Optimization and Hardware Implementation of Learning Assisted Min-Sum Decoders for Polar Codes
- PDF / 1,629,370 Bytes
- 12 Pages / 595.224 x 790.955 pts Page_size
- 17 Downloads / 193 Views
Optimization and Hardware Implementation of Learning Assisted Min-Sum Decoders for Polar Codes Ning Lyu1 · Bin Dai2 · Hongfei Wang2 · Zhiyuan Yan1 Received: 10 May 2019 / Revised: 7 February 2020 / Accepted: 26 May 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Polar codes have received a lot of attention due to their capacity-achieving performance and low complexity. This paper proposes a novel scaling offset min-sum (SOMS) algorithm and adapts the offset min-sum (OMS) algorithm for polar codes, and both algorithms are improved via learning. For all message updates, conventional min-sum decoding algorithms use the same scaling factor or offset, which is usually obtained by numerical simulations. By modeling the data flow of minsum algorithms as a deep neural network, the parameters used in the message passing updates of min-sum decoders can be different for each message update, and are obtained by training and optimizing the corresponding deep neural network. The simulation results show that the proposed SOMS algorithm based on deep learning performs better than all existing BP-based algorithms. Moreover, this paper presents an efficient hardware architecture of the proposed SOMS algorithm. The K-Means clustering algorithm is applied to reduce the number of possible parameters of the neural network, leading to reduced energy consumption and memory requirement with negligible error performance degradation. The proposed architecture of the SOMS algorithm for a (256, 128) polar code is implemented and validated on the Xilinx Artix-7 field-programmable gate array. Keywords Polar codes · Deep learning · Belief Propagation (BP) · Hardware implementation
1 Introduction Polar codes [1] have been regarded as a breakthrough due to their capacity-achieving performance. Recently, polar codes have received great attention and are used in advanced wireless communication protocols such as the fifth generation mobile communication system (5G) [27]. Two families of decoding algorithms for polar codes are successive cancellation (SC) [2] and belief propagation (BP) [4]. The SC algorithm works in serial mode and suffers from long latency and low throughput. Compared with SC, BP decoders for polar codes are attractive due to their low latency and high throughput. However, BP
Ning Lyu
[email protected] 1
Department of Electrical and Computer Engineering, Lehigh University, Bethlehem, PA 18015, USA
2
School of Electronic and Information Engineering, Beihang University, Beijing, China
decoding suffers from a higher computational complexity as well as error performance degradation. Hence, the minsum (MS) approximation [3] for polar codes was proposed to reduce complexity of BP decoding at the expense of error performance degradation. How to improve the error performance of BP decoders remains a challenge. Deep learning has demonstrated significant performance improvements over traditional hand-crafted solutions in many applications, such as image classification [9], speech recognition [6], ga
Data Loading...
