Scholarlink

This paper proposed a low-density parity check (LDPC) bit flipping algorithm based on deep double dueling Q -network (D3QN). The algorithm employs deep reinforcement learning for decoding, transforming the decoding process into an interaction between agent and the environment. By defining appropriate state and action spaces, different actions are selected to flip bits of codewords, resulting in different codeword states. After multiple rounds of learning, the agent produces the decoding results. Simulation results demonstrate that the LDPC bit flipping algorithm based on D3QN outperforms traditional bit decoding algorithms. In recent years, deep reinforcement learning (DRL) has shown significant potential in solving complex decision-making problems. Applying DRL to LDPC decoding presents a novel approach, transforming the decoding process into an interactive learning problem.

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

LDPC Decoding Algorithm Based on D3QN Deep Reinforcement Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Weiqi Wang,
  • Hongzheng Zeng,
  • Qizhen Sun,
  • Wei Wei,
  • Wen Zeng,
  • Xue Feng

摘要

This paper proposed a low-density parity check (LDPC) bit flipping algorithm based on deep double dueling Q -network (D3QN). The algorithm employs deep reinforcement learning for decoding, transforming the decoding process into an interaction between agent and the environment. By defining appropriate state and action spaces, different actions are selected to flip bits of codewords, resulting in different codeword states. After multiple rounds of learning, the agent produces the decoding results. Simulation results demonstrate that the LDPC bit flipping algorithm based on D3QN outperforms traditional bit decoding algorithms. In recent years, deep reinforcement learning (DRL) has shown significant potential in solving complex decision-making problems. Applying DRL to LDPC decoding presents a novel approach, transforming the decoding process into an interactive learning problem.