冲击噪声下基于Lorentzian范数的CSR参数估计  

Parameter estimation for CSR under impulsive noise based on Lorentzian norm

作　　者：代林[1] 崔琛[1] 余剑[1] 梁浩[1] 

机构地区：[1]电子工程学院通信对抗系,安徽合肥230037

出　　处：《华中科技大学学报（自然科学版）》2015年第7期66-71,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(60702015)

摘　　要：针对现有重构算法及其改进算法在压缩感知雷达(CSR)参数估计中存在的稳健性不强、适用性不广等问题,提出了一种适用于冲击噪声背景的鲁棒性算法——Lorentzian-ISL0(基于Lorentzian范数的改进光滑l0范数).建立CSR参数估计的稀疏线性模型,并基于Lorentzian范数和高斯函数稀疏正则化,构造冲击噪声下稳健的优化目标函数;修正优化目标函数的牛顿方向,并沿修正方向对估计值进行更新,直至收敛.仿真实验结果表明:与已有算法相比,本文方法计算复杂度更小,支撑集重构更精确,信号重构精度更高.A novel parameter estimation algorithms used in impulsive noise environment—Lorentzian based improved smoothed l0norm（Lorentzian-ISL0）was proposed to improve the robustness and universality of compressed sensing radar（CSR）with most existing sparse recovery algorithms and their derivants.Firstly,the sparse linear model（SLM）was built for parameter estimation,and the robust target function was derived based on the Lorentzian norm and the Gaussian function sparse regularization.Then the Newton direction of the robust target function was modified to iteratively update the parameter until the target function was converged.The proposed method outperforms other existing method in terms of computational complexity,support and signal recovery.The effectiveness is demonstrated by computer simulations.

关 键 词：压缩感知雷达 冲击噪声 对称α稳定分布 Lorentzian范数 改进的光滑l0范数 

分 类 号：TN958.8[电子电信—信号与信息处理]