SUMMARY:
UCLA researchers in the Department of Electrical and Computer Engineering have developed a neural network-based approach to decode Low Density Parity Codes (LDPCs) that is capable of accurately transmitting large bit number data faster and with a lower frame rate error that conventional approaches.
BACKGROUND:
The transmission of data at high speeds is limited due to the time it takes to check and correct for transmission errors. As the application of wireless telecommunication services continues to increase, so does the need to correct errors in the transmitted messages. When recovering stored data from a device or sending data to a new device, there is an inherent risk for errors to be introduced. Errors can lead to the misinterpretation of data by the receiving device which can result in information loss and can lead to a number of software failures. This directly affects a number of technologies including wireless communication with satellites as well as wired technologies like data storage. One method often used to correct these errors involves the use of parity bits which are additional bits attached to the end of the data that assist in correcting errors. Extensive work has been done to design efficient Low Density Parity Codes (LDPC) which allow for a minimal number of parity bits to protect a large amount of data. Even more work has been done on computational approaches to use these LDPCs to recover the original data such as MinSums. While these approaches can effectively recover errors, they are often slow. Additionally, the maximum length of data that can be recovered with approaches like this limit data transmission speeds. There is a clear and present need to develop more efficient algorithms for decoding LDPCs which will allow the accurate transmission of larger data packets at faster speeds.
INNOVATION:
The present invention is a Neural Network based MinSum decoding approach called Neural Normalized MinSum (N-NMS). This novel method offers better frame error rate performance especially on linear block codes compared to conventional normalized MinSum. N-NMS also allows for decoding of data with longer block lengths and has been tested using 3096 and 1032 bit LDPCs. The approach consists of a reduced number of parameters that allow for efficient convergence and robust implementation. N-NMS offers a solution to efficiently reduce the number of transmission errors while ensuring high-speed transmission rates are maintained. This innovation poses increased value in a wide array of fields including telecommunication and aerospace applications.
POTENTIAL APPLICATIONS:
- Ethernet
- Satellite communications
- Television
- 5G networks
- Real-time video transmission
ADVANTAGES:
- Better Frame Error Rate
- Larger Data Block Size
- Minimal Parameters
- Fast Convergence
DEVELOPMENT-TO-DATE:
N-NMS has been successfully demonstrated including extensive simulation.
RELATED PAPERS:
[1] E. Nachmani, Y. Be’ery, and D. Burshtein, “Learning to decode linear codes using deep learning,” in 2016 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton), Sep. 2016, pp.341–346.
[2] L. Lugosch and W. J. Gross, “Neural offset min-sum decoding,” in 2017 IEEE International Symposium on Information Theory (ISIT), Jun. 2017,pp. 1361–1365.
[3] E. Nachmani, E. Marciano, D. Burshtein, and Y. Be’ery, “RNN decoding of linear block codes,” CoRR, vol. abs/1702.07560, 2017.[Online]. Available: http://arxiv.org/abs/1702.07560
[4] E. Nachmani, E. Marciano, L. Lugosch, W. J. Gross, D. Burshtein, andY. Be’ery, “Deep learning methods for improved decoding of linear codes,” IEEE J. Sel. Top. Signal Process., vol. 12, no. 1, pp. 119–131,Feb. 2018.
[5] F. Liang, C. Shen, and F. Wu, “An iterative BP-CNN architecture for channel decoding,” IEEE J. Sel. Top. Signal Process., vol. 12, no. 1, pp.144–159, Feb. 2018.
[6] X. Wu, M. Jiang, and C. Zhao, “Decoding optimization for 5G LDPC codes by machine learning,” IEEE Access, vol. 6, pp. 50 179–50 186,2018.
[7] L. Lugosch and W. J. Gross, “Learning from the syndrome,” in 201852nd Asilomar Conference on Signals, Systems, and Computers, Oct.2018, pp. 594–598.
[8] W. Lyu, Z. Zhang, C. Jiao, K. Qin, and H. Zhang, “Performance evaluation of channel decoding with deep neural networks,” in 2018 IEEE International Conference on Communications (ICC), May 2018,pp. 1–6.
[9] X. Xiao, B. Vasic, R. Tandon, and S. Lin, “Finite alphabet iterative decoding of LDPC codes with coarsely quantized neural networks,” in 2019 IEEE Global Communications Conference (GLOBECOM), Dec.2019, pp. 1–6.
[10] C. Deng and S. L. Bo Yuan, “Reduced-complexity deep neural network aided channel code decoder: A case study for BCH decoder,” in ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), May 2019, pp. 1468–1472.
[11] A. Abotabl, J. H. Bae, and K. Song, “Offset min-sum optimization for general decoding scheduling: A deep learning approach,” in 2019 IEEE 90th Vehicular Technology Conference (VTC2019-Fall), Sep. 2019, pp.1–5.
[12] A. Buchberger, C. Häger, H. D. Pfister, L. Schmalen, and A. G. i Amat,“Pruning and quantizing neural belief propagation decoders,” IEEE Journal on Selected Areas in Communications, 2020.
[13] Q. Wang, S. Wang, H. Fang, L. Chen, L. Chen, and Y. Guo, “A Model-Driven deep learning method for normalized Min-Sum LDPC decoding,”in 2020 IEEE International Conference on Communications Workshops(ICC Workshops), Jun. 2020, pp. 1–6.
[14] Juntan Zhang, M. Fossorier, Daqing Gu, and Jinyun Zhang, “Improved min-sum decoding of ldpc codes using 2-dimensional normalization,” in GLOBECOM ’05. IEEE Global Telecommunications Conference, 2005.,vol. 3, 2005, pp. 1187– 1192.
[15] P. J. Werbos, “Generalization of backpropagation with application to are current gas market model,” Neural Netw., vol. 1, no. 4, pp. 339–356,Jan. 1988.
[16] “UCLA communications systems laboratory,” http://www.seas.ucla.edu/csl/#/publications/published-codes-and-design-tools, accessed: 2020-12-15.
[17] T. Chen, K. Vakilinia, D. Divsalar, and R. D. Wesel, “Protograph-based raptor-like ldpc codes,” IEEE Transactions on Communications, vol. 63,no. 5, pp. 1522–1532, 2015.
[18] A. Anastasopoulos, “A comparison between the sum product and the min-sum iterative detection algorithms based on density evolution,” in GLOBECOM ’01. IEEE Global Telecommunications Conference (Cat.No.01CH37270), vol. 2, 2001, pp. 1021–1025 vol.2.
[19] J. Rissanen, “Order estimation by accumulated prediction errors,” J.Appl. Probab., vol. 23, pp. 55–61, 1986.
[20] K. J. Preacher, G. Zhang, C. Kim, and G. Mels, “Choosing the optimal number of factors in exploratory factor analysis: A model selection perspective,” Multivariate Behav. Res., vol. 48, no. 1, pp. 28–56, Jan.2013.