WO2021110054A1 - Procédé de décodage de canal de décomposition double de pénalité assisté par réseau neuronal multicouche - Google Patents
Procédé de décodage de canal de décomposition double de pénalité assisté par réseau neuronal multicouche Download PDFInfo
- Publication number
- WO2021110054A1 WO2021110054A1 PCT/CN2020/133414 CN2020133414W WO2021110054A1 WO 2021110054 A1 WO2021110054 A1 WO 2021110054A1 CN 2020133414 W CN2020133414 W CN 2020133414W WO 2021110054 A1 WO2021110054 A1 WO 2021110054A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- penalty
- neural network
- polyhedron
- dual decomposition
- decoding
- Prior art date
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 61
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 59
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000009977 dual effect Effects 0.000 title claims abstract description 39
- 238000013507 mapping Methods 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 19
- 238000007476 Maximum Likelihood Methods 0.000 claims abstract description 13
- 210000002569 neuron Anatomy 0.000 claims description 9
- 230000003190 augmentative effect Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000000750 progressive effect Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 2
- 238000010801 machine learning Methods 0.000 abstract description 2
- 230000001131 transforming effect Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 9
- 230000001537 neural effect Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/01—Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
- H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
- H03M13/1105—Decoding
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
- H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
- H03M13/1148—Structural properties of the code parity-check or generator matrix
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
- H04L1/0045—Arrangements at the receiver end
- H04L1/0054—Maximum-likelihood or sequential decoding, e.g. Viterbi, Fano, ZJ algorithms
Definitions
- This application belongs to the field of wireless communication channel coding and decoding, and relates to a penalty-dual decomposition channel decoding method assisted by a multilayer neural network.
- Channel decoding is a question of how to make a decision on the received symbol message.
- the message received by the sink is not necessarily the same as the message sent by the source, and the sink needs to know which source message is sent by the source at this time, so the message received by the sink needs to be based on a certain This kind of rule decides to correspond to a certain one of the source symbol message set.
- the linear programming (LP) decoder is based on the linear relaxation of the original maximum likelihood decoding problem, and is a popular decoding technique for binary linear codes. Because the linear programming decoder has a strong guarantee for the decoding performance in theory, it has received extensive attention from the academic and industrial circles, especially the decoding of low-density parity-check (LDPC) codes. However, compared with the classic Belief Propagation (BP) decoder, the LP decoder has higher computational complexity and lower error correction performance in the low signal-to-noise ratio (SNR) area.
- SNR signal-to-noise ratio
- the deep learning method since the deep learning method has been successfully applied to many other fields, such as image processing, natural language processing, etc., it has also begun to be applied to the field of wireless communication as a potential technology, such as signal detection, channel estimation, and channel coding. Wait.
- the purpose of this application is to improve the decoding performance in the channel decoding process, and propose a penalty dual decomposition channel decoding method assisted by a multilayer neural network.
- This application first proposes to use the penalty dual decomposition method to solve the maximum likelihood channel decoding problem to further improve the decoding performance.
- the neural network is introduced into the iterative polyhedron mapping in this method to reduce the number of iterations to reduce decoding delay.
- a multi-layer neural network-assisted penalty-dual decomposition channel decoding method includes:
- step 2 2. Introduce the basic polyhedron into the parity check constraint condition in step 1, and transform the maximum likelihood decoding problem into a decoding optimization problem based on the parity check polyhedron;
- step 4 Use the multi-layer neural network-assisted penalty dual decomposition channel decoder obtained in step 4 to perform online real-time channel decoding.
- the penalty-dual decomposition channel decoder obtained in step 3 has a check polyhedron mapping, and a check polyhedron mapping operation It is the most time-consuming part. Iterative verification of polyhedral mapping causes decoding delays due to iterations.
- the polyhedral mapping based on the multilayer neural network proposed in this application reduces the decoding delay by reducing the number of iterations.
- the multilayer neural network is composed of three layers: an input layer, an output layer and a hidden layer; the input layer contains d j neurons, and the hidden layer contains A neuron, the output layer contains 1 neuron, the hidden layer and the output layer are followed by an activation function, which is defined as:
- the weights and biases in the network are the network parameters that need to be learned; therefore, the mapping between the input and output realized by the multilayer neural network is expressed as:
- the loss function is:
- ⁇ W a ,w b ⁇ is quantified as It is a set of natural numbers; the multiplication operation can be cancelled or transformed into a less complex shift operation.
