LLM-Driven Large-Scale Spectrum Access

TL;DR

This paper introduces a hierarchical LLM-based framework for large-scale spectrum management, enabling efficient high-dimensional reasoning and superior performance in dense wireless networks.

Contribution

It develops a novel hierarchical state serialization mechanism and a code-driven paradigm to overcome prompt length limitations and improve scalability.

Findings

Eliminates the cold-start bottleneck in spectrum optimization.

Achieves superior scaling laws and spectral utility in dense networks.

Outperforms classical solvers and heuristics under limited compute budgets.

Abstract

Efficient spectrum management in massive-scale wireless networks is increasingly challenged by explosive action spaces and the computational intractability of traditional optimization. This study proposes a Large-Scale LLM-Driven Spectrum Access (LSA) framework rooted in Group Relative Policy Optimization (GRPO). To overcome the computational collapse caused by ultra-long prompts in large-scale scenarios, we develop a hierarchical state serialization mechanism that synthesizes global environment statistics with localized critical constraints, enabling the LLM to perform high-dimensional reasoning within a bounded context window. Simulation results under strictly time-bounded inference protocols reveal that the code-driven paradigm eliminates the SFT cold-start bottleneck and leverages direct execution feedback to achieve superior scaling laws. The framework maintains robust spectral utility and generalization across varying network scales, yielding consistent and empirically superior performance over non-deterministic heuristics, and surpassing partitioned classical solvers in ultra-dense regimes under matched compute budgets.