Abstract: Robustness of Constrained Cyclic Adaptive Beam-Forming (CCCAB) included in Cyclic Adaptive Beam-Forming (CAB) algorithms will become lower under bad signals environments because its correlation matrix is not very suitable for cyclostationary signals. An improved algorithm for CCAB is presented as a solution to the problem in this paper. The theoretic formula of the correlation matrix for cyclostationary signals is induced firstly, after the inexactness of the old correlation matrix and the positive effect of forgetting factor are analyzed, the recursive formula of the new correlation matrix is concluded. The comparsion of computer simulations shows that the improved algorithm can receive weak signals, depress interferences and decreace the sensitivity to cyclic frequency error more effectively.