脉冲信道下基于深度学习的BP译码方法

doi:10.3969/j.issn.1001-506X.2020.09.30

摘要/Abstract

摘要：

在脉冲信道下,针对置信传播(belief propagation, BP)算法对短码长的低密度奇偶校验(low-density parity-check, LDPC)码译码性能差的问题,提出了一种基于深度学习的BP译码方法。首先，根据Tanner图构建两种深度神经网络模型,通过对Tanner图中边的权重重新赋值来提升译码性能。然后，简化信道对数似然比(log-likelihood ratio, LLR)的计算方法,通过模型训练优化近似计算参数,得到对参数γ鲁棒的译码模型。最后，构造鲁棒训练集,训练得到对参数α和γ鲁棒的译码模型。仿真结果表明,在高码率时,该方法相对于传统BP译码算法性能显著提升,且在近似计算信道LLR值时,译码性能在不同参数的脉冲信道下均具有鲁棒性。

关键词: 深度学习, 脉冲信道, 信道译码, 低密度奇偶校验码

Abstract:

Aiming at the poor decoding performance of the belief propagation (BP) algorithm for low-density parity-check (LDPC) codes with short block lengths, a BP decoding method based on deep learning is presented in impulsive channels. Firstly, two deep neural network models are constructed by the Tanner graph, and the weights of edges in the Tanner graph are reassigned to improve the decoding performance. Then, the calculation method of the log-likelihood ratio (LLR) of the channel is simplified, the model is trained to optimize approximate calculation parameters, and the decoding model robust to parameter γ is obtained. Finally, a robust training set is constructed to obtain the decoding model robust to parameter α and γ. Simulation results show that, at a high code rate, the performance of the proposed method is significantly improved compared with traditional BP algorithms. Furthermore, when approximately calculating the LLR value of the channel, the decoding performance is robust in impulsive channels with different parameters.

Key words: deep learning, impulsive channel, channel decoding, low-density parity-check (LDPC) code

中图分类号:

TN911.7

潘睿, 袁磊. 脉冲信道下基于深度学习的BP译码方法[J]. 系统工程与电子技术, 2020, 42(9): 2116-2122.

Rui PAN, Lei YUAN. BP decoding method based on deep learning in impulsive channels[J]. Systems Engineering and Electronics, 2020, 42(9): 2116-2122.

图/表 13

图1

图2

图3

图4

表1

表2

图5

图6

图7

图8

表3

图9

图10

参考文献 19

1	肖扬, 徐丹. 准循环LDPC好码设计[J]. 系统工程与电子技术, 2009, 31 (5): 1011- 1017. doi: 10.3321/j.issn:1001-506X.2009.05.003
	XIAO Y , XU D . Design of good quasi-cyclic LDPC codes[J]. Systems Engineering and Electronics, 2009, 31 (5): 1011- 1017. doi: 10.3321/j.issn:1001-506X.2009.05.003
2	SHIRVANIMOGHADDAM M , MOHAMMADI M S , ABBAS R , et al. Short block-length codes for ultra-reliable low latency communications[J]. IEEE Communications Magazine, 2019, 57 (2): 130- 137. doi: 10.1109/MCOM.2018.1800181
3	MAO Q , HU F , HAO Q . Deep learning for intelligent wireless networks: a comprehensive survey[J]. IEEE Communications Surveys & Tutorials, 2018, 20 (4): 2595- 2621.
4	NACHMANI E, BEERY Y, BURSHTEIN D. Learning to decode linear codes using deep learning[C]//Proc.of the 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton), 2016: 341-346.
5	NACHMANI E , MARCIANO E , LUGOSCH L , et al. Deep learning methods for improved decoding of linear codes[J]. IEEE Journal of Selected Topics in Signal Processing, 2017, 12 (1): 119- 131.
6	GRUBER T, CAMMERER S, HOYDIS J, et al. On deep learning-based channel decoding[C]//Proc.of the 51st Annual Conference on Information Sciences and Systems, 2017.
7	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. doi: 10.1109/JSTSP.2018.2794062
8	KUAI X Y , SUN H S , ZHOU S L , et al. Impulsive noise mitigation in underwater acoustic OFDM systems[J]. IEEE Trans.on Vehicular Technology, 2016, 65 (10): 8190- 8202. doi: 10.1109/TVT.2016.2516539
9	ANKIT D , RANJAN K M , ROBERT S . Performance analysis of a multi-hop power line communication system over log-normal fading in presence of impulsive noise[J]. IET Communications, 2015, 9 (1): 1- 9. doi: 10.1049/iet-com.2014.0464
10	SACUTO F , LABEAU F , AGBA B L . Wide band time-correlated model for wireless communications under impulsive noise within power substation[J]. IEEE Trans.on Wireless Communications, 2014, 13 (3): 1449- 1461. doi: 10.1109/TWC.2014.011714.130685
11	李旭涛, 朱光喜, 王首勇, 等. Alpha稳定分布的参数表征及仿真[J]. 信号处理, 2007, 23 (6): 814- 817. doi: 10.3969/j.issn.1003-0530.2007.06.004
	LI X T , ZHU G X , WANG S Y , et al. Parameterizations and simulation of alpha stable distribution[J]. Signal Processing, 2007, 23 (6): 814- 817. doi: 10.3969/j.issn.1003-0530.2007.06.004
12	JOHN P N . Numerical calculation of stable densities and distribution functions[J]. Communications in Statistics. Stochastic Models, 1997, 13 (4): 759- 774. doi: 10.1080/15326349708807450
13	GONZALEZ J G , PAREDES J L , ARCE G R . Zero-order statistics: a mathematical framework for the processing and cha-racterization of very impulsive signals[J]. IEEE Trans.on Signal Processing, 2006, 54 (10): 3839- 3851. doi: 10.1109/TSP.2006.880306
14	HOU Y , LIU R K , DAI B , et al. Joint channel estimation and LDPC decoding over time-varying impulsive noise channels[J]. IEEE Trans.on Communications, 2018, 66 (6): 2376- 2383. doi: 10.1109/TCOMM.2018.2800748
15	BEN M H , GOUPIL A , CLAVIER L , et al. Clipping demapper for LDPC decoding in impulsive channel[J]. IEEE Communications Letters, 2013, 17 (5): 968- 971. doi: 10.1109/LCOMM.2013.031913.130060
16	DIMANCHE V , GOUPIL A , CLAVIER L , et al. On detection method for soft iterative decoding in the presence of impulsive interference[J]. IEEE Communications Letters, 2014, 18 (6): 945- 948. doi: 10.1109/LCOMM.2014.2320921
17	DAI B , LIU R K , MEI Z . Novel non-linear demapper for soft decision decoder of LDPC codes in impulsive noise[J]. IET Communications, 2019, 13 (6): 796- 801. doi: 10.1049/iet-com.2018.5362
18	RICHARDSON T J, RüDIGER L, URBANKE. Modern coding theory[M]. Cambridge University Press, 2008: 210-215.
19	HU X Y , ELEFTHERIOU E , ARNOLD D M . Regular and irregular progressive edge-growth tanner graphs[J]. IEEE Trans.on Information Theory, 2005, 51 (1): 386- 398. doi: 10.1109/TIT.2004.839541

