在无线通信系统中，信道编译码技术被用来抵抗信息传输过程中噪声的影响，以降低误码率为目标，进而改善通信系统的性能。然而各种编译码方案的优缺点和复杂度均不同，准循环-低密度奇偶校验码（Quasi-Cyslic Low-Density Parity-Check Codes，QC-LDPC）只有在码长较长时性能才能凸显出来，但其译码算法复杂度与码长成线性关系。随着深度学习技术的发展，神经网络为解决传统信道译码中存在的难题提供了新的思路。因此，论文基于卷积神经网络研究了QC-LDPC的译码算法。主要工作有以下几个方面：

（1）传统的QC-LDPC在码长增加时，译码算法的计算复杂度也随之增加。针对这一问题，提出了一种基于卷积神经网络的QC-LDPC译码方法，该方法能够在码长和译码算法复杂度之间进行折中。采用卷积神经网络来取代传统的QC-LDPC译码算法，结合信道译码需求，设计了DenseNet中的稠密块和过渡层，在QC-LDPC编码之后的数据集上进行了仿真实验。比较了该方法在不同传统译码方法、码长、码率、调制方式、神经网络结构和信道环境下的性能。结果表明，基于卷积神经网络的QC-LDPC译码算法比传统的QC-LDPC译码算法误码率性能更好。

（2）为了进一步降低模型的训练和推理时间，设计了轻量型卷积神经网络模型来实现QC-LDPC的译码方法，其中，MobileNetV3通过引入一个浅层特征提取模块，来获取更多的浅层特征信息，并借鉴了DenseNet中密集连接的思想进行深度特征的提取。同时引入了解耦全连接注意力机制以加强特征连接。比较了该方法在不同码长、码率、神经网络结构和信道环境下的性能。仿真结果表明，基于轻量型卷积神经网络的QC-LDPC译码方法可实现较低的系统误码率，且模型更小，复杂度更低。

论文将卷积神经网络与QC-LDPC译码相结合，充分发挥卷积神经网络能够从数据中学习到更深层次的特征的优点，提高了QC-LDPC码字的译码性能，也对其他译码方法的性能提升有一定的借鉴作用。

In wireless communication systems, channel decoding technology is used to resist the influence of noise during information transmission, with the goal of reducing the bit error rate, thereby improving the performance of communication systems. However, the advantages, disadvantages and complexity of various encoding and decoding are different, and the performance of Quasi-cyslic Low-density Parity-Check Codes (QC-LDPC) can only be highlighted when the code length is long, but the complexity of the decoding algorithm is linearly related to the code length. With the development of deep learning Xi technology, neural networks provide new ideas for solving the problems existing in traditional channel decoding. Therefore, based on convolutional neural network, the decoding algorithm of QC-LDPC was studied. The main work includes the following aspects:

(1)In traditional QC-LDPC, when the code length increases, the computational complexity of the decoding algorithm also increases. In order to solve this problem, a QC-LDPC decoding method based on convolutional neural network was proposed, which can make a compromise between code length and decoding algorithm complexity. Convolutional neural network is used to replace the traditional QC-LDPC decoding algorithm, combined with the channel decoding requirements, the dense block and transition layer in DenseNet are designed, and simulation experiments are carried out on the dataset after QC-LDPC encoding. The performance of the proposed method in different traditional decoding methods, code length, code rate, modulation mode, neural network structure and channel environment is compared. The results show that the QC-LDPC decoding algorithm based on convolutional neural network has better bit error rate performance than the traditional QC-LDPC decoding algorithm.

(2)In order to further reduce the training and inference time of the model, a lightweight convolutional neural network model was designed to realize the decoding method of QC-LDPC, in which MobileNetV3 introduced a shallow feature extraction module to obtain more shallow feature information, and borrowed the idea of dense connection in DenseNet for deep feature extraction. At the same time, the decoupling of the full connection attention mechanism is introduced to strengthen the feature connection. The performance of the proposed method in different code lengths, code rates, neural network structures and channel environments is compared. The simulation results show that the QC-LDPC decoding method based on lightweight convolutional neural network can achieve a lower system bit error rate, and the model is smaller and less complex.

