An improved adaptive noise cancelling method

The limitations of Widrow-Hoff's adaptive noise cancelling (ANC) LMS method, in some cases, are discussed and an improved algorithm MLMS is presented. On this basis, the MMLMS algorithm with additional inertia is created and some calculation results by this method are given. It is concluded that the improved LMS algorithm (MLMS) can achieve the goal of noise cancelling by a filter of lower order due to the introduction of delay between signal v/sub 0/(n) and v/sub 1/(n). Compared to the LMS algorithm, the MLMS algorithm speeds up the convergence rate of iteration and remarkably reduces the amount of calculation. With a suitable choice of coefficient /spl alpha/, the MLMS algorithm with inertia term (MMLMS) can also raise the convergence rate of iteration. These algorithms are very useful for acoustic signal analysis.<
>