实验编号	深度神经网络模型的网络层	权重数量	优化算法	学习率	mini-batch
1	输入层、准确LLR层、5个变量层、5个校验层、5个判决层	5(3h+n)	RMSProp	0.01	50
1	输入层、准确LLR层、循环层、判决层	3h+n
2	输入层、近似LLR层、循环层、判决层	3h+n+2
3	输入层、近似LLR层、循环层、判决层	3h+n+2

码字类型	R	α值	(E_b/N₀)/dB
全零码字	1/2	1	{9, 10, 11, 12, 13, 14}
	1/2	1.5	{6, 7, 8, 9, 10, 11}
	3/4、4/5	1	{20, 22, 24, 26, 28, 30}
	3/4、4/5	1.5	{14, 16, 18, 20, 22, 24}

R	码字类型	α值	E_b/N₀/dB	数量
3/4	全零码字	0.8，1，1.5	{20, 22, 24}	270 000
1/2	全零码字	0.8，1，1.5	{9, 10, 11}	270 000

[1]	韩啸, 陈世文, 陈蒙, 杨锦程. 基于互易点学习的LPI信号开集识别[J]. 系统工程与电子技术, 2022, 44(9): 2752-2759.
[2]	张立民, 谭凯文, 闫文君, 张聿远. 基于多级跳线残差网络的雷达辐射源识别[J]. 系统工程与电子技术, 2022, 44(7): 2148-2156.
[3]	陈发堂, 李贺宾, 李平安. 基于偏移最小和的LDPC译码改进算法[J]. 系统工程与电子技术, 2022, 44(7): 2350-2356.
[4]	金国栋, 薛远亮, 谭力宁, 许剑锟. 基于孪生神经网络的目标跟踪算法进展研究[J]. 系统工程与电子技术, 2022, 44(6): 1805-1822.
[5]	赵晓枫, 徐叶斌, 吴飞, 牛家辉, 蔡伟, 张志利. 基于全局感知机制的地面红外目标检测方法[J]. 系统工程与电子技术, 2022, 44(5): 1461-1467.
[6]	邹虹, 白陈阳, 何鹏, 崔亚平, 王汝言, 吴大鹏. 基于分布式深度学习的边缘服务放置策略[J]. 系统工程与电子技术, 2022, 44(5): 1728-1737.
[7]	陈冬, 句彦伟. 基于语义分割实现的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(4): 1195-1201.
[8]	孙晶明, 虞盛康, 孙俊. 基于深度学习的HRRP识别姿态敏感性分析[J]. 系统工程与电子技术, 2022, 44(3): 802-807.
[9]	刘恒燕, 张立民, 闫文君, 钟兆根, 凌青, 梁晓军. 基于WBP-CNN算法的LDPC译码[J]. 系统工程与电子技术, 2022, 44(3): 1030-1035.
[10]	姚云翔, 陈莹. 注意力机制下双模态交互融合的目标跟踪网络[J]. 系统工程与电子技术, 2022, 44(2): 410-419.
[11]	高涌荇, 王旭东, 汪玲, 朱岱寅, 郭军, 孟凡旺. 基于RCNN的双极化气象雷达天气信号检测[J]. 系统工程与电子技术, 2022, 44(11): 3380-3387.
[12]	王任之, 潘克刚, 赵瑞祥. 跳频抗干扰通信系统中LDPC码的编码优化设计[J]. 系统工程与电子技术, 2022, 44(11): 3548-3555.
[13]	周毅恒, 杨军, 夏赛强, 吕明久. 闪烁现象下旋翼目标微动参数估计方法[J]. 系统工程与电子技术, 2022, 44(1): 54-63.
[14]	付蓉, 黄天耀, 刘一民. 基于深度学习的捷变相参雷达1-Bit块稀疏重构[J]. 系统工程与电子技术, 2022, 44(1): 70-75.
[15]	张聿远, 张立民, 闫文君. 基于互相关特征图和扩张稠密卷积网络的SFBC-OFDM识别方法[J]. 系统工程与电子技术, 2021, 43(9): 2657-2664.