- the calculation method of the penalty dual decomposition channel decoder assisted by the multilayer neural network is as follows:
- sgn( ⁇ ) is a step function. If the number of elements with 1 in
- ⁇ i: ⁇ i 1 ⁇
- This application makes full use of the penalty function decomposition method to solve the maximum likelihood channel decoding optimization problem, and improves the decoding performance.
- machine learning methods are used to further optimize the iterative polyhedral mapping method in the channel decoding method based on penalty function decomposition, and a neural network is introduced to reduce the number of iterations to reduce the decoding delay.
- the network is very easy to train, and the training process requires less training time and hardware platform.
- Figure 1 is a CPP-net structure diagram based on [96,48]MacKey 96.33.964 code
- Figure 2 shows the BP decoder, the decoder based on the alternating direction multiplier method (ADMM L2), the penalty dual decomposition channel decoding method (PDD), and the multi-layer neural network assisted penalty dual decomposition channel decoding method (PDD with neural CPP) BLER curve diagram of code group error rate in Rayleigh channel environment.
- a multi-layer neural network-assisted penalty-dual decomposition channel decoding method proposed for this system includes the following steps:
- the method specifically includes the following steps:
- Step 1 For a binary linear code of length N Each codeword is specified by the M ⁇ N parity check matrix H, Represents the transmitted codeword, and y represents the received signal; the maximum likelihood decoding problem is constructed based on channel decoding, expressed as the form described in the following formula (1):
- Pr( ⁇ ) represents the conditional probability
- Step 2 Introduce a basic polyhedron into the parity check constraint condition, and relax the maximum likelihood decoding problem (1) into the following linear constraint problem:
- P j represents a d j ⁇ N selection matrix, which is used to select the elements in the vector x participating in the j-th check equation, Represents the parity check polyhedron with degree d j, the expression is:
- Step 3 Use the penalty dual decomposition method to introduce a set of auxiliary variables Convert the constraints in (4) into the following equivalent forms:
- the block continuous upper bound minimization algorithm BSUM is used to process the inner loop, and the dual variable and the penalty parameter ⁇ m are updated in the outer loop value;
- the BSUM algorithm processing steps are as follows:
- ⁇ i is a vector The i-th element of
- ⁇ [0,1] represents the Euclidean mapping on the interval [0,1]
- the dual variable is updated by the following formula:
- the decoding method in this step is called a penalty-dual decomposition channel decoder, and the penalty-dual decomposition channel decoder has a check polyhedron map.
- Step 4 Design the verification polyhedron mapping calculation based on the multilayer neural network It is divided into the following steps:
- sgn( ⁇ ) is a step function. If the number of elements with 1 in
- ⁇ i: ⁇ i 1 ⁇
- the CPP-net consists of three layers: an input layer, an output layer and a hidden layer; the input layer contains d j neurons, and the hidden layer contains A neuron, the output layer contains 1 neuron, the hidden layer and the output layer are followed by an activation function, which is defined as:
- the weights and biases in the network are the network parameters that need to be learned; therefore, the mapping between input and output realized by CPP-net is expressed as:
- check polyhedron mapping based on multi-layer neural network The calculation method described in this step is called check polyhedron mapping based on multi-layer neural network, and its implementation code is:
- Step 5 Use the penalty dual decomposition channel decoder assisted by the multilayer neural network to perform online real-time channel decoding.
- Figure 2 shows the BP decoder, the decoder based on the alternating direction multiplier method (ADMM L2), the penalty dual decomposition channel decoding method (PDD), and the penalty dual assisted by the multilayer neural network in the Rayleigh channel environment Decompose the code group error rate BLER of the channel decoding method (PDD with neural CPP).
- the penalty dual decomposition channel decoding method (PDD) and the multi-layer neural network-assisted penalty dual decomposition channel decoding method (PDD with neural CPP) have obtained the best translation. Code performance.
- Table 1 compares the average iteration times (iterM) of the iterative polyhedron mapping algorithm and the neural network-based polyhedron mapping algorithm. It can be seen that the neural network-based polyhedron mapping algorithm included in this application can effectively reduce the number of iterations.
- this application proposes a multi-layer neural network-assisted penalty dual decomposition channel decoding method.
- the above are only specific implementations for specific applications, but the true spirit and scope of this application are not limited to this. Any person skilled in the art can modify, equivalently replace, improve, etc., to achieve channel decoding for different applications. method. These modifications, equivalent replacements and improvements also fall within the protection scope of the claims of this application.
