Generalized Query Planner for Distribute Fusion Navy SBIR FY2013.2 Sol No.: Navy SBIR FY2013.2 Topic No.: N132-135 Topic Title: Generalized Query Planner for Distribute Fusion Proposal No.: N132-135-0123 Firm: Harmonia Holdings Group 2020 Kraft Drive, Suite 1000 Blacksburg, Virginia 24060-6491 Contact: Rich Kadel Phone: (540) 951-5900 Web Site: www.harmonia.com Abstract: Our military's ability to dominate the battlespace depends on the ability to first "see" the battlespace-that is, for our military commanders to have awareness of the environment, entities, activities, and intent of both our adversaries and ourselves. The things we don't control must be observed, interpreted, and analyzed to understand a situation and predict potential outcomes. "Observations" come from the sensors our expeditionary, afloat, and air forces bring with them into contested areas; as well as from cooperating services and agencies with remote sensors, data centers, and analytical capabilities. In contested areas, the sensor capabilities will likely overwhelm our ability to centralize information. We must move analytics closer to the sensors, while still providing the commander a fused view of the battlespace. We propose a new Fusion Framework that leverages Harmonia's technology for distributing MapReduce jobs across geographically distributed clusters, and our partner CUBRC's extensive applied research in fusion to create the Generalized Query Planner for Distributed Fusion (GQPDF). The GQPDF query planner will use both shared and machine-learned knowledge of data localization to route analytical processes (e.g., MapReduce) to the best combination of tactical or enterprise fusion-data nodes, as one-time queries, or standing subscriptions for continuous awareness. Benefits: Harmonia's Generalized Query Planner for Distributed Fusion (GQPDF) will enable the warfighter access to the full spectrum of intelligence on our adversary, the environment, and the dynamic situation by fusing a variety of information collected at remote tactical sites, and information available from national sources. With limited bandwidth due to geographic or adversary-imposed communication constraints and increasing amounts of data inputs at the tip of the spear, GQPDF supports a distributed, step-wise fusion process to minimize the requirements to move information to common locations, routing the most relevant information available to support centralized, global or theater-wide situational awareness. We will develop a generalized framework to allow different fusion engines to be used collectively, both within a single cluster and as applied across clusters. The Harmonia Team will create a new variant of database "query planning" techniques, informed by machine-learned characteristics and locality of available data, to construct optimized query plans using a normalized language to drive the fusion processes and results. Return