Wireless Power Control Based on Large Language Models
Authors: Jiacheng Wang, Yucheng Sheng, Le Liang, Hao Ye, Shi Jin
Organizations: School of Information Science and Engineering, Southeast University, Nanjing 210096, China · Department of Electrical and Computer Engineering, University of California, Santa Cruz, CA 95064, USA
This paper investigates the power control problem in wireless networks by repurposing pre-trained large language models (LLMs) as relational reasoning backbones. In hyper-connected interference environments, traditional optimization methods face high computational cost, while standard message passing neural networks suffer from aggregation bottlenecks that can obscure critical high-interference structures. In response, we propose PC-LLM, a physics-informed framework that augments a pre-trained LLM with an interference-aware attention bias. The proposed bias tuning mechanism injects the physical channel gain matrix directly into the self-attention scores, enabling explicit fusion of wireless topology with pre-trained relational priors without retraining the backbone from scratch. Extensive experiments demonstrate that PC-LLM consistently outperforms both traditional optimization methods and state-of-the-art graph neural network baselines, while exhibiting exceptional zero-shot generalization to unseen environments. We further observe that topology-relevant relational reasoning is concentrated in shallow layers, whereas deeper layers encode task-irrelevant semantic noise. Motivated by this finding, we develop a lightweight adaptation strategy that reduces model depth by 50%, significantly lowering inference cost while preserving state-of-the-art spectral efficiency.
Figures & tables
Fig. 1: Graph representation of the wireless interference network. Communication links within transmitter-receiver pairs are modeled as nodes, while the mutual interference between pairs is represented by directed edges, constituting a fully connected interference graph G .
Fig. 2: Overall architecture of the proposed PC-LLM. The framework aligns physical channel features with the latent space of the model through an input projection layer, a bias-modulated Transformer backbone, and a power inference head.
Fig. 3: Performance comparison of the proposed PC-LLM framework against baseline algorithms across varying network densities (number of D2D pairs K ) and interference environments (distance ranges [dmin,dmax] ). The results are organized by optimization objectives: (a)–(c) sum-rate maximization, (d)–(f) proportional fairness, and (g)–(i) harmonic maximization. The performance is normalized with respect to the WMMSE-Best algorithm for the max sum rate and proportional fairness tasks, whereas the harmonic mean rate is normalized relative to the optimal PCGNN-Large baseline. For learning-based methods, only the best performance achieved under their respective optimal training configurations is reported.
Fig. 4: Performance comparison of the proposed PC-LLM and baseline algorithms under multi-dataset (darker bars) and single-dataset (lighter bars) training configurations. The performance is normalized with respect to the WMMSE-Best algorithm for the max sum rate and proportional fairness tasks, whereas the harmonic mean rate is normalized relative to the multi-dataset trained PCGNN-Large baseline. The bar heights represent the average performance across 15 distinct scenarios with varying user densities and channel distributions. The error bars denote the standard deviation, quantifying the performance variance and validating the universal robustness of each method against extreme macro-level topological heterogeneity.
Optimization Objective
Algorithm
Number of D2D Pairs K
20
35
50
65
80
Max Sum Rate
PC-LLM (zero-shot)
4.7741
3.8514
3.3893
3.0993
2.9070
PC-LLM (full-shot)
4.7731
3.8183
3.3760
3.0914
2.9011
WMMSE-Best
4.7452
3.8152
3.3561
3.0701
2.8811
Prop. Fairness
PC-LLM (zero-shot)
2.1553
1.3818
1.0601
0.8838
0.7776
PC-LLM (full-shot)
2.1536
1.3511
1.0482
0.8830
0.7702
TABLE I: Generalization Performance Across Varying Network Densities with [dmin,dmax]=[1,100] m
Fig. 5: Performance comparison with varying numbers of LLM layers L under the setup of K=50 and [dmin,dmax]=[2,65] m. The performance is normalized with respect to the WMMSE-Best algorithm for the max sum rate and proportional fairness tasks, whereas the harmonic mean rate is normalized relative to PCGNN-Large.
Fig. 6: Ablation results of the proposed PC-LLM framework with varying numbers of D2D pairs K under the [dmin,dmax]=[2,65] m scenario. The performance is normalized with respect to the WMMSE-Best algorithm.
