LLM-Driven Automated Reward Design for Reinforcement Learning-Based Routing in LEO Satellite Networks
2026-08-03 • Networking and Internet Architecture
Networking and Internet Architecture
AI summaryⓘ
The authors address the challenge of routing data through fast-changing Low Earth Orbit (LEO) satellite networks using reinforcement learning (RL), which depends heavily on designing good reward functions. They introduce LARGE, a system that uses large language models (LLMs) to create and improve these reward functions automatically by testing them in simulations and learning from the results. Their method reaches performance close to expert-made designs without needing manual tuning. This shows that combining LLMs with feedback loops can help automate complex RL tasks in dynamic environments like satellite networks.
Low Earth Orbit (LEO)satellite networksroutingreinforcement learning (RL)reward functionlarge language models (LLMs)simulation-in-the-loopgoodputend-to-end delayautomated reward design
Authors
Walter P. Casas, Nelson L. S. da Fonseca, and Carlos A. Astudillo
Abstract
Routing in Low Earth Orbit (LEO) satellite networks is challenging due to highly dynamic topologies and spatio-temporal network conditions. Reinforcement Learning (RL) has emerged as a promising approach for adaptive routing; however, its performance critically depends on reward function design, which must balance objectives such as goodput and end-to-end delay. In practice, reward design remains a complex manual process requiring significant domain expertise and extensive trial-and-error. Recent works have explored Large Language Models (LLMs) for automated reward design, but their application to highly dynamic systems such as LEO satellite networks remains largely unexplored. We propose LARGE, a framework that automates reward design for RL-based routing by combining LLM- driven generation with iterative simulator-in-the-loop evaluation. LARGE generates an initial reward from LLM prior knowledge and iteratively refines it using simulation feedback. This loop enables exploration of diverse reward formulations while aligning them with network objectives. Results show that LARGE improves reward quality within a few iterations through feedback-driven refinement. Across different backbones, the framework achieves performance comparable to an expert-designed baseline, with the best-performing configuration reaching goodput within approximately 3% of the baseline and slightly lower end-to-end delay, without manual reward engineering. These results indicate that effectiveness emerges from the iterative feedback-driven process enabled by LARGE, highlighting the potential of framework-driven LLM-in-the-loop optimization for RL-based routing in dynamic satellite networks.