Abstract: To address the issue of multiple coherent signal sources incident on a polarization-sensitive array， an efficient direction of arrival （DOA） estimation method that decorrelates multiple coherent signal sources is proposed. By utilizing the cooperative effect of a moving platform and an orthogonal dipole array， the method reduces the required number of observations， thereby improving the efficiency and accuracy of DOA estimation for coherent signals. The algorithm first smooths the data received in the polarization domain by an array composed of multiple pairs of orthogonal dipoles and restores the rank of the signal source covariance matrix. Smoothing in the polarization domain effectively reduces the correlation between signals because of the strong polarization sensitivity of the array， which helps recover the rank of the covariance matrix. By moving the platform to induce motion into the arrays， the covariance matrix of the received signals was calculated for each moving array after polarization smoothing. Using a method similar to spatial smoothing， the average covariance matrix of these arrays was computed， to restore the signal covariance matrix to full rank. The multiple signal classification （MUSIC） algorithm was applied to the recovered covariance matrix for DOA estimation， thereby obtaining the direction information of multiple signal sources. By jointly resolving decorrelation in the polarization domain and smoothing in the time domain， the algorithm reduces the number of observations required for decorrelation by the moving platform to half the original number， thereby significantly improving the efficiency of DOA estimation for coherent signals. Simulation results demonstrated that the proposed algorithm effectively reduces the estimation errors caused by signal coherence from multiple coherent signal sources in DOA estimation， thereby significantly improving estimation accuracy. The simulation results validated the effectiveness of the proposed method. In conclusion， the proposed method successfully improves the efficiency and accuracy of DOA estimation for multiple coherent signal sources by leveraging joint decorrelation in the polarization domain and time domain smoothing.