北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 1998, Vol. 21 ›› Issue (2): 46-50.

• 学术论文 • 上一篇    下一篇

一类线性分组码的神经网络译码*

马晓敏1, 杨义先1,章照止2   

  1. 1北京邮电大学信息工程系, 北京 100876; 第一作者33岁, 男, 博士生;2中国科学院系统科学研究所, 北京 100080
  • 收稿日期:1997-06-09 出版日期:1998-03-10
  • 基金资助:
    国家自然科学基金资助项目

Decoding a Kind of Linear Block Codes Using Neural Network

Ma Xiaomin1, Yang Yixian1, Zhang Zhaozhi2   

  1. 1Department of Information Engineering, Beijing University of Posts and Telecommunications, Beijing 100876;2Institute of System Science, Academic Simica, Beijing 100080)
  • Received:1997-06-09 Online:1998-03-10
  • Supported by:
     

摘要: 讨论了线性分组码互补码的特性及分解, 提出了基于神经网络的互补码分解译码方案, 此方案利用神经网络的吸引子和吸引域进行纠错译码, 并把互补码分解为子码及其若干陪集, 更进一步减小神经网络译码的复杂性及规模, 从而实现高效实时硬判决译码.给出了实现原理及步骤, 并对其译码性能进行了分析比较.

关键词: 信道译码, 纠错编码, 分组码, 神经网络

Abstract: After characteristics and decomposition of complementary codes are discussed, A decomposition decoding strategy for complementary codes based on neural network is presented. The attractors of neural network are introduced to implement error correction decoding. The complementary codes are decomposed into a subcode and several cosets of the subcode, which can further decrease complexityof decoding using neural networks, carrying out high efficiency decoding in realtime. The paper demonstrates the principle and the procedure of the decoding structure, and analysis the performance of the decoding.

Key words: channel decoding, error correcting coding, block codes, neural networks

中图分类号: