This paper focuses on jamming cancelling and signal enhancement for measurement and control signals based on adaptive beamforming. Firstly, the conventional LMS algorithm and its improved versions are introduced. The tradeoff between the speed of convergence of the LMS algorithm and its residual error floor is described, which causes difficulties in obtaining proper adaptation step size. To solve the computation complexity and low speed of convergence in dynamic communication environment, a new reduced complexity LMS algorithm based on second-order loop filter (LF-LMS) is proposed. The new algorithm improves the reliability of the adaptive beamforming array and reduces the error variance of convergence by using proper step size and variable error estimation. The reduced computation complexity can also simplify the realization on DSP platforms. Simulation results demonstrate the effectiveness of our proposed algorithm.