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Error Detection And Correction (AREA)
Abstract
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911238262.0A CN110995277B (zh) | 2019-12-06 | 2019-12-06 | 一种多层神经网络辅助的罚对偶分解信道译码方法 |
CN201911238262.0 | 2019-12-06 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021110054A1 true WO2021110054A1 (fr) | 2021-06-10 |
Family
ID=70090843
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2020/133414 WO2021110054A1 (fr) | 2019-12-06 | 2020-12-02 | Procédé de décodage de canal de décomposition double de pénalité assisté par réseau neuronal multicouche |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110995277B (fr) |
WO (1) | WO2021110054A1 (fr) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113824478A (zh) * | 2021-10-11 | 2021-12-21 | 北京邮电大学 | 离散透镜天线阵列辅助的宽带毫米波多用户大规模mimo上行频谱效率优化方法 |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110995277B (zh) * | 2019-12-06 | 2021-06-01 | 浙江大学 | 一种多层神经网络辅助的罚对偶分解信道译码方法 |
CN117335815B (zh) * | 2023-11-29 | 2024-03-15 | 广东工业大学 | 基于改进原模图神经译码器的训练方法及装置 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102122962A (zh) * | 2011-04-01 | 2011-07-13 | 山东大学 | 基于预测-修正原对偶内点法的ldpc码的lp译码器 |
CN102130745A (zh) * | 2011-04-07 | 2011-07-20 | 山东大学 | 一种改进的ldpc码的线性规划译码方法 |
CN105959015A (zh) * | 2016-04-22 | 2016-09-21 | 西安电子科技大学 | 基于最小多面体模型的ldpc码线性规划译码方法 |
CN108964669A (zh) * | 2018-07-06 | 2018-12-07 | 西安电子科技大学 | 基于度分解和交替乘子法的ldpc码二次规划译码方法 |
CN110995277A (zh) * | 2019-12-06 | 2020-04-10 | 浙江大学 | 一种多层神经网络辅助的罚对偶分解信道译码方法 |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
ATE336107T1 (de) * | 2003-04-30 | 2006-09-15 | Freescale Semiconductor Inc | Verfahren und einrichtung zur digital- analog/analog-digital wandlung mit reduziertem energieverbrauch |
US20120051452A1 (en) * | 2010-09-01 | 2012-03-01 | Nec Laboratories America, Inc. | Modified coded hybrid subcarrier amplitude phase polarization modulation |
US8650459B1 (en) * | 2011-02-16 | 2014-02-11 | Sk Hynix Memory Solutions Inc. | Soft input, soft ouput mappers and demappers for block codes |
CN103581924B (zh) * | 2013-11-22 | 2017-02-08 | 北京无线电测量研究所 | 一种提高蜂窝网通信系统中双向中继传输容量的方法 |
EP3680820B1 (fr) * | 2015-10-04 | 2021-12-01 | Atomwise Inc. | Procédé d'application de réseau convolutionnel à des données spatiales |
CN105792258B (zh) * | 2016-04-13 | 2019-02-22 | 浙江理工大学 | 无线传感网中速率和可靠性协同的跨层优化方法 |
CN107689801B (zh) * | 2017-09-07 | 2019-10-25 | 西安电子科技大学 | Ldpc码admm迭代译码的早停止方法 |
CN108199721B (zh) * | 2017-12-22 | 2019-10-25 | 西安电子科技大学 | 基于badmm的低密度奇偶检验码线性规划译码方法 |
CN109167600A (zh) * | 2018-10-12 | 2019-01-08 | 哈尔滨工业大学 | 基于深度置信网络的抗音调干扰ldpc码的译码方法 |
-
2019
- 2019-12-06 CN CN201911238262.0A patent/CN110995277B/zh active Active
-
2020
- 2020-12-02 WO PCT/CN2020/133414 patent/WO2021110054A1/fr active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102122962A (zh) * | 2011-04-01 | 2011-07-13 | 山东大学 | 基于预测-修正原对偶内点法的ldpc码的lp译码器 |
CN102130745A (zh) * | 2011-04-07 | 2011-07-20 | 山东大学 | 一种改进的ldpc码的线性规划译码方法 |
CN105959015A (zh) * | 2016-04-22 | 2016-09-21 | 西安电子科技大学 | 基于最小多面体模型的ldpc码线性规划译码方法 |
CN108964669A (zh) * | 2018-07-06 | 2018-12-07 | 西安电子科技大学 | 基于度分解和交替乘子法的ldpc码二次规划译码方法 |
CN110995277A (zh) * | 2019-12-06 | 2020-04-10 | 浙江大学 | 一种多层神经网络辅助的罚对偶分解信道译码方法 |
Non-Patent Citations (2)
Title |
---|
