基于CIM的偏差补偿稀疏NLMAD算法研究  

Research on CIM based bias-compensated sparse NLMAD algorithm

作　　者：马占军[1] 张佳庚 马文涛[3] 桂冠[4] Ma Zhanjun;Zhang Jiageng;Ma Wentao;Gui Guan(School of Information&Engineering,Shaanxi Institute of International Trade&Commerce,Xi’an 712046,China;Network Center,Xi’an Jiaotong University,Xi’an 710049,China;School of Automation&Information Engineering,Xi’an University of Technology,Xi’an 710048,China;College of Telecommunications&Information Engineering,Nanjing University of Posts&Telecommunications,Nanjing 210003,China)

机构地区：[1]陕西国际商贸学院信息与工程学院,西安712046 [2]西安交通大学网络中心,西安710049 [3]西安理工大学自动化与信息工程学院,西安710048 [4]南京邮电大学通信与信息工程学院,南京210003

出　　处：《计算机应用研究》2018年第9期2736-2740,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61401069);西安理工大学博士启动基金资助项目(103-451116011)

摘　　要：针对输入信号受噪声干扰和输出观测噪声具有脉冲特征的稀疏系统辨识问题,提出一种基于CIM的偏差补偿(normalized least mean absolute deviation,NLMAD)算法,利用NLMAD算法可有效抵御脉冲输出观测噪声。首先应用无偏准则设计偏差补偿NLMAD算法来有效解决由于输入噪声导致的估计偏差问题;考虑到稀疏系统辨识问题,将CIM作为稀疏约束惩罚项引入到偏差补偿NLMAD算法提出了新的稀疏自适应滤波算法——CIMBCNLMAD算法。将所提算法应用于输入和输出均含有噪声的稀疏系统辨识和回声干扰抵消场景中,实验表明CIMBCNLMAD算法的稳态性能优于其他自适应滤波算法,说明该方法具有较强的鲁棒性且可应用于工程实践。For the sparse system identification problem under impulsive output measurement noise and noisy input environment,this paper developed a bias-compensated NLMAD algorithm based on correntropy induced metric(CIM)penalty constraints.Taking advantage of the NLMAD algorithm to resist impulsive output measurement noise interference.First,this paper designed a bias-compensated NLMAD(BCNLMD)algorithm using the unbiasedness criterion to solve the problem of the biased estimation generated by the noisy input.Second,considering the sparse system identification issue,it introduced the CIM into the BCNLMAD algorithm to propose a novel sparse adaptive filtering algorithm CIMBCNLMAD.Finally,it applied the proposed algorithms in sparse system identification and echo cancellation under astable measurement noise and input noise environments.Simulation results show that the proposed CIMBCNLMAD algorithm outperforms other adaptive filtering algorithms which means that it has nice robustness and can be used in engineering practical.

关 键 词：偏差补偿 稀疏系统辨识 互相关熵诱导的维度(CCIM) NLMAD算法 脉冲噪声 

分 类 号：TN911.7[电子电信—通信与信息系统]