Scholarlink

<p>Large kernel polar codes can provide outstanding error-correction performance for finite-length coding, which is expected to support ultra-high reliability and ultra-low latency communication (URLLC) for 6G. However, the complexity of decoding for large kernel polar codes grows exponentially with the kernel size. In this paper, we study Tanner-graph-assisted (TGA) decoding to further reduce complexity while maintaining satisfactory error-correction performance. We first construct a low-complexity parallel TGA belief propagation (TGA-BP) decoder. The decoder takes the kernel matrix that achieves the optimal exponent as the Tanner graph and selects key nodes for iterative decoding. In particular, the Tanner graph and iterative decoding equations for arbitrary dimensional linear binary kernels are derived. Then, a two-step Tanner graph optimization strategy is further proposed to enhance the TGA-BP. It includes a kernel matrix generation method that obtains different kernels with the optimal exponent and a kernel matrix selection method that maximizes the mean log-likelihood ratios (mLLR). The simulation results demonstrate that our scheme achieves significant complexity reduction and error-correction improvement compared to large kernel polar codes under successive cancellation (SC) decoding over the additive white Gaussian noise (AWGN) channel.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Tanner-graph-assisted belief propagation decoding for large kernel polar codes: low-complexity design and enhancement method

  • 期刊论文

  • 发表日期：

    2025年9月16日

查看全文
  • Yangyang Liu,
  • Yu Zhang,
  • Guanghua Liu,
  • Jiaxi Zhou,
  • Lixia Xiao,
  • Tao Jiang

摘要

Large kernel polar codes can provide outstanding error-correction performance for finite-length coding, which is expected to support ultra-high reliability and ultra-low latency communication (URLLC) for 6G. However, the complexity of decoding for large kernel polar codes grows exponentially with the kernel size. In this paper, we study Tanner-graph-assisted (TGA) decoding to further reduce complexity while maintaining satisfactory error-correction performance. We first construct a low-complexity parallel TGA belief propagation (TGA-BP) decoder. The decoder takes the kernel matrix that achieves the optimal exponent as the Tanner graph and selects key nodes for iterative decoding. In particular, the Tanner graph and iterative decoding equations for arbitrary dimensional linear binary kernels are derived. Then, a two-step Tanner graph optimization strategy is further proposed to enhance the TGA-BP. It includes a kernel matrix generation method that obtains different kernels with the optimal exponent and a kernel matrix selection method that maximizes the mean log-likelihood ratios (mLLR). The simulation results demonstrate that our scheme achieves significant complexity reduction and error-correction improvement compared to large kernel polar codes under successive cancellation (SC) decoding over the additive white Gaussian noise (AWGN) channel.