Richard Stottler, Stottler Henke Associates, Inc.; Rakshith Mandayam, Stottler Henke Associates, Inc.; Brandon Liu, Stottler Henke Associates, Inc.; William Lin, Stottler Henke Associates, Inc.
https://doi.org/10.64861/WUPH2040
Keywords: Space Situational Awareness (SSA), Space Domain Awareness (SDA), Cislunar, Routing Protocol, Link Optimization, Delay/Disruption Tolerant Network (DTN), Scheduling, Reinforcement Learning (RL), Communication Network
Abstract:
Cislunar Space Situational Awareness (SSA) and Space Domain Awareness (SDA) will depend on the timely and reliable movement of data among heterogeneous spacecraft, sensors, relays, and ground systems operating over intermittent, bandwidth-constrained links. This paper presents an approach for autonomous data communication scheduling, routing, and link optimization intended to improve the efficiency and resilience of such networks.
The scheduling approach operates in a distributed manner, allowing network nodes to coordinate schedules, propagate changes, resolve resource conflicts, maintain custody of data, and dynamically reschedule traffic in response to unexpected outages or newly arriving data. The routing and scheduling algorithms incorporate congestion awareness, support multicast communications, and use predicted future traffic to reserve network capacity and improve decisions before packets arrive. An accompanying reinforcement learning (RL)-based link optimization capability adapts communication parameters to changing link conditions.
Our software, titled DREAMS (Delay/Disruption Tolerant Reinforcement Learning and Aurora Based Communication System), has been integrated with the Common Open Research Emulator (CORE) network simulator paired with the Extendable Mobile Ad-hoc Network Emulator (EMANE), a high-fidelity wireless packet transmission emulator. DREAMS has also been integrated with NASA’s High-rate Delay Tolerant Network (HDTN) implementation, enabling high-fidelity evaluation using existing networking protocols and software capable of running in space.
Experimental results show an increase in throughput of about 20% compared to the fastest-route baseline, as well as the ability to improve throughput when network traffic is predictable. The results demonstrate the value of combining congestion-aware routing, distributed autonomous scheduling, traffic prediction, and standards-compatible networking to support resilient data movement for future cislunar SSA/SDA architectures.
Date of Conference: September 15-18, 2026
Track: AI for SDA Applications