无字典的超分辨率多径稀疏信道估计方法

发布时间：2018-10-07 19:20

【摘要】：使用有网格的观测字典进行稀疏信道估计是近年来常用的多径稀疏信道估计方法,而网格的存在使得这种方法存在估计精度较差的问题,尤其在网格间距较大时,这种方法的劣势更加明显。本文针对这个问题,抛弃了传统的观测字典,基于连续压缩感知理论,构建出可以施加原子范数最优化的原子集,提出了更加精确的多径稀疏信道估计方法。这种方法避免了网格化带来的误差,实现了高精度、超分辨率的估计。本文首先对此进行了理论阐述,进而在两种不同的多径稀疏信道模型下进行了仿真试验,并从估计精度、计算效率等方面与其他有网格稀疏估计方法以及去网格估计方法进行了对比。仿真结果证明,采用本文提出的方法进行多径稀疏信道估计时,相比其他算法可以更加精确地估计出信道冲激响应。
[Abstract]:Sparse channel estimation using a mesh observation dictionary is a commonly used multipath sparse channel estimation method in recent years. However, the existence of meshes makes the estimation accuracy worse, especially when the mesh spacing is large. The disadvantages of this method are even more obvious. In order to solve this problem, the traditional observation dictionary is abandoned, and based on the theory of continuous compression perception, the atomic set which can be optimized by applying atomic norm is constructed, and a more accurate multipath sparse channel estimation method is proposed. This method avoids the error caused by gridding and achieves high precision and super resolution estimation. In this paper, first of all, the theory is expounded, and then the simulation experiments are carried out under two different multipath sparse channel models, and the estimation accuracy is analyzed. The computational efficiency is compared with other mesh sparse estimation methods and de-mesh estimation methods. Simulation results show that the proposed method can estimate channel impulse response more accurately than other algorithms.
【作者单位】：中国科学院声学研究所;中国科学院水下航行器信息技术重点实验室;中国科学院大学;中国科学院信息工程研究所;
【基金】：国家自然科学基金项目(61471352,61431020,61571434)资助
【分类号】：TN911

【相似文献】