Control of sensing by navigation on information gradients
Sofia Suvorova, Bill Moran, Stephen D. Howard, Douglas Cochran
Abstract
Sofia Suvorova, Bill Moran, Stephen D. Howard, Douglas Cochran
Abstract
In estimation of parameters residing in a smooth manifold from sensor data, the Fisher information induces a Riemannian metric on the parameter manifold. If the collection of sensors is reconfigured, this metric changes. In this way, sensor configurations are identified with Riemannian metrics on the parameter manifold. The collection of all Riemannian metrics on a manifold forms a (weak) Riemannian manifold, and a smooth trajectory of sensor configurations manifests as a smooth curve in this space. This paper develops the idea of sensor management by following trajectories in the space of sensor configurations that are defined locally by gradients of the metric this space inherits from the abstract space of all Riemannian metrics on the parameter manifold. Theory is developed and computational examples that illustrate sensor configuration trajectories arising from this scheme are presented.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In estimation of parameters residing in a smooth manifold from sensor data, the Fisher information induces a Riemannian metric on the parameter manifold. If the collection of sensors is reconfigured, this metric changes. In this way, sensor configurations are identified with Riemannian metrics on the parameter manifold. The collection of all Riemannian metrics on a manifold forms a (weak) Riemannian manifold, and a smooth trajectory of sensor configurations manifests as a smooth curve in this space. This paper develops the idea of sensor management by following trajectories in the space of sensor configurations that are defined locally by gradients of the metric this space inherits from the abstract space of all Riemannian metrics on the parameter manifold. Theory is developed and computational examples that illustrate sensor configuration trajectories arising from this scheme are presented.
Key concepts: Statistical manifold, Information geometry, Manifold (fluid mechanics), Riemannian manifold, Pseudo-Riemannian manifold, Metric (unit), Riemannian geometry, Fisher information metric