This paper combines convolutional neural network with QC-LDPC decoding, gives full play to the advantages of convolutional neural network to learn deeper features from data, improves the decoding performance of QC-LDPC codewords, and also has a certain reference effect on the performance improvement of other decoding methods.

[1]Gallager R. Low-Density Parity-Check Codes[J]. IRE Transactions on Information Theory, 1962, 8(1): 21-28.

[2]Zhao Z, Jiao X, Mu J, et al. Momentum-Based Symbol Flipping Decoding Algorithms for Non-Binary LDPC Codes[J]. IEEE Transactions on Vehicular Technology, 2023: 1-6.

[3]Lu T, He X, Kang P, et al. Parity Check Matrix Partitioning for Layered Decoding of QC-LDPC Codes[C]//2021 IEEE Globecom Workshops (GC Wkshps). IEEE, 2021: 1-6.

[4]Zhang L, Wang J. Construction of QC-LDPC Codes from Sidon Sequence Using Permutation and Segmentation[J]. IEEE Communications Etters, 2022, 26(8): 1710-1714.

[5]Lu T, He X, Kang P, et al. Parity-Check Matrix Partitioning for Efficient Layered Decoding of QC-LDPC Codes[J]. IEEE Transactions on Communications, 2023: 1-1.

[6]Zhao C, Yang F, Waweru D.K, et al. Optimized Design of Distributed Quasi-Cyclic LDPC Coded Spatial Modulation[J]. Sensors, 2023, 23(7): 1424-8220.

[7]陈发堂,王永航,张翰卿.改进的非规则QC-LDPC译码算法和结构[J].光通信研究, 2020(06): 1-4+16.

[8]Weng Z, Qin Z, Tao X, et al. Deep Learning Enabled Semantic Communications with Speech Recognition and Synthesis[J]. IEEE Transactions on Wireless Communications, 2023: 1-1.

[9]Tang X, Reviriego P, Tang W, et al. Joint Learning and Channel Coding for Error-Tolerant IoT Systems Based on Machine Learning[J]. IEEE Transactions on Artificial Intelligence, 2023: 1-12.

[10]Yasir M, Jianhua W, Mingming X, et al. Ship Detection Based on Deep Learning Using SAR Imagery: A Systematic Literature Review[J]. Soft Computing, 2023, 27(1): 63-84.

[11]Gu W, Bai S, Kong L. A Review on 2D Instance Segmentation Based on Deep Neural Networks[J]. Image and Vision Computing, 2022, 120: 104401.

[12]Erpek T, O’Shea T J, Sagduyu Y E, et al. Deep Learning for Wireless Communications[J]. Development and Analysis of Deep Learning Architectures, 2020, 223-266.

[13]Xiao Z, Li L, Xu J, et al. Construction of Protograph LDPC Codes Based on the Convolution neural Network[J]. China Communications, 2023, 1-9.

[14]He J. A Deep Learning-Aided Post-Processing Scheme to Lower the Error Floor of LDPC Codes[C]//2020 IEEE 20th International Conference on Communication Technology (ICCT). IEEE, 2020: 1587-1592.

[15]赵梦.基于深度学习的LDPC码ADMM译码算法研究[D].西安电子科技大学, 2020.

[16]Verma A, Shrestha R. Low Computational-Complexity SOMS-Algorithm and High-Throughput Decoder Architecture for QC-LDPC Codes[J]. IEEE Transactions on Vehicular Technology, 2023, 72(1): 66-80.

[17]Sun Y, Huang P, Zhang Y, et al. Multiple Flipping Strategy LISBF for LDPC Codes[C]//2023 6th World Conference on Computing and Communication Technologies (WCCCT). IEEE, 2023: 61-67.

[18]Li Y, Ye N, et al. High Throughput Priority-Based Layered QC-LDPC Decoder with Double Update Queues for Mitigating Pipeline Conflicts[J]. Sensors, 2022, 22(9): 3508.

