A systemic approach to cislunar spatial domain awareness: building intuition towards improved fidelity, adaptability, and constellation geometry

Abstract

With a significant proliferation of space objects expected in cislunar space, legacySpatial Domain Awareness (SDA) systems like the Deep Space Network are ill-suited to track to a high level of accuracy nor deal with high traffic. Thus, the design of an improved cislunar SDA architecture has become a major focus for NASA and the USSF. This research takes a systemic approach to the design and optimization problem such that rigorous parametric analysis is achieved and critical trade offs are quantified to inform mission architecture. First an in-depth analysis of heuristic optimization and nonlinear filtering methods was conducted to build intuition into the dynamics of the system and the design space in general. The Lunar Gateway (Southern NRHO) coasting orbit, a time optimal low thrust transfer trajectory which transitions through the lunar gravitational influence, and a stable Southern Butterfly orbit were chosen as tracking targets for their ability to showcase different aspects of the design space. Custom genetic algorithms and the classic Extended Kalman Filter (EKF) in tandem were found to be the best techniques to solve the tracking embedded optimization problem. They also highlighted the non-linear dynamics that occur in close Lunar Orbit. Once a strong intuition was built, a deep dive into higher fidelity and more complex tracking systems was conducted. Looking at different geometries of space-based observer systems showed the limits of the most common Cislunar SDA systems. Then hybrid constellations fusing space based observers with Lunar Surface Observers (LSOs) were optimized, and proved to be more effective for tracking targets that get close to the lunar surface. Additionally, valuable insights into symmetry emerged with respect to target type and observer distributions over different sensor configurations. Armed with a deeper understanding of the interplay between system geometry, observer number, filter type and system dynamics, it was possible to introduce novel concepts which further exploited the existing architecture intuitions. Dynamic cadence was introduced to be an adaptive measurement system which does not invoke noisy measurements when pure physical propagation is sufficiently accurate, and incorporates denser measurements when the system becomes unpredictable. This essentially baselines every single optimized SDA configuration to a similar tracking error and uncertainty, just with varying measurement densities. Superior systems with more observers required seven times fewer measurements in the most complex case, thus speaking to the overall capacity and resource use of a particular system. Additionally, the selections for particular LSO geometries reemphasized the advantages of a hybrid constellation from a purely estimation perspective. Dynamic Cadence combined with extended mission modeling validated previous findings and improved the robustness of the solutions found, leading to optimization of adaptive architectures.