首页|基于非均匀量化的极化码SCL译码器FPGA实现

基于非均匀量化的极化码SCL译码器FPGA实现

扫码查看
针对现有均匀量化的连续消除列表(Successive Cancellation List,SCL)译码算法中存储资源消耗大、布线延迟高的问题,提出了一种采用 5 bit非均匀量化方案的SCL译码算法。该算法保留均匀量化中的对数似然比(Log-Like-lihood Ratio,LLR)迭代计算方法,采用5 bit非均匀量化LLR,在LLR计算模块中设计查找表(Look-Up-Table,LUT)转为6 bit均匀量化LLR用于计算。仿真结果表明,提出的 5 bit非均匀量化SCL译码相比于 6 bit均匀量化 SCL译码器,在码率R=1/2、列表宽度L=2 和L=4 时,误帧率(Frame Erasure Rate,FER)性能损失在0。1dB以内。在硬件资源消耗方面,与 6 bit均匀量化译码器相比,5 bit非均匀量化方案译码器在 L=2 时触发器(Flip-Flop,FF)和块随机存取存储器(Block Random Access Memory,BRAM)存储资源消耗分别减少了 10。9%和 22%,吞吐量增加了 24%;L=4 时 FF和BRAM分别减少了 10%和 18。1%,吞吐量增加了 17。5%。
Design of SCL Decoder Based on Non-uniform Quantized Polarization Code of FPGA
In order to solve the problem of large storage resource consumption and high wiring delay in the existing uniformly quan-tized Successive Cancellation List(SCL)decoding algorithms,a SCL decoding algorithm using 5 bit non-uniform quantization scheme is proposed.The algorithm retains the iterative calculation method of Log-Likelihood Ratio(LLR)in uniform quantization,adopts 5 bit non-uniform quantization of LLR,and designs a Look-Up Table(LUT)in the LLR calculation module to convert to 6 bit uniformly quantized LLR for calculation.Simulation results show that compared with the 6 bit uniformly quantized SCL decoder,the performance loss of Frame Erasure Rate(FER)frame error rate is less than 0.1 dB when the bitrate R=1/2 list width L=2 and L=4 are used.In terms of hardware resource consumption,compared with the 6 bit uniform quantization decoder,the Flip-Flop(FF)and Block Random Access Memory(BRAM)storage resource consumption of the 5 bit non-uniform quantization scheme decoder are reduced by 10.9%and 22%,respectively,and the throughput is increased by24%when L=2.At L=4,FF and BRAM decreased by10%and18.1%,respec-tively,and throughput increased by 17.5%.

polar codeSCL decodingnon-uniform quantificationFPGA

魏少圣、熊启金、郑绍华、陈平平

展开 >

福州大学物理与信息工程学院,福建福州 350108

极化码 连续消除列表译码 非均匀量化 现场可编程逻辑门阵列

2024

无线电通信技术
中国电子科技集团公司第五十四研究所

无线电通信技术

北大核心
影响因子:0.745
ISSN:1003-3114
年,卷(期):2024.50(6)