On zero locations for sampled stochastic systems

Zero locations of stochastic models have significant influence on optimal prediction algorithms as well as on the convergence speed of identification method. The possible zero locations for a sampled stochastic system are examined. In particular, it is shown and numerically illustrated that the sampled zeros can appear only in a restricted set within the unit circle. A complete analysis of second-order systems is given, and all possible zero locations are characterized. It is shown how the zero location in this case depends on the continuous-time model parameters and the sampling interval.<
>