CHEN ZI-QIANG, OUYANG SHAN ,LI MIN-ZHENG, ZANG LAN, XIAO HAI-LIN: "A Hybrid Decoding Algorithm Based on Improved Linear Programming for LDPC Codes", JOURNAL OF CIRCUITS AND SYSTEMS, vol. 18, no. 1, 1 February 2013 (2013-02-01), pages 107 - 112, XP055819140, ISSN: 1007-0249 * |
HAIYANG LIU ; WENZE QU ; BIN LIU ; JIE CHEN: "On the Decomposition Method for Linear Programming Decoding of LDPC Codes", IEEE TRANSACTIONS ON COMMUNICATIONS, IEEE SERVICE CENTER, PISCATAWAY, NJ. USA., vol. 58, no. 12, 1 December 2010 (2010-12-01), PISCATAWAY, NJ. USA., pages 3448 - 3458, XP011340171, ISSN: 0090-6778 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113824478A (zh) * | 2021-10-11 | 2021-12-21 | 北京邮电大学 | 离散透镜天线阵列辅助的宽带毫米波多用户大规模mimo上行频谱效率优化方法 |
CN113824478B (zh) * | 2021-10-11 | 2023-07-04 | 北京邮电大学 | 离散透镜天线阵列辅助的宽带毫米波多用户大规模mimo上行频谱效率优化方法 |
Also Published As
Publication number | Publication date |
---|---|
CN110995277A (zh) | 2020-04-10 |
CN110995277B (zh) | 2021-06-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021110054A1 (fr) | Procédé de décodage de canal de décomposition double de pénalité assisté par réseau neuronal multicouche | |
WO2021093866A1 (fr) | Procédé de décodage de canal d'apprentissage profond basé sur un procédé de direction alternée de multiplicateurs | |
CN110474716B (zh) | 基于降噪自编码器的scma编解码器模型的建立方法 | |
Zhang et al. | A unified multi-task semantic communication system for multimodal data | |
Choukroun et al. | Error correction code transformer | |
CN109361404A (zh) | 一种基于半监督深度学习网络的ldpc译码系统及译码方法 | |
CN111835364B (zh) | 一种极化码的低复杂度神经bp译码方法 | |
Seo et al. | Semantics-native communication via contextual reasoning | |
CN115309869A (zh) | 一种一对多的多用户语义通信模型及通信方法 | |
Nemati et al. | VQ-VAE empowered wireless communication for joint source-channel coding and beyond | |
Tseng et al. | Deep learning based decoding for polar codes in Markov Gaussian memory impulse noise channels | |
Lugosch | Learning algorithms for error correction | |
Nachmani et al. | Autoregressive belief propagation for decoding block codes | |
Aiyer | Robust image compression using Gauss mixture models | |
US8571092B2 (en) | Interconnect coding method and apparatus | |
CN111049531B (zh) | 一种基于分段线性惩罚函数的交替方向乘子法的深度学习信道译码方法 | |
CN113872610A (zh) | 一种ldpc码神经网络训练、译码方法及其系统 | |
Tang et al. | Normalized Neural Network for Belief Propagation LDPC Decoding | |
CN108737298B (zh) | 一种基于图像处理的scma盲检测方法 | |
Chen et al. | Boosting belief propagation for LDPC codes with deep convolutional neural network predictors | |
Liu | Task-oriented explainable semantic communication based on semantic triplets | |
Stark et al. | Neural information bottleneck decoding | |
Fonał et al. | Fast recursive nonnegative standard and hierarchical Tucker decomposition | |
Ivanov et al. | Method of critical set construction for successive cancellation list decoder of polar codes based on deep learning of neural networks | |
Devamane et al. | Performance analysis of neural network-based polar decoding algorithms with different code rates |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20895541 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20895541 Country of ref document: EP Kind code of ref document: A1 |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20895541 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 30.01.2023) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20895541 Country of ref document: EP Kind code of ref document: A1 |