For the multisensor systems with unkonwn noise statistics, using the modern time series analysis method,based on on- line identification of the autoregressive moving average(ARMA)innovation model,and based on the solution of the matrix equations for correlation function, the noise statistics can on- line be estimated, and further under the linear minimum variance optimal information fusion criterion weighted by scalars, a self- tuning information fusion Kalman filter weighted by scalars is presented . It has asymptotic optimality,and its accuracy is higher than each local self- tuning Kalman filter. Its algorithm is simple,and is suitable for real time applicatons. A simulation example for a target tracking system shows its effectiveness.