TECHNOLOGY AREA(S): Air Platform, Battlespace, Electronics

ACQUISITION PROGRAM: PMA290 Maritime Surveillance Aircraft

OBJECTIVE: Augment traditional first order logic sensor resource management approaches by employing Bayesian inference approaches that leverage information that is accumulated over a surveillance mission in a confined area of interest.

DESCRIPTION: Recent advances [Ref 3] in machine learning (ML), deep learning (DL), and other artificial intelligence (AI) techniques have shown great promise in delivering significant improvement in radar system performance for both surveillance (detection and tracking) and imaging functions. So-called “cognitive” systems seek to combine the optimization of sensor resources and capabilities with ML and data mining techniques to provide an autonomous system that, given a high-level descriptor (e.g., mission plan, Operational Situation/Tactical Situation (OPSIT/TACSIT)), will automatically adjudicate the target environment and provide human operators with actionable information or even take certain actions on its own (e.g., modification of platform flight pattern) in response to what has been learned. The objective is to develop a software-based system, prototyped in MATLAB with the final product in Java, that can make any given radar system “cognitive” by automatically understanding its native hardware capabilities and executing the most appropriate radar function at any given moment in the operational timeline in response to dynamic in-situ conditions present in a typical Navy maritime surveillance environment. Bayesian inference is a strong basis for this application as learning or experience can be used to update the probability for a hypothesis that is guiding radar tasking. Many of the Navy maritime surveillance missions involve surveilling the same geographical area over a mission or across multiple missions. Such operations offer the opportunity to significantly enhance mission success through learning-based resource management. Clearly demonstrating how the proposed approach enhances performance beyond that possible from first order logic expert driven approaches and how the proposed approach is trained and maintained are considered critical.

PHASE I: Complete a top-level design and demonstrate the feasibility of its approach to improve sensor utilization using Bayesian inference as an addition to first order logic approaches in sensor resource management. In order to facilitate the analysis, the sensor suite may be limited to radar only but the approach should be easily expandable to other Navy sensor systems such as electro-optic/infrared (EO/IR) and electronic support measures (ESM). Perform an analysis that assumes operation of the airborne sensor system is in a geographically constrained operational area with multiple revisits over the course of a mission. Operational maritime environment information will be provided by the Navy. The Phase I effort will include prototype plans to be developed under Phase II.

PHASE II: Develop a prototype system based upon the Phase I design to provide and demonstrate that legacy radar systems can be modified to provide the improved timeline utilization and mission success. Perform performance assessments quantifying vessel tracking, track maintenance, and imaging using target layouts and behaviors representative of operational maritime environments provided by the Navy. Deliver a detailed report and prototype system.

PHASE III DUAL USE APPLICATIONS: Complete development, perform final testing, and integrate and transition the final solution to Naval airborne maritime surveillance platforms. The high-level control logic to be utilized here is applicable to a wide range of applications including law enforcement and border control surveillance operations.

REFERENCES:

1. Guerci, J. Cognitive Radar: The Knowledge-Aided Fully Adapted Approach. Boston: Artech House, 2010. http://www.gbv.de/dms/ilmenau/toc/629620326.PDF

2. Haykin, S. Cognitive Dynamics Systems: Radar, Control, and Radio. Canada: Cambridge University Press, 2012. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6218166

3. Abad, R. et.al. “Basic Understanding of Cognitive Radar.” IEEE ANDESCON, 19-21 October 2016. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7836270

KEYWORDS: Sensor Resource Management; Maritime Surveillance; Artificial Intelligence; Machine Learning; Command and Control; Cognition