Bayesian Variable Resolution Modeling for Autonomous Satellite Cluster Data Fusion

The micro-satellite cluster is a revolutionary technology with potential to transform the way we view and deploy space-based applications. Tightly-coupled, highly autonomous satellite clusters promise enhanced mission capabilities and marked reductions inoperational costs. However, ground-based data-processing and control present obstacles to the full realization of this technology.CleverSet proposes to apply its expertise in Bayesian methods for modeling and distributed data fusion, with support from Boeing Space Systems, to design, prototype, and evaluate a coherent distributed architecture for variable-resolution model-based datafusion for autonomous satellite clusters. The prototype system will efficiently predict, detect, and monitor ongoing processes and events, both internal (e.g. health status) and external (e.g. objects to be detected and/or classified).CleverSet's will apply variable resolution modeling using distributed relational Bayesian networks. Variable-resolution modeling provides the proven effectiveness of Bayesian data fusion in an efficient, scalable, distributed architecture. CleverSetproposes to exploit a recent breakthrough, relational Bayesian networks - a modular form of Bayesian networks ideally suited for dynamic tasks such as diagnosis, situation assessment, and distributed target identification. The results of the proposed work will be a concrete architecture for scalable, distributed data fusion, together with a near-flyable prototype instantiation of the architecture and a set of studies evaluating its performance. Distributed Bayesianinference will support scalable, distributed data-fusion with algorithms that provide real-time global data-fusion. Adaptive, model-based compression will provide efficient utilization of limited bandwidth by exploiting knowledge about the structure ofcommunications for distributed Bayesian data fusion to produce superior data compression performance. Model-based decision-theoretic control will provide efficient use of processor and communication bandwidth resources. The work performed in the Phase Ieffort will allow extension of CleverSet's enterprise system monitoring software to widely-distributed systems and systems of systems.