A novel de-coherence algorithm,called cross-correlation vector reconstruction by changing the reference element(CVRCRE) algorithm, was proposed to estimate the DOA of coherent signals. Using cross-correlation information of received data between the whole array and the reference element, an equivalent covariance matrix which had excellent de-coherence performance was reconstruted. Combined with the subspace algorithms, such as MUSIC and ESPRIT, the DOA of coherent signals could be estimated. The proposed algorithm avoides the loss of array physical aperture. Computer simulation confirmed that the CVRCRE algorithm achieves better performance than Toeplitz algorithm and FBSS algorithm under scenarios of low SNR and deficient snapshots.
