Multi-Turn Reasoning LLMs for Task Offloading in Mobile Edge Computing

TL;DR

This paper introduces COMLLM, a foresighted decision-making framework for task offloading in Mobile Edge Computing, leveraging large language models with multi-step look-ahead to optimize latency and load balancing.

Contribution

It presents a novel generative approach combining policy optimization and simulation for scalable, long-term MEC task offloading without retraining on new topologies.

Findings

COMLLM achieves near-optimal latency performance.

It outperforms existing SFT, DRL, and heuristic methods.

The model generalizes to larger, unseen network topologies.

Abstract

Emerging computation-intensive applications impose stringent latency requirements on resource-constrained mobile devices. Mobile Edge Computing (MEC) addresses this challenge through task offloading. However, designing effective policies remains difficult due to dynamic task arrivals, time-varying channels, and the spatio-temporal coupling of server queues. Conventional heuristics lack adaptability, while Deep Reinforcement Learning (DRL) suffers from limited generalization and architectural rigidity, requiring retraining when network topology changes. Although Large Language Models (LLMs) offer semantic reasoning capabilities, standard Supervised Fine-Tuning (SFT) yields myopic policies that greedily minimize immediate latency without accounting for long-term system evolution. To address these limitations, we propose COMLLM, a generative framework that enables foresighted decision-making in MEC systems. COMLLM integrates Group Relative Policy Optimization (GRPO) with a Look-Ahead Collaborative Simulation (LACS) mechanism, which performs multi-step Monte Carlo rollouts while jointly modeling server queue dynamics. By incorporating these rollouts into the reward design, the framework captures the long-term impact of current decisions on future system states. Experimental results demonstrate that COMLLM achieves near-optimal latency and improved load-balancing fairness. Notably, it exhibits zero-shot topological scalability, allowing a model trained on small-scale networks to generalize to larger, unseen topologies without retraining, outperforming SFT, DRL, and heuristic baselines.