In the presence of input interference, the Wiener solution for impulse resp
onse estimation is biased. It is proved that bias removal can be achieved b
y proper scaling of the optimal filter coefficients and a modified least me
an squares (LMS) algorithm is then developed for accurate system identifica
tion in noise. Simulation results show that the proposed method outperforms
two total least squares (TLS) based adaptive algorithms under nonstationar
y interference conditions.