Abstract: ‍ ‍Adaptive beamforming is an effective approach for suppressing interferences at the array receiver. However， when the interferences from the different direction-of-arrival （DOA） are strongly coherent， general adaptive beamformers cannot build the weight vector orthogonal to the interference subspace， thus resulting in the failure of optimal interference suppression. The residual interference significantly affects the array output’s reliability for subsequent processing， such as target detection and localization. A robust adaptive beamformer using interference covariance matrix reconstruction and the spatial differencing technique （SDT） is proposed to address the considerable performance degradation of general adaptive beamformers in a complex coherent interference scenario where uncorrelated interferences and multiple groups of self-coherent interferences coexist. Based on the minimum variance distortionless response （MVDR） beamformer， the algorithm first constructs the uncorrelated interference-plus-noise covariance matrix （UNCM） with Capon spectral power estimation and then applies the UNCM to the MVDR beamformer to obtain a modified weight vector for suppressing uncorrelated interferences. Second， using the SDT， a coherent interference covariance matrix （CICM） for restoring the actual coherent interference subspace is constructed， and then the modified weight vector for the ideal suppression of coherent interferences is obtained by applying the CICM to the MVDR beamformer. Finally， the proposed beamformer for simultaneous suppression of all interferences in the received signal is derived by combining the two weight vectors above. Simulation results confirm that the proposed beamformer can adaptively obtain accurate and deep interference nulling without DOA information and presents low sidelobes， thus reflecting its better interference suppression performance compared with existing algorithms. Additionally， the proposed beamformer is robust against small snapshots and high SNRs in the received signal. Results of theoretical analysis show that the proposed beamformer can achieve the ideal interference suppression performance even when the number of interferences exceeds that of array sensors， which is supported by the simulation for a typical design case.