The performance of least mean squares（LMS） algorithm degraded dramatically in the presence of even slight steering vector mismatches.In order to overcome the shortage

a novel robust adaptive beamforming algorithm based on the variable diagonal loading technique was proposed.The signal steering vector was obtained via the gradient-descent method and the diagonal loading term was incorporated at each recursive step.Then

the optimal weight vector was de-rived.To account for the contradiction of convergence rate and steady errors

a function between the step size and input signals was built to obtain the variable step size.The proposed algorithm had a low complexity cost without the inverse matrix and eigendecomposition.The proposed algorithm offered faster convergence rate

provided a sufficient robustness against the mismatches and made the mean output array SINR consistently close to the optimal one.The theoretical analysis and simulation results demonstrate that the performance of the proposed algorithm can outperform that of the conventional algorithm.