[19]Mahdi A, Paliouras V. A Low Complexity-High Throughput QC-LDPC Encoder[J]. IEEE Transactions on Signal Processing, 2014, 62(10): 2696-2708.

[20]Song L, Yu S, Huang Q. Low-Density Parity-Check Codes: Highway to Channel Capacity[J]. China Communications, 2023, 20(2): 235-256.

[21]Wang X, Ge T, Li J, et al. Efficient Multi‐Rate Encoder of QC‐LDPC Codes Based on FPGA for WIMAX Standard[J]. Chinese Journal of Electronics, 2017, 26(2): 250-255.

[22]Yuan R J, Bai B M, Tong S. FPGA-Based Design of LDPC Encoder with Throughput Over 10 Gbps[J]. Journal of Electronics and Information Technology, 2011, 33(12): 2942-2947.

[23]He Z, Zhao Q, Xu H, et al. An Encoder with Speed Over 40Gbps for RC LDPC Codes with Rates Up to 0.96[J]. Chinese Journal of Electronics, 2016, 25(5): 921-927.

[24]Bai H, Zhai K, Liu G, et al. A Design Method of High Parallelism QC-LDPC Decoder Based on FPGA[C]//2022 5th International Conference on Information Communication and Signal Processing (ICICSP). IEEE, 2022: 441-447.

[25]El Ouakili H, Touati H, Kadi A, et al. New Low-Density-Parity-Check Decoding Approach Based on the Hard and Soft Decisions Algorithms[J]. International Journal of Electrical and Computer Engineering, 2023, 13(2): 1639.

[26]Kschischang F R, Frey B J, Loeliger H A. Factor Graphs and the Sum-Product Algorithm[J]. IEEE Transactions on Information Theory, 2001, 47(2): 498-519.

[27]Petrović V L, Marković M M, El Mezeni D M, et al. Flexible High Throughput QC-LDPC Decoder with Perfect Pipeline Conflicts Resolution and Efficient Hardware Utilization[J]. IEEE Transactions on Circuits and Systems I: Regular Papers, 2020, 67(12): 5454-5467.

[28]Ma L, Sham C W, Zhan J, et al. Implementation for JSCC Scheme Based on QC-LDPC Codes[C]//2022 IEEE 11th Global Conference on Consumer Electronics (GCCE). IEEE, 2022: 292-295.

[29]袁瑞佳,白宝明.基于FPGA的LDPC码编译码器联合设计[J].电子与信息学报, 2012, 34(01): 38-44.

[30]张顺根,仰枫帆.基于FPGA的随机构造QC-LDPC分层译码器设计[J].无线电通信技术, 2015, 41(01): 41-45.

[31]刘振,黎勇.基于QR码构造的QC-LDPC码译码器设计与实现[J].重庆邮电大学学报(自然科学版), 2020, 32(03): 419-425.

[32]张晓芳,黎勇.GPU的QC-LDPC码译码器设计与实现[J].重庆邮电大学学报(自然科学版), 2021, 33(04): 577-583.

[33]徐斌,贺玉成.高吞吐量QC-LDPC码分层译码设计[J].计算机工程, 2019, 45(07): 121-125+133.

[34]Hsieh K, Lin Y W, Chu S I, et al. A Simple Neural-Network-Based Decoder for Short Binary Linear Block Codes[J]. Applied Sciences, 2023, 13(7): 4371.

[35]Wang L, Chen S, Nguyen J, et al. Neural-Network-Optimized Degree-Specific Weights for LDPC Minsum Decoding[J], 2021.

[36]Li G, Yu X, Luo Y, et al. A Bottom‐Up Design Methodology of Neural Min‐Sum Decoders for LDPC Codes[J]. IET Communications, 2023, 17(3): 377-386.

[37]Zhang J, Jiang W, Zhou J, et al. An Iterative BP-CNN Decoder for Optical Fiber Communication Systems[J]. 2023.

[38]Wadayama T, Takabe S. Proximal Decoding for LDPC Codes[J]. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, 2023, 106(3): 359-367.

[39]刘恒燕,张立民,闫文君,等.基于WBP-CNN算法的LDPC译码[J].系统工程与电子技术, 2022, 44(03): 1030-1035.

