On performance limitations in estimation

In this paper, we address the performance limitation issues in estimation problems. Specifically, we study the performance limitations in four cases: (1) estimating the output of an LTI system under the corruption of a white noise, (2) estimating a Brownian motion input of an LTI system from the output measurement, (3) estimating the output of an LTI system under the corruption of a Brownian noise, (4) estimating a white noise input of an LTI system from the output measurement. In each case we find and characterize explicitly how the best achievable estimation errors may relate to certain simple, intrinsic system characteristics.