ABEL: an Adaptive Belief Engine for Satellite Cluster Data Fusion

"To support autonomous scenarios, future constellations of satellites must manage multiple sources of information carrying various levels of uncertainty. Multi-mode payloads will be autonomously configured based on fusion of evidences provided byindependent cooperative agents. This will require an advanced architecture to loosely couple distributed knowledge sources. The Adaptive Belief Engine will concurrently manage uncertain information from onboard processing agents as part of a ClusterManager's intelligent reasoning. The same engine will concurrently support integrated Fault Management at the vehicle and/or cluster level. It will manage, corroborate or refute evidences with varying degrees of certainty. These evidences are dynamicallygathered from various diagnostics and prognostics engines providing unparalleled confidence in spacecraft automation. Results will be accumulated into the embedded shared database. Significantly, a cooperating expert system will evaluate rules,dynamically uploaded, that will trigger based on user-tunable thresholds of certainty associated with the current set of hypotheses. The real-time executive will then concurrently process specified scripts to intelligently task the cluster payload andelements or even recover from newly detected faults. This component will be integrated within a distributed blackboard architecture required to allow interchange of information across heterogeneous elements s