Aiming at the case of the known knowledge of the direction of arrival of the desired signal,by using the constant modulus feature,a new blind adaptive beamforming algorithm is proposed under the frame of the Kalman filter. According to the Lagrange multiplier method,the optimal estimation expression of the system states can be derived by minimizing the constrained cost function. Then,the optimal weight vector of the adaptive beamformer can be obtained by using the iteration and update equations of the unscented Kalman filter. In the simulation,the proposed algorithm is compared with the constrained constant modulus recursive least square (CCM-RLS) and constrained minimum variance recursive least square (CMV-RLS) to demonstrate its effectiveness in the terms of the convergence speed,signal to interference plus noise ratio,robustness to changeable environments and tracking capability in the non-stationary condition.