基于 GSVD 的分布式 MIMO 雷达测向算法

针对分布式多输入多输出（multi-input multi-output, MIMO）雷达测向中存在的数据信息提取不充分、运算量偏大等问题，开展了基于广义奇异值分解（generalized singular value decomposition, GSVD）的测向算法研究，以提高低信噪比条件下的角度估计性能。首先，建立了分布式阵列MIMO雷达回波信号的统一化表征模型；其次，将分布式MIMO雷达系统接收阵列数据的多线程GSVD问题转换为一个联合优化问题，运用交替最小二乘（alternating least squares, ALS）技术实现阵列信号流行矩阵的拟合，并引入子空间类算法实现目标角度联合估计；最后，对优化问题增加l1范数约束，避免了每次迭代中进行的奇异值分解运算，降低了算法运算量。仿真实验从角度联合估计、均方误差、运算时间等方面验证了所提算法的有效性。

GSVD-based distributed MIMO radar direction finding algorithm

In order to solve insufficient data information extraction and large amount of operations in distributed multi-input multi-output(MIMO) radar direction-finding, this paper studied the direction-finding algorithm based on generalized singular value decomposition(GSVD), so as to improve the performance of target angle estimation under low signal-to-noise ratio (SNR). Firstly, the distributed MIMO radar echo signal model was established. Then, the multilinear GSVD problem of the receiving array data was converted into an optimization problem and alternating least squares(ALS) algorithm was applied to solve it, achieving the fitting of channel matrix. Besides, the subspace algorithm was introduced to realize the joint estimation of the target angles. Finally, the l₁ constraint was provided to avoid the singular value decomposition operation in each iteration and the computational complexity was reduced. Simulation experiments have demonstrated that the effectiveness of the proposed algorithm in terms of joint angle estimation, mean square error and operation time.