摘要
在极化敏感圆阵通过降维长矢量多重信号分类(DRLV-MUSIC)算法实现DOA估计的现场可编程门阵列(FPGA)实现时,为了进一步缩短计算时间的同时节省硬件占用资源,提出一种特征值与特征向量并行计算的FPGA实现结构,同时利用矩阵相乘的重复性,提出一种并行矩阵运算方法,缩短FPGA的计算时间同时节约硬件资源。该FPGA方案由预处理模块、协方差矩阵计算模块、并行Jacobi算法计算特征值模块、并行特征向量计算模块组成。实验结果表明,与非并行的DRLV-MUSIC算法相比,该方案在减少计算时间、降低硬件占用资源的同时保证了测角精度。
In order to solve the problem that directly embedded FPGA will consume lots of hardware resources and overlong computing time to achieve angle measurement through polarization dimensionality reduction multiple signal classification(dimensionality reduction long vector multiple signal classification,DRLV-MUSIC)algorithm,an FPGA implementation structure of parallel calculation of eigenvalues and eigenvectors was proposed.Otherwise parallel matrix operation method was proposed to shorten the calculation time of FPGA by using the repeatability of mutability.The FPGA scheme was comprised of a preprocessing module,a covariance matrix calculation module,a parallel Jacobi algorithm computing eigenvalue module and a parallel eigenvector computing module.Experimental results showed that the scheme reduced the calculation time and hardware resources while ensuring the angular measurement accuracy compared with the non-parallel DRLV-MUSIC algorithm.
作者
李晓璇
孙闽红
LI Xiaoxuan;SUN Minhong(School of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310018,China)
出处
《探测与控制学报》
CSCD
北大核心
2024年第5期50-56,共7页
Journal of Detection & Control
基金
国家自然科学基金项目(61901149)
国防特色学科发展项目(JCKY2019415D002)。