Integrating large language models (LLMs) into wireless communication optimization is a promising yet challenging direction. Existing approaches either use LLMs as black-box solvers or code generators, tightly coupling them with numerical computation. However, LLMs lack the precision required for physical-layer optimization, and the scarcity of wireless training data makes domain-specific fine-tuning impractical. We propose BeamAgent, an LLM-aided MIMO beamforming framework that explicitly decouples semantic intent parsing from numerical optimization. The LLM serves solely as a semantic translator that converts natural language descriptions into structured spatial constraints. A dedicated gradient-based optimizer then jointly solves the discrete base station site selection and continuous precoding design through an alternating optimization algorithm. A scene-aware prompt enables grounded spatial reasoning without fine-tuning, and a multi-round interaction mechanism with dual-layer intent classification ensures robust constraint verification. A penalty-based loss function enforces dark-zone power constraints while releasing optimization degrees of freedom for bright-zone gain maximization. Experiments on a ray-tracing-based urban MIMO scenario show that BeamAgent achieves a bright-zone power of 84.0,dB, outperforming exhaustive zero-forcing by 7.1 dB under the same dark-zone constraint. The end-to-end system reaches within 3.3 dB of the expert upper bound, with the full optimization completing in under 2 s on a laptop.
Xiucheng Wang, Yue Zhang, Nan Cheng
State Key Laboratory of ISN and School of Telecommunications Engineering, Xidian University, Xi’an 710071, China
The aggressive densification of modern wireless networks necessitates judicious resource allocation to mitigate severe mutual interference. However, classical iterative algorithms remain computationally prohibitive for real-time applications requiring rapid responsiveness. While recent deep learning-based methods show promise, they typically function as task-specific solvers lacking the flexibility to adapt to different objectives and scenarios without expensive retraining. To address these limitations, we propose a graph foundation model for resource allocation (GFM-RA) based on a pre-training and fine-tuning paradigm to extract unified representations, thereby enabling rapid adaptation to different objectives and scenarios. Specifically, we introduce an interference-aware Transformer architecture with a bias projector that injects interference topologies into global attention mechanisms. Furthermore, we develop a hybrid self-supervised pre-training strategy that synergizes masked edge prediction with negative-free Teacher-Student contrastive learning, enabling the model to capture transferable structural representations from massive unlabeled datasets. Extensive experiments demonstrate that the proposed framework achieves state-of-the-art performance and scales effectively with increased model capacity. Crucially, leveraging its unified representations, the foundation model exhibits exceptional sample efficiency, enabling robust few-shot adaptation to diverse and unsupervised downstream objectives in out-of-distribution (OOD) scenarios. These results demonstrate the promise of pre-trained foundation models for adaptable wireless resource allocation and provide a strong foundation for future research on generalizable learning-based wireless optimization.
Yucheng Sheng, Jiacheng Wang, Le Liang +2
National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, China · Department of Electrical and Computer Engineering, University of California, Santa Cruz, CA 95064, USA
Next-generation large language models (LLMs) are expanding from the cloud to ubiquitous edge devices. However, edge devices typically either lack the memory to store increasingly large LLM weights or, even with enough memory, spend unaffordable energy on loading the weights. This raises our question: can an edge device run an LLM without storing or loading its weights, but receive them over the air and consume them on the fly? Inspired by wireless broadcasting, we present AIR-LLM, an LLM inference architecture for edge devices, which is composed of: (i) a central radio (e.g., 5G base stations) that broadcasts the LLM weights into the air, and (ii) the edge user that receives the weights and completes the general matrix-vector multiplication (GEMV) of LLM inference directly in the radio frequency (RF) domain using RF mixers. To further shorten the airtime, AIR-LLM exploits MIMO spatial multiplexing and proposes an energy-efficient precoder-postcoder pair on the edge to calibrate its own wireless channel. Since the central radio stays user-unaware, AIR-LLM is user-scalable so that one broadcast serves unlimited users within its coverage. We implement AIR-LLM on the NVIDIA Sionna ray-traced channels of two real-world urban scenes and the profiling of a real RF mixer. With a WikiText-2 perplexity degradation of 4.0% on LLaMA-3.1-8B, AIR-LLM saves the energy by 157.7x/40.4x against the FP16 and weight-only quantization baselines; with 20 users, its airtime is 104.1x/26.0x shorter, respectively.
Zhihui Gao, Tingjun Chen, Dirk Englund
Massachusetts Institute of Technology Cambridge, MA, USA · Duke University Durham, NC, USA