Abstract: In the presence of signal steering vector mismatches,least mean squares(LMS) algorithm displays such problems as low convergence speed,degraded output performance and instability.In order to overcome the shortages and to improve the traditional LMS algorithm,this paper presents a robust constrained-LMS algorithm based on Bayesian approach.By using prior knowledge,the proposed algorithm can estimate the actual signal steering vector,thus effectively reduces the influence of signal steering vector mismatches and improves the system robustness.The mean output array signal-to-interference-plus-noise ratio(SINR) is improved,and is closer to the optimal value.Simulation results are given to demonstrate the effectiveness and feasibility of the proposed algorithm.