Big Data Mining for Maritime Situational Awareness - 20-029

The Minotaur Family of Services (MFoS) provides correlation tools and some degree of automatic processing for radar, electronic, video, and Automatic Identification System (AIS) signals to help manage thousands of tracks, yet the operator needs additional tools to identify the most important tracks.  The Tactical All-Source Replay/Repository (TASR) within Minotaur can leverage advances in processing power, big data handling, and Artificial Intelligence (AI) / Machine Learning (ML) algorithms to highlight these threat contacts based upon their tactical relevance.  This will provide Minotaur operators with usable cues to act upon: focusing available sensors (e.g., Inverse Synthetic Aperture Radar (ISAR) or Electro Optical / Infrared (EO/IR) scans), augmenting information with additional external intelligence feeds, and sharing first-pass intelligence and targeting data.  To get to this state, a prudent implementation plan must take into account raw sensor ability, off-board feeds, message formats, network characteristics, algorithm training, post-processing, storage requirements, hardware upgrade cycles, and finally human-in/on-the-loop concerns. Metron will create a mission scenario to use as a canvas for employing data mining and ML algorithms, and for determining the tradespace where incorporating big data mining, filtering, and tactically relevant cueing increases mission performance. The main vignette will require sifting through many tracks to find a covert surveillance ship amongst legitimate cruise and container ships.  This mission scenario will include simulation using the Executable Architecture Management System (ExAMS), which seamlessly integrates a data-centric framework with a robust mission simulation tool.  It provides traceability between changes to sensor performance, data architecture, and information flow to mission impacts tested in a discrete-event, Monte Carlo simulation environment.   ExAMS has sophisticated tracking algorithms and information management functionality, which will allow it to serve as an excellent surrogate for MFoS.  Phase I will inform the following objective questions: How can the availability of additional sensor feeds and big data analytic techniques help an operator perform tactical tasks such as finding dark targets, identifying anomalous behavior, and confirming ID? Phase I data feeds include historical AIS, other open source data, and sensor/track data generated within ExAMS.   What are the data conditioning and algorithm training functions required to employ representative AI/ML data mining techniques in a practical mission scenario? Phase I algorithms will be run internal and external to the ExAMS model.