超分辨测向中奇异值分解的FPGA实现

奇异值分解是超分辨测向技术的核心组成部分，现有的并行实现方案适用范围窄，运算量大，迭代时间长。为了满足测向接收机系统的高实时性需求，结合双边Jacobi算法的交换策略和单边Jacobi算法的求角结构，提出了一种改进的实现方法。该实现方法修正了脉动阵列的收敛性问题，提高了复数矩阵的收敛速度。同时，给出了算法的现场可编程门阵列(FPGA)实现结构。仿真结果证明该方案耗时在百微秒以内，能够应用于电子侦察设备。

FPGA implementation of singular value decomposition algorithm in super-resolution direction-finding

Singular value decomposition(SVD) is the core of super-resolution direction-finding(DF) technique.The existing parallel implementation methods have narrow applicability,intensive computation and lengthy iteration. In order to meet the application requirements of DF receiver system,a modified implementation method is proposed according to the exchange policy of bilateral Jacobi algorithm and the angle calculation structure of unilateral Jacobi algorithm. In this implementation method,the ergodic problem in Brent-Lu-Van(BLV) array is modified,and the convergence rate in complex matrix is improved.An algorithm structure on field programmable gate array(FPGA) is presented.Simulation results show that the method consumes less than 100 ms and can be applied to electronic reconnaissance equipment.