I Can Hear You Coming: RF Sensing for Uncooperative Satellite Evasion

2025-04-04

AI Summary

In non-cooperative satellite confrontations, autonomous collision avoidance remains inadequate, while conventional space situational awareness (SSA) suffers from high dependency and resource constraints. Method: This paper establishes a “cat-and-mouse” game-theoretic framework and, for the first time, integrates intercepted radio-frequency (RF) communication signals with spacecraft dynamical states as multimodal inputs, proposing a lightweight multimodal reinforcement learning (RL) avoidance paradigm. The approach synergizes proximal policy optimization (PPO) and soft actor-critic (SAC), RF temporal feature extraction, multimodal fusion modeling, and an MPC/A* variant obstacle-avoidance strategy optimized using real-world Space Surveillance Network (SSN) measurements. Results: Evaluated on realistic low-Earth-orbit (LEO) trajectory data, the method achieves a 37% higher collision avoidance success rate than conventional approaches, with response latency under 8 seconds—enabling real-time, adaptive, and cost-effective satellite deployment in adversarial scenarios.

Technology Category

• Multiagent Systems: Adversarial Agents
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Uncooperative satellite engagements with nation-state actors prompts the need for enhanced maneuverability and agility on-orbit. However, robust, autonomous and rapid adversary avoidance capabilities for the space environment is seldom studied. Further, the capability constrained nature of many space vehicles does not afford robust space situational awareness capabilities that can inform maneuvers. We present a"Cat&Mouse"system for training optimal adversary avoidance algorithms using Reinforcement Learning (RL). We propose the novel approach of utilizing intercepted radio frequency communication and dynamic spacecraft state as multi-modal input that could inform paths for a mouse to outmaneuver the cat satellite. Given the current ubiquitous use of RF communications, our proposed system can be applicable to a diverse array of satellites. In addition to providing a comprehensive framework for an RL architecture capable of training performant and adaptive adversary avoidance policies, we also explore several optimization based methods for adversarial avoidance on real-world data obtained from the Space Surveillance Network (SSN) to analyze the benefits and limitations of different avoidance methods.

Problem

• Enhancing on-orbit maneuverability against uncooperative satellites
• Developing autonomous evasion using RF sensing and RL
• Improving space situational awareness for constrained spacecraft

Innovation

• Reinforcement Learning for satellite evasion
• RF communication as multi-modal input
• Optimization methods for adversarial avoidance