Non asymptotic estimation lower bounds for LTI state space models with Cram\'er-Rao and van Trees

TL;DR

This paper establishes sharp non-asymptotic lower bounds for the estimation error in LTI state-space models with Gaussian noise, extending classical bounds to more general settings and providing explicit constants.

Contribution

It introduces new non-asymptotic lower bounds for estimation risk in LTI models, incorporating general noise covariances and explicit constants, with novel concentration and geometric techniques.

Findings

Bounds are sharp when the system matrix has no eigenvalues on the unit circle.

Results extend classical bounds to systems with general noise covariance.

Bounds are rate-optimal under certain spectral conditions.

Abstract

We study the estimation problem for linear time-invariant (LTI) state-space models with Gaussian excitation of an unknown covariance. We provide non asymptotic lower bounds for the expected estimation error and the mean square estimation risk of the least square estimator, and the minimax mean square estimation risk. These bounds are sharp with explicit constants when the matrix of the dynamics has no eigenvalues on the unit circle and are rate-optimal when they do. Our results extend and improve existing lower bounds to lower bounds in expectation of the mean square estimation risk and to systems with a general noise covariance. Instrumental to our derivation are new concentration results for rescaled sample covariances and deviation results for the corresponding multiplication processes of the covariates, a differential geometric construction of a prior on the unit operator ball of small Fisher information, and an extension of the Cram\'er-Rao and van Treesinequalities to matrix-valued estimators.