| 最低误码率非线性均衡器的快速自适应学习算法 |
| |
| 引用本文： | 朱仁祥,吴乐南.最低误码率非线性均衡器的快速自适应学习算法[J].电路与系统学报,2012,17(2):88-94. |
| |
| 作者姓名： | 朱仁祥 吴乐南 |
| |
| 作者单位： | 1. 宁波工程学院电子与信息工程学院,浙江宁波,315016
2. 东南大学信息科学与工程学院,江苏南京,210096 |
| |
| 基金项目： | 国家自然科学基金资助课题(60872075);宁波市自然科学基金资助课题(2010A610176,2011A610206);浙江省自然科学基金资助课题(Y1100377,T1110086) |
| |
| 摘 要： | 针对最低误码率非线性均衡器的参数在线自适应学习问题,本文提出基于拟牛顿方法的快速自适应学习算法。采用Parzen窗函数方法估计误码率,通过设定切换条件,使参数学习在滑窗随机梯度法与滑窗拟牛顿法之间切换。这既增加了新算法的数值稳定性,又可提高收敛速度。通过对拟牛顿方法进行修改,还使新算法既可以在线自适应学习,也可用于高维参数的快速学习。仿真采用最低误码率非线性均衡器对通信系统进行干扰抑制和信道均衡,结果表明了新算法的高效性。
|
| 关 键 词： | 非线性均衡器 最低误码率 学习算法 通信信号处理 |
Adaptive learning algorithms with fast convergence rate for minimum bit error rate nonlinear equalizers |
| |
| ZHU Ren-xiang
,
WU Le-nan.Adaptive learning algorithms with fast convergence rate for minimum bit error rate nonlinear equalizers[J].Journal of Circuits and Systems,2012,17(2):88-94. |
| |
| Authors: | ZHU Ren-xiang WU Le-nan |
| |
| Affiliation: | 1.School of Electronics and Information Engineering,Ningbo University of Technology,Ningbo 315016,China; 2.School of Information Science and Engineering,Southeast University,Nanjing 210096,China) |
| |
| Abstract: | Fast converged and adaptive algorithms based on quasi Newton method are proposed in this paper for online training of minimum bit error rate nonlinear equalizers.Bit error rate is estimated by Parzen window.Learning can be switched between sliding window-stochastic gradient algorithm and sliding window-quasi Newton algorithm by a decisive variable.This makes the new algorithms much more stable with a fast convergence rate.Moreover,by modifying the quasi Newton method,the new algorithms can be used both for adaptive learning and for high-dimensional parameters training.Minimum bit error rate nonlinear equalizers are used for interference suppression and channel equalization in communications,and their high efficiency is proved by simulation results. |
| |
| Keywords: | nonlinear equalizers minimum bit error rate learning algorithm communication signal processing |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
