To support heterogeneous services in 5G and beyond, multi-radio access technology (multi-RAT) mobile edge computing (MEC) systems must jointly handle user association and radio resource allocation under stringent delay requirements. This problem becomes particularly challenging when ultra-reliable low-latency communication (URLLC) and enhanced mobile broadband (eMBB) users coexist, since the former requires strict deadline satisfaction while the latter emphasizes low average latency under computation-intensive workloads. In this paper, we propose a hierarchical multi-RAT MEC framework with large language model (LLM)-assisted evolutionary computation (EC). The outer layer uses an LLM to generate user–base station association decisions, while the inner layer employs EC for bandwidth allocation under the given association pattern. This design combines the global reasoning capability of LLMs with the continuous optimization strength of EC. Simulation results show that the proposed LLM-hRAT scheme converges faster and achieves lower average cost than conventional single-layer EC methods and heuristic non-LLM baselines in both uniform and non-uniform user deployment scenarios.
Zhao-Kun Shao, Kangyu Gao, Zhi-hui Zhan et al.· International Conference on...· 0 citations
Large language models (LLMs) have made automated heuristic design (AHD) increasingly practical by generating executable heuristic code from task descriptions and evaluator feedback. Yet under a limited query and evaluation budget, search efficiency depends critically on a pre-generation decision. Before each LLM query and black-box evaluation, the system must choose which archived heuristics to reuse as parents and which generation operator should transform them. Existing methods typically choose such actions with predefined rules, leaving the expected outcome of each concrete operator-parent action only indirectly modeled. Therefore, we propose \emph{\fullmethod{}} (\method{}), a surrogate-guided action-selection module for operator-parent selection in LLM-based AHD. \method{} guides the LLM code-generation process by scoring pre-generation actions with two complementary surrogates. Specifically, a transition surrogate is proposed to predict the latent distribution of the child representation induced by an operator-parent action, while an instance-conditioned utility surrogate is proposed to estimate the expected performance of sampled child latents. Moreover, we propose an uncertainty-aware acquisition rule that combines predicted utility, utility uncertainty, and transition uncertainty to select the next LLM generation action. Across a diverse heuristic-design suite, \method{} is competitive with strong LLM-AHD baselines, and ablation and action-selection analyses suggest that its behavior goes beyond simple archive ranking or fixed operator preferences.
Yuhang Wang, Chaoda Peng, Xingyu Wu et al.· 0 citations