基于数据预处理的改进GS正交化波束形成

针对常规Gram-Schmidt（GS）正交化算法在训练快拍中混有期望信号时，自适应波束会出现期望信号相消的问题，提出了基于数据预处理的改进GS正交化波束形成算法. 该算法构造阻塞矩阵进行数据预处理剔除期望信号，估计对应的协方差矩阵，并对其进行GS正交化重构干扰子空间，将静态加权矢量向干扰子空间作正交投影得到自适应权矢量. 同时，为准确估计干扰子空间，对协方差矩阵的正交化自适应门限进行了修正. 仿真结果表明，所提算法的输出信干噪比（SINR）比其它GS正交化算法有2 dB以上的性能改善. 

Improved Gram-Schmidt Orthogonalization Beamforming Algorithm Based on Data Preprocessing

When the desired signal is mixed in the training data, the conventional Gram-Schmidt orthogonalization beam-forming algorithm will result in the desired signal cancellation. In this paper, an improved Gram-Schmidt orthogonalization beam-forming algorithm based on data preprocessing was proposed to resolve the desired signal cancellation. In the proposed algorithm, the training data are firstly preprocessed to remove the desired signal by the designed block matrix, then the corresponding covariance matrix was estimated, and the interference subspace was reconstructed by Gram-Schmidt orthogonalization of the columns of the covariance matrix. Finally, the adaptive weight vector was obtained by orthogonally projecting the quiescent weight vector into the interference subspace. Moreover, the orthogonalization adaptive threshold of the covariance matrix was re-designed for accurate interference subspace estimation. Simulation results show that the output signal to interference plus noise ratio (SINR) of the proposed algorithm is improved above 2 dB comparing with the current Gram-Schmidt orthogonalization methods.