Abstract: PVFS (Particle Velocity Field Smoothing) algorithm based on acoustic vector sensor array is an effective decorrelation algorithm. However, the performance will seriously degrade or even fails when there are a large number of coherent sources. Based on PVFS, the MSS (Matrix Square Smoothing) algorithm is proposed as the amelioration of PVFS. The proposed algorithm first squared and partitioned the data covariance matrix constructed by PVFS. Then, the partitioned matrix was cross-multiplied by each other. Finally, the decorrelation ability of PVFS algorithm was enhanced and more coherent sources can be distinguished. Computer simulation indicated that the proposed algorithm achieved a similar performance as the Spatial Smoothing PVFS algorithm. Moreover, the DOA estimation accuracy of the proposed algorithm was much higher when the signal-to-noise ratio was low.