Аннотация
In reality, there are Multi-mode Noise (as a whole, they are non- Gaussian noise), it causes serious damage and loss to the signals. Traditional LMS algorithm can not suppress the noise well, not only slowing down the convergence, but also occurring a big steady-state error and the offset amount. This paper combined with the advantages of LMS algorithm and the Newton algorithm, to propose the amendment of autocorrelation matrix and take variable step-size which changes by the SNR. Simulation results show that the proposed algorithm in this paper has greatly improved in convergence rate, steady state error and the offset amount. It also can well suppress multi - mode noise to take useful signal.