摘要
针对稀疏重构中正交匹配追踪(OMP)算法解相干问题,利用接收数据构造目标矩阵奇异值分解(SVD)后的大特征值对应的特征矢量,提出了两种改进解相干算法(NSO算法和MNSO算法).首先根据稀疏重构的框架下的阵列DOA估计模型,理论上分析了经典OMP算法、NSO算法和MNSO算法的运算量和重构精度,然后给出了算法性能的仿真结果.仿真结果表明,相对于经典OMP算法,两种改进算法的运算速度更快,稀疏重构效果更优.理论分析和仿真结果验证了两种改进算法的良好性能.
In order to solve the coherent problem of orthogonal matching pursuit (OMP) algorithm in sparse reconstruction, this paper puts forward two ameliorated de-coherence algorithms (namely the NSO algorithm and the MNSO algorithm) by using the eigenvectors corresponding to large eigenvalues after singular value decomposition (SVD) of the object matrix of the received data construction. Firstly, based on the array DOA estimation model under the sparse reconstruction framework, the paper theoretically analyzes the amount of calculation and reconstruction precision of the classical OMP algorithm, the NSO algorithm and the MNSO algorithm, and then provides the simulation results of the algorithm performance. The simulation results show that the two ameliorated algorithms are faster and more effective in sparse reconstruction than the classical OMP algorithm. Theoretical analysis and simulation results verify the good performance of the two ameliorated algorithms.
作者
季正燕
陈辉
张佳佳
陆晓飞
JI Zhengyan;CHEN Hui;ZHANG Jiajia;LU Xiaofei(Air Force EarlyWarning Academy,Wuhan 430019, China)
出处
《空军预警学院学报》
2017年第1期5-10,共6页
Journal of Air Force Early Warning Academy
关键词
稀疏重构
解相干
正交匹配追踪算法
奇异值分解算法
sparse reconstruction
de-coherence
orthogonal matching pursuit (OMP) algorithm
singular value decomposition (SVD) algorithm
