An Improved MUSIC Algorithm Implemented with High-speed Parallel Optimization for FPGA
An Improved MUSIC Algorithm Implemented with High-speed Parallel Optimization for FPGA
复制标题
DOI：
10.1109/isape.2006.353475
复制
发表时间：
2006-10
期刊：
影响因子：
--
通讯作者：
Zhou Zou;Hongyuan Wang;Guowen Yu
中科院分区：
文献类型：
--
作者：
Zhou Zou;Hongyuan Wang;Guowen Yu
本文提出了一种改进的音乐算法，并针对FPGA进行了高速并行优化。尽管音乐算法是一种高性能，经典的DOA方法，但它需要协方差矩阵的估计和特征结构分解，该矩阵的计算成本很耗时，不适合FPGA实施。在本文中，作者提出了一种优化算法，而没有协方差矩阵的特征结构分解。与音乐相比，该算法的计算成本要低得多，而牺牲性能降低了。通过关注相关矩阵估计和频谱峰搜索的并行预处理，引入了FPGA实现，并通过理论分析，仿真和硬件实现被证明是有效的。
This paper proposes an improved MUSIC algorithm with high-speed parallel optimization for FPGA. Although MUSIC algorithm is a high-performance, classic DOA method, it needs estimation and eigenstructure decomposition of covariance matrix, which is time-consuming with high computation cost and not suitable for FPGA implementations. In this paper, the authors present an optimization algorithm without the eigenstructure decomposition of the covariance matrix. This algorithm offers far lower computation cost compared to MUSIC at the expense of little performance decrease. With parallel preprocessing focusing on correlation matrices estimation and spectral peak search, an FPGA implementation is introduced, and proved to be efficient through theoretical analysis, simulation and hardware implementation.