Since the directions of arrival (DOAs) of interferences change rapidly in high dynamic conditions, the traditional anti-jamming algorithm based on the linearly constrained minimum power (LCMP) criterion cannot suppress the interferences effectively, which may lead to the acquisition failure of GNSS receivers. To slove this problem, a covariance matrix taper (CMT) null widening algorithm was proposed, which can make the DOAs of interferences always in the widening nulls. Firstly, the DOAs of interferences which are unpredictable in practice were assumed to be triangular distributed. Then, the conversion relationship between the virtual covariance matrix and the real covariance matrix was obtained according to the assumption. Consequently, the virtual sample covariance matrix was derived, based on which the weight vector was acquired. Finally, the propsed algorithm was compared with the algorithms that adopt uniform distribution model and Laplace distribution model. The research results show that the proposed algorithm can widen the nulls effectively. Compared with uniform distribution model and Laplace distribution model, the proposed model achieved a superior balance between the null width and the null depth, and it had a higher array output signal to interference plus noise ratio (SINR). Moreover, software GNSS receiver was used to validate the effectiveness of the proposed algorithm.