a second set of learned parameters for use in the iterative algorithm, depending on one or both of a modulation order and a code rate.

14 . The receiver apparatus according to claim 13 , wherein
the second set is generated by training the network using a machine learning technique with a second loss function; the second loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function, the first set is used when the code rate is a first value or when the modulation order is a second value, and the second set is used when the code rate is lower than the first value or when the modulation order is higher than the second value.

15 . The receiver apparatus according to claim 12 , wherein the at least one processor is configured to select between the first set and a second set of learned parameters for use in the iterative algorithm, depending on a signal to noise power ratio (SNR).

16 . The receiver apparatus according to claim 15 , wherein
the third set is generated by training the network using a machine learning technique with a third loss function, the third loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function, the first set is used when the SNR is a third value, and the third set is used when the SNR is less than the third value.

17 . The receiver apparatus according to claim 12 , wherein the at least one processor is configured to use the first set for demodulation process on initial transmission and to use a fourth set of learned parameters for demodulation process on retransmission.

18 . The receiver apparatus according to claim 17 , wherein
the fourth set is generated by training the network using a machine learning technique with a fourth loss function, and the fourth loss function is defined to take into account the non-Gaussianity of the LLR distribution more deeply than the first loss function.

19 . A method performed by a receiver apparatus, the method comprising:
performing on a plurality of received signals an iterative algorithm that uses a

Assignees

• Nec Corp
• Univ Osaka
• The Doshisha

Inventors

• Muraoka Kazushi
• Doi Takanobu
• Ishii Naoto
• Takahashi Takumi
• Sampei Seiichi
• Ibi Shinsuke

Classifications

• H04L1/005 Primary

  Iterative decoding, including iteration between signal detection and decoding operation · CPC title

• H04B7/0413

  MIMO systems · CPC title

• H04B7/0452

  Multi-user MIMO systems · CPC title

• H04L1/0045 Primary

  Arrangements at the receiver end · CPC title