Optimization-as-a-Service via Multi-Agent Large Language Model for Radio Access Networks

The article presents a novel approach to physical resource block (PRB) allocation in Radio Access Networks (RANs) through an Optimization-as-a-Service (OaaS) framework utilizing a multi-agent large language model (LLM-MA). This system features a closed-loop architecture with agents that dynamically formulate optimization problems and objectives, incorporating a one-shot reflection distillation mechanism to minimize computational latency. Experimental results indicate that the proposed framework achieves near-optimal resource allocation with significantly reduced inference latency, addressing the challenges posed by the dynamic conditions of sixth-generation (6G) environments.