| dc.description.abstract | Despite recent advancements in remote sensing technologies, the United States Coast Guard (USCG) and other organizations face significant challenges in detecting, identifying, and monitoring small, non-cooperating vessels at sea. Recent increases in daily maritime activity, both legal and illegal, have only intensified the demand for technological solutions that provide instantaneous position, velocity, and time data (PVTD) in near-real time to support an adequate response. Environmental and operational conditions, such as adverse weather, cloud cover, sea state, signal interference, poor lighting, and extended data processing times, continue to limit sensor performance and reduce the timeliness of PVTD data across these use cases. Although significant research around remote sensing technologies currently exists, there exists limited public research investigating the challenges and potential solutions associated with these specific use cases. Furthermore, the recent trend of more cost-efficient space launch costs and the potential incorporation of AI/ML algorithms and other tools for processing high-volume data sets lend the possibility or opportunity of considering new, larger constellation architectures not previously thought of due to previously high costs. Therefore, this thesis explores the challenges associated with identifying and monitoring small objects at sea and proposes a set of potential system architectures and concepts robust to dynamic environmental and operational conditions through an argument of plausibility that adequately addresses stakeholder needs using preexisting and emerging technologies. An analysis of hundreds of design permutations using a modeling framework capable of evaluating architectures under uncertainty identified three unique architectures that supported sub-hour data collection while maintaining cost-effectiveness. Additionally, a framework for implementation outlines actionable steps such as the phased deployment of potential architecture constellations, integration of AI/ML-enabled functionality to reduce overall latency, leveraging commercial launch partnerships to reduce timelines and costs, and considerations for single- vs. multi-sensor payloads when considering cost performance under uncertainty. Notably, the performance-cost analysis revealed that multi-sensor architectures, despite their common use in modern, small constellation systems, failed to appear on any Pareto frontiers, with system complexity and cost penalties outweighing performance gains, thus reinforcing the value of smaller, simpler single-sensor designs. | |
