Real-time implementation of MCAO Kalman filters

Determine whether a Kalman filter for multi-conjugate adaptive optics can be implemented on a real-time computer.

Background

Kalman filter-based reconstructors are presented as potentially optimal because they incorporate temporal prediction and can model the spatial and temporal properties of atmospheric turbulence and measurement noise. However, their computational requirements remain a major obstacle for deployment in real-time MCAO systems. Existing approaches rely on approximations to the turbulence and noise models, such as the spatially invariant approximation used by the distributed Kalman filter, which can prevent accurate treatment of the non-uniform and correlated noise produced by elongated laser-guide-star spots. The paper therefore identifies real-time implementation of an MCAO Kalman filter as an unresolved issue, particularly when computational efficiency must be balanced against model fidelity.