An Efficient Software List Sphere Decoder for Polar Codes

被引:2
作者
Zhou, Huayi [1 ,2 ]
Fu, Yuxiang [3 ]
Zhang, Zaichen [1 ,2 ]
Gross, Warren J. [4 ]
You, Xiaohu [1 ,2 ]
Zhang, Chuan [1 ,2 ]
机构
[1] Southeast Univ, Lab Efficient Architectures Digital Commun & Sign, Natl Mobile Commun Res Lab, Quantum Informat Ctr, Nanjing 210096, Jiangsu, Peoples R China
[2] Purple Mt Labs, Nanjing 210096, Jiangsu, Peoples R China
[3] Nanjing Univ, Sch Elect Sci & Engn, Nanjing, Jiangsu, Peoples R China
[4] McGill Univ, Dept Elect & Comp Engn, Montreal, PQ, Canada
来源
JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY | 2020年 / 92卷 / 05期
关键词
Polar codes; List sphere decoding; Path pruning; Efficient sorting; Software decoder;
D O I
10.1007/s11265-019-01506-0
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Polar codes, first achieving the capacity of symmetric binary-input discrete memoryless channels (B-DMCs), have been standardized for eMBB control channels. Since 5G cellular requires flexible architecture which is realized by the software defined networking paradigm, efficient polar decoder is anticipated. Though successive cancellation list (SCL) decoder achieves satisfactory performance, it requires a large amount of memory. For short control channel codes, sphere decoder (SD) is an alternative, but costs unbearable time complexity at low signal-to-noise ratio. List sphere decoder (LSD) abandons the radius and keeps a list of best paths to gain a fixed complexity. However, LSD needs a large list size L for satisfactory performance. In this paper, an efficient software LSD with path pruning and efficient sorting is proposed. We recall the radius as the bound to delete the paths out of the sphere at very early levels. Since L is dynamic, efficient sorting is proposed to reduce the copy operations. Implemented with C++, the proposed decoder can reduce up to 65.3% latency compared with the original LSD, with the same performance and lower complexity.
引用
收藏
页码:517 / 528
页数:12
相关论文
共 22 条
[1]   On the rate of channel polarization [J].
Arikan, Erdal ;
Telatar, Emre .
2009 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY, VOLS 1- 4, 2009, :1493-+
[2]   Channel Polarization: A Method for Constructing Capacity-Achieving Codes for Symmetric Binary-Input Memoryless Channels [J].
Arikan, Erdal .
IEEE TRANSACTIONS ON INFORMATION THEORY, 2009, 55 (07) :3051-3073
[3]  
FINCKE U, 1985, MATH COMPUT, V44, P463, DOI 10.1090/S0025-5718-1985-0777278-8
[4]  
Guo J, 2015, IEEE INT SYMP INFO, P236, DOI 10.1109/ISIT.2015.7282452
[5]   Fast and Flexible Successive-Cancellation List Decoders for Polar Codes [J].
Hashemi, Seyyed Ali ;
Condo, Carlo ;
Gross, Warren J. .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2017, 65 (21) :5756-5769
[6]   A Fast Polar Code List Decoder Architecture Based on Sphere Decoding [J].
Hashemi, Seyyed Ali ;
Condo, Carlo ;
Gross, Warren J. .
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS, 2016, 63 (12) :2368-2380
[7]  
Hashemi SA, 2016, IEEE INT SYMP CIRC S, P1730, DOI 10.1109/ISCAS.2016.7538902
[8]  
Hashemi SA, 2015, 2015 49TH ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS AND COMPUTERS, P1346, DOI 10.1109/ACSSC.2015.7421362
[9]   On the sphere-decoding algorithm I. Expected complexity [J].
Hassibi, B ;
Vikalo, H .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2005, 53 (08) :2806-2818
[10]   Reduced Latency ML Polar Decoding via Multiple Sphere-Decoding Tree Searches [J].
Husmann, Chistopher ;
Nikolaou, Panagiotis Chatzi ;
Nikitopoulos, Konstantinos .
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2018, 67 (02) :1835-1839