[40]冯俊鹏.LDPC码神经网络译码模型研究[D].西南交通大学, 2021.

[41]Gui G, Liu M, Tang F, et al. 6G: Opening New Horizons for Integration of Comfort, Security, and Intelligence[J]. IEEE Wireless Communications, 2020, 27(5): 126-132.

[42]You X, Wang C X, Huang J, et al. Towards 6G Wireless Communication Networks: Vision, Enabling Technologies, and New Paradigm Shifts[J]. Science China Information Sciences, 2021, 64: 1-74.

[43]Guo X, Chang T H, Wang Y. Model-Driven Deep Learning ADMM Decoder for Irregular Binary LDPC Codes[J]. IEEE Communications Letters, 2023, 27(2): 571-575.

[44]Deng P, Zhang T, Ma B, et al. Research on Blind Recognition Algorithm of Channel Coding Based on One-Dimensional Convolutional Neural Network Under the Low SNR Regime[J]. Neural Processing Letters, 2023: 1-21.

[45]Wang Q, Liu Q, Wang S, et al. Normalized Min-Sum Neural Network for LDPC Decoding[J]. IEEE Transactions on Cognitive Communications and Networking, 2022: 70-81.

[46]Liang F, Shen C, Wu F. An Iterative BP-CNN Architecture for Channel Decoding[J]. IEEE Journal of Selected Topics in Signal Processing, 2018, 12(1): 144-159.

[47]Li J, Zhao X, Fan J, et al. A Double-CNN BP Decoder on Fast Fading Channels Using Correlation Information[C]//2019 IEEE/CIC International Conference on Communications in China (ICCC). IEEE, 2019: 53-58.

[48]Karami A R, Attari M A, Tavakoli H. Multi Layer Perceptron Neural Networks Decoder for LDPC Codes[C]//2009 5th International Conference on Wireless Communications, Networking and Mobile Computing. IEEE, 2009: 1-4.

[49]Wang X, Li J, Chang H, et al. Optimization Design of Polar-LDPC Concatenated Scheme Based on Deep Learning[J]. Computers & Electrical Engineering, 2020, 84: 106636.

[50]O’shea T, Hoydis J. An Introduction to Deep Learning for the Physical Layer[J]. IEEE Transactions on Cognitive Communications and Networking, 2017, 3(4): 563-575.

[51]Nachmani E, Be'ery Y, Burshtein D. Learning to Decode Linear Codes Using Deep Learning[C]//2016 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton). IEEE, 2016: 341-346.

[52]Gruber T, Cammerer S, Hoydis J, et al. On Deep Learning-Based Channel Decoding[C]//2017 51st Annual Conference on Information Sciences and Systems (CISS). IEEE, 2017: 1-6.

[53]林景栋,吴欣怡,柴毅,等.卷积神经网络结构优化综述[J].自动化学报, 2020, 46(01): 24-37.

[54]Szegedy C, Liu W, Jia Y, et al. Going Deeper with Convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2015: 1-9.

[55]He K, Zhang X, Ren S, et al. Deep Residual Learning for Image Recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016: 770-778.

[56]Huang G, Liu Z, Van Der Maaten L, et al. Densely Connected Convolutional Networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 4700-4708.

[57]Howard A, Sandler M, Chu G, et al. Searching for Mobilenetv3[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019: 1314-1324.

[58]Howard A G, Zhu M, Chen B, et al. Mobilenets: Efficient Convolutional Neural Networks for Mobile Vision Applications[J]. Computer Vision and Pattern Recognition, 2017.

[59]Sandler M, Howard A, Zhu M, et al. Mobilenetv2: Inverted Residuals and Linear Bottlenecks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2018: 4510-4520.

[60]Han K, Wang Y, Tian Q, et al. Ghostnet: More Features from Cheap Operations[C]// Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition. 2020: 1580-1589.

[61]Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-v4, Inception-Resnet and the Impact of Residual Connections on Learning[C]//Proceedings of The AAAI Conference on Artificial Intelligence. 2017, 31(1).