一种改进的基于深度前馈神经网络的极化码BP译码算法
|
摘要
近年来高性能和低复杂度的信道译码算法一直是5G移动通信的核心技术之一,深度学习方法因在译码性能方面表现突出已成为研究热点。基于深度神经网络的极化码译码器使用多尺度置信传播算法可以得到较低复杂度和延迟性能,但其译码性能依旧有待提高。在多尺度置信传播译码算法的基础上提出了一种具有多偏移因子的最小和极化码译码算法,通过使用交叉熵损失函数与提出的交叉熵多损失函数对深度神经网络译码器进行训练,生成的深度神经网络译码器可以降低复杂度和时延,显著提高译码性能。
In recent years, channel decoding algorithm with high performance and low complexity has been one of the core technologies of 5 G mobile communications. Deep learning methods have become a research hotspot because of the outstanding performance in decoding performance. The polar code decoder based on deep neural network can obtain better decoding performance by using multi-scale Belief Propagation(BP) algorithm, and have lower complexity and delay, but the decoding performance still needs to be improved. Based on the multi-scale BP decoding algorithm, a minimum sum decoding algorithm with multiple offset factors is proposed in this paper. The deep neural network decoder is trained by cross-entropy loss function and the proposed cross-entropy multi-loss function, respectively. The generated deep neural network decoder can not only reduce the complexity and delay, but also significantly improve the decoding performance.
In recent years, channel decoding algorithm with high performance and low complexity has been one of the core technologies of 5 G mobile communications. Deep learning methods have become a research hotspot because of the outstanding performance in decoding performance. The polar code decoder based on deep neural network can obtain better decoding performance by using multi-scale Belief Propagation(BP) algorithm, and have lower complexity and delay, but the decoding performance still needs to be improved. Based on the multi-scale BP decoding algorithm, a minimum sum decoding algorithm with multiple offset factors is proposed in this paper. The deep neural network decoder is trained by cross-entropy loss function and the proposed cross-entropy multi-loss function, respectively. The generated deep neural network decoder can not only reduce the complexity and delay, but also significantly improve the decoding performance.
引文
[1]Goodfellow I,Bengio Y,Courville A,et al.Deep learning[M].Cambridge:MIT press,2016.
[2]Polar码成为5G新的控制信道编码[J].上海信息化,2016(12):87-88.
[3]Pamuk A.An FPGA implementation architecture for decoding of polar codes[C]//2011 8th International Symposium on Wireless Communication Systems.Aachen,Germany,2011:437-441.
[4]Nachmani E,Beery Y,Burshtein D.Learning to decode linear codes using deep learning[C]//2016 54th Annual Allerton Conference on Communication,Control and Computing.Monticello,IL,2016:341-346.
[5]Lugosch L,Gross WJNeural off setmin-sum decoding[C]//2017 IEEE International Symposium on Information Theory(ISIT).Aachen,Germany,2017:1361-1365.
[6]Gruber T,Cammerer S,Hoydis J,et al.On deep learning based channel decoding[C]//2017 51st Annual Conference on Information Sciences and Systems.Baltimore,MD,2017:1-6.
[7]Cammerer S,Gruber T,Hoydis J,et al.Scaling deep learning-based decoding of polar codes via partitioning[C]//2017 IEEE Global Communications Conference.Singapore,2017:1-6.
[8]Xu W,Wu Z,Ueng Y L,et al.Improved polar decoder based on deep learning[C]//2017 IEEE International Workshop on Signal Processing Systems.Lorient,France,2017:1-6.
[9]Angarita F,Valls J,Almenar V,et al.Reduced-complexity min-sum algorithm for decoding LDPC codes with low error-floor[J].IEEE Transactions on Circuits and Systems I:Regular Papers,2014,61(7):2150-2158.
[10]Abadi M,Agarwal A,Barham P,et al.Tensorflow:Largescale machine learning on heterogeneous distributed systems[Z].2016.
[11]Kingma D P,Ba J.Adam:A method for stochastic optimization[Z].2014.
[2]Polar码成为5G新的控制信道编码[J].上海信息化,2016(12):87-88.
[3]Pamuk A.An FPGA implementation architecture for decoding of polar codes[C]//2011 8th International Symposium on Wireless Communication Systems.Aachen,Germany,2011:437-441.
[4]Nachmani E,Beery Y,Burshtein D.Learning to decode linear codes using deep learning[C]//2016 54th Annual Allerton Conference on Communication,Control and Computing.Monticello,IL,2016:341-346.
[5]Lugosch L,Gross WJNeural off setmin-sum decoding[C]//2017 IEEE International Symposium on Information Theory(ISIT).Aachen,Germany,2017:1361-1365.
[6]Gruber T,Cammerer S,Hoydis J,et al.On deep learning based channel decoding[C]//2017 51st Annual Conference on Information Sciences and Systems.Baltimore,MD,2017:1-6.
[7]Cammerer S,Gruber T,Hoydis J,et al.Scaling deep learning-based decoding of polar codes via partitioning[C]//2017 IEEE Global Communications Conference.Singapore,2017:1-6.
[8]Xu W,Wu Z,Ueng Y L,et al.Improved polar decoder based on deep learning[C]//2017 IEEE International Workshop on Signal Processing Systems.Lorient,France,2017:1-6.
[9]Angarita F,Valls J,Almenar V,et al.Reduced-complexity min-sum algorithm for decoding LDPC codes with low error-floor[J].IEEE Transactions on Circuits and Systems I:Regular Papers,2014,61(7):2150-2158.
[10]Abadi M,Agarwal A,Barham P,et al.Tensorflow:Largescale machine learning on heterogeneous distributed systems[Z].2016.
[11]Kingma D P,Ba J.Adam:A method for stochastic optimization[Z].2014.