Combining an autoregressive (AR) model with the FD-TD algorithm for improved computational efficiency

An autoregressive (AR) model is combined with the finite-difference time-domain (FD-TD) algorithm. It is shown that this greatly improves the computational efficiency of the FD-TD technique. The AR model can be set up using a short segment of the time-domain trace generated by the FD-TD algorithm. Through extrapolation, future segments of the response can be accurately predicted. It is shown that the AR model is more robust and accurate than other prediction techniques used with electromagnetic modeling. A number of numerical examples are shown.<
>