北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 1999, Vol. 22 ›› Issue (1): 29-33.

• 分层与分割概念及其在ATM网管系统中的应用 • 上一篇    下一篇

前向神经网络用于CDMA中的单频干扰抑制

杜光政, 杜丽冰   

  1. 北京邮电大学电信工程学院, 北京 100876
  • 收稿日期:1998-05-20 出版日期:1999-01-10

Feedforward Neural Networks Used to Suppress#br# the Single-Tone Interference in CDMA

Du Guangzheng Du Libing   

  1. School of Telecommunication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876
  • Received:1998-05-20 Online:1999-01-10

摘要: 提出并尝试将基于误差反传的前向神经网络用于CDMA系统的窄带干扰抑制, 构成前向神经网络非线性判决反馈干扰抑制器(FNDFIS); 利用优化计算方法求解了误差反传网络的最佳权值, 进而得出FNDFIS的信噪比改善, 并同传统的非线性判决反馈干扰抑制器(DFIS)作了比较, 其性能优于DFIS.

关键词: 码分多址联接, 神经网络, 单频干扰, 判决反馈, 优化计算

Abstract: Two questiones are discussed.Firstly, a new type ofinterference suppressor was used to achieve the tone interference cancellation in DS-CDMA systems, which is constructed by app-
lying feedforwardneural networksbased on backpropagation algorithm to the nonlinear decision feedback interference suppressor(FNDFIS). Secondly, we also calculate the optimum weights of the networks by using optimization method. The performances were compared between FNDFIS and DFIS, and FNDFIS acts much better than DFIS.

Key words: CDMA, neural networks, tone interference, decisionfeedback, optimization calculation

中图分类号: