电子学报 ›› 2019, Vol. 47 ›› Issue (2): 428-433.DOI: 10.3969/j.issn.0372-2112.2019.02.024

• 学术论文 • 上一篇    下一篇

基于IBED和仿生算法的LT码度分布设计

姚渭箐1,2, 胡凡2   

  1. 1. 国网湖北省电力有限公司信息通信公司, 湖北武汉 430077;
    2. 武汉大学电子信息学院, 湖北武汉 430072
  • 收稿日期:2018-05-13 修回日期:2018-12-03 出版日期:2019-02-25 发布日期:2019-02-25
  • 通讯作者: 姚渭箐
  • 作者简介:胡凡 男,1988年生,湖北随州人,博士,研究方向为高分辨率遥感图像解译、模式识别及、深度学习.
  • 基金资助:
    中国博士后科学基金(No.2017M622519)

The Design of Degree Distribution for LT Codes Based on IBED and Bionic Algorithm

YAO Wei-qing1,2, HU Fan2   

  1. 1. State Grid Hubei Information & Telecommunication Company Limited, Wuhan, Hubei 430077, China;
    2. School of Electronic Information, Wuhan University, Wuhan, Hubei 430072, China
  • Received:2018-05-13 Revised:2018-12-03 Online:2019-02-25 Published:2019-02-25

摘要: 度分布是影响Luby变换(Luby transform,LT)码性能优劣的关键因素,为了得到更优的度分布,提出一种基于改进的二进制指数分布(improved binary exponential distribution,IBED)和仿生算法的LT码度分布设计方案.采用IBED度分布进行LT编码,初始译码成功率可高达90%以上,但随着译码开销增大,译码成功率上升速度减缓,而此时鲁棒孤子分布(robust soliton distribution,RSD)的译码性能表现更佳.为了将IBED和RSD的优点有机结合,利用人工鱼群算法(artificial fish swarm algorithm,AFSA)在两种度分布间寻优,基于期望可译集构建该算法的目标函数,通过最大化目标值得到更优的度分布.仿真结果表明,与类似方法及传统的RSD相比,采用新度分布进行LT编码可明显降低译码开销,并节约编译码耗时.

关键词: LT码, 改进的二进制指数分布, 鲁棒孤子分布, 人工鱼群算法, 可译集

Abstract: The degree distribution greatly influences the performance of Luby transform(LT)codes.In order to obtain a better degree distribution,a method is proposed to design the degree distribution for LT codes based on improved binary exponential distribution(IBED)and bionic algorithm.The decoding performance of IBED outperforms robust soliton distribution(RSD)when the overhead is small,and its successful decoding rate reaches more than 90%.The successful decoding rate for IBED rises slowly as the overhead increases,while RSD performs better.In order to combine the advantages of IBED and RSD,we use artificial fish swarm algorithm(AFSA)to search the optimal degree distribution between these two.The expected ripple size is employed to present the objective function for AFSA,and the optimal degree distribution is obtained by maximizing the objective value.Simulation results show that compared to other optimized degree distributions and RSD,the new degree distribution reduces the overhead,and saves the average consuming time per encoding/decoding process.

Key words: luby transform (LT) codes, improved binary exponential distribution (IBED), robust soliton distribution (RSD), artificial fish swarm algorithm (AFSA), ripple size

中图分类号: