**1.1.2 Robust approaches for general mismatch**

Several other approaches are known to provide the improved robustness against more general types of mismatches, for example, the algorithms that use the diagonal loading of the sample covariance matrix [14][16], the eigenspace-based beamformer [11][30][31], and the covariance matrix taper (CMT) approach [32]-[34]. For the diagonal loading method, a serious drawback is that there is no reliable way to choose the diagonal loading level, F.Vincent and O.Besson propose the method to select the optimal loading level with a view to maximizing the signal-to-noise ratio (SNR) in the presence of steering vector errors and it is shown that the loading is negative, but they can't give the exact solution, instead of the approximate solution, moreover, they can't give the expression of steering vector errors [35]. The eigenspace-based approach is essentially restricted in its performance at low SNR and when the dimension of the signal-plus-interference subspace is high, and the dimension must be known in the latter technique [31]. The CMT approach is known to provide an excellent robustness in scenarios with nonstationary interferers, however, its robustness against mismatches of the desired signal array response may be unsatisfactory, furthermore, it can also be explained as the diagonal loading [33].
