Bridging Network Fragmentation: A Semantic-Augmented DRL Framework for UAV-aided VANETs

Gaoxiang Cao; Wenke Yuan; Huasen He; Yunpeng Hou; Xiaofeng Jiang; Shuangwu Chen; Jian Yang

cs.AIMar 19, 2026

Bridging Network Fragmentation: A Semantic-Augmented DRL Framework for UAV-aided VANETs

Authors: Gaoxiang Cao, Wenke Yuan, Huasen He, Yunpeng Hou, Xiaofeng Jiang, Shuangwu Chen, Jian Yang

Organizations: Department of Automation, University of Science & Technology of China

Abstract

Urban Vehicular Ad-Hoc Networks (VANETs) can become fragmented because buildings obstruct wireless links and vehicle mobility continuously changes the network topology. Unmanned Aerial Vehicles (UAVs) can serve as mobile relays, but Deep Reinforcement Learning (DRL)-based deployment often suffers from inefficient exploration because it lacks road-topology guidance. To address this problem, we propose Semantic-Augmented DRL (SA-DRL), which models network fragmentation over the road topology and aligns a pretrained Large Language Model (LLM) to generate a topology-dependent action prior from dynamic traffic states. The resulting Semantic-Augmented PPO (SA-PPO) algorithm combines this prior with the PPO policy through Logit Fusion, guiding exploration toward promising intersections while retaining adaptation through environmental returns. Simulations driven by real-world urban trajectories show that SA-PPO reaches the final converged reward of Vanilla PPO using only 28.6% of its training episodes. It improves the average number of vehicles in connected components and the average connected-component size by 7.9% and 8.7%, respectively, while reducing UAV energy consumption by 21.3%.

Figures & tables

Fig. 2: The four-stage pipeline of the SA-DRL framework.

Fig. 3: Schematic diagram of the LoRA fine-tuning mechanism.

Fig. 4: Schematic diagram of the SA-PPO training and execution mechanism based on Logit Fusion.

Symbol	Parameter	Value
$\tau$	Time-slot duration	100 s
$T_{\tau}$	Total task duration	5000 s
$v_{u}$	UAV flight speed	20 m/s
$\varepsilon_{1}$	UAV hovering power	100 W
$\varepsilon_{2}$	UAV communication power	5 W
$\varepsilon_{3}$	UAV propulsion power	200 W

TABLE I: Simulation Environment Parameters

Fig. 5: The RTG corresponding to the road network used in simulation.

LLM	Qwen2.5	Qwen2.5	Qwen3	Llama3.2	Gemma3
Parameters	3B	3B	4B	3B	4B
Alignment	Prompt-only	LoRA	LoRA	LoRA	LoRA
BLEU-4	0.79	75.10	70.32	44.64	55.50
ROUGE-L	18.55	81.19	78.42	69.81	74.84
Kendall’s $\tau$	–	86.14	78.82	85.60	88.89
JSON PSR	0.00	100.00	100.00	100.00	100.00

TABLE II: Performance Comparison of Prompt-Only and LoRA-Based Knowledge Alignment. Kendall’s

\tau

, JSON PSR, and

HR_{10}

are reported as percentages.

Fig. 6: Inference efficiency comparison for different LLM architectures as batch size varies.

Fig. 7: Heatmap comparison of action scores predicted by different fine-tuned LLMs against ground truth.

Fig. 8: Comparison of action score predictions by Qwen2.5-3B against ground truth.

Fig. 9: Training reward curves for different algorithms.

Fig. 10: Comparison of VANET connectivity and UAV energy consumption metrics.

Panel of Figure 13 — Fig. 11: UAV trajectory visualization under different algorithms.

Panel of Figure 14 — Fig. 12: Ablation study results comparing SA-PPO with its variants.

Fig. 13: Sensitivity of SA-PPO to the semantic fusion coefficient

\lambda

. The reward and connectivity metrics are normalized by their values under the default setting

\lambda=0.5

, while the UAV energy consumption is reported in absolute values. “Decay” denotes a schedule that progressively reduces

\lambda

during training.

Average number of vehicles in connected components $\uparrow$
Method	08:00	10:00	12:00	14:00	16:00	18:00
Soft Actor-Critic	36.1	59.3	44.9	76.3	82.1	60.7
Vanilla PPO	38.4	58.2	44.0	74.7	86.2	62.2
GAT-PPO	37.1	58.6	44.5	75.1	78.8	61.4
SA-PPO	40.4	59.9	51.8	77.4	84.5	63.9
Average connected-component size $\uparrow$

TABLE III: Temporal robustness under different task start times. The best result at each start time is shown in bold.

Fig. 14: Training performance on the Jinan road network.

Intelligent Transportation Systems

Intelligent Transportation Systems apply sensing, communication, AI, and optimization to make transport networks safer and more efficient.

See all 297 papers

Dynamic Mask Enhanced Intelligent Multi-UAV Deployment for Urban Vehicular Networks
Mar 19, 2026Gaoxiang Cao, Wenke Yuan, Yunpeng Hou +3Cooperative MARL V2X Communication
Department of Automation, University of Science and Technology of China, Hefei, China · Institute of Artificial Intelligence, Hefei Comprehensive National Science Center, Hefei, China
TRUAV: Distributed Multi-Agent Reinforcement Learning for Trajectory Planning and Routing Enhancement in UAV-Aided IoT-Enabled VANETs
Jul 26, 2026Muhammad Umar Farooq Qaisar, Lin Zhang, Zhen Chen +3UAV Navigation Decentralized MARL
Hangzhou International Innovation Institute of Beihang University, Hangzhou 311115, China · Hangzhou International Innovation Institute of Beihang University, Hangzhou 311115, China, the School of Automation Science and Electrical Engineering at Beihang University, Beijing 100191, China, and the State Key Laboratory of Intelligent Manufacturing Systems Technology, Beijing 100854, China · School of Intelligent Science and Engineering and the School of Transportation, Southeast University, Nanjing 211189, China +3
Spatio-Temporal Attention Enhanced Multi-Agent DRL for UAV-Assisted Wireless Networks with Limited Communications
Mar 23, 2026Che Chen, Lanhua Li, Shimin Gong +3Spatio-Temporal Attention Multi-Agent Reinforcement Learning
School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen University, Shenzhen 518000, China, and Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 519082, China · Department of Equipment Management and Unmanned Aerial Vehicle Engineering, Air Force Engineering University, Xi’an · School of Computing and Artificial Intelligence, Jiangxi University of Finance and Economics, Nanchang 330032, China +1

Mar 19, 2026J/K move · Enter open · S save

Dynamic Mask Enhanced Intelligent Multi-UAV Deployment for Urban Vehicular Networks

Gaoxiang Cao, Wenke Yuan, Yunpeng Hou, Huasen He+2

Department of Automation, University of Science and Technology of China, Hefei, China · Institute of Artificial Intelligence, Hefei Comprehensive National Science Center, Hefei, China

Cooperative MARL V2X Communication Multi-Robot Systems

Open paper AskPDF arXiv

Vehicular Ad Hoc Networks (VANETs) play a crucial role in realizing vehicle-road collaboration and intelligent transportation. However, urban VANETs often face challenges such as frequent link disconnections and subnet fragmentation, which hinder reliable connectivity. To address these issues, we dynamically deploy multiple Unmanned Aerial Vehicles (UAVs) as communication relays to enhance VANET. A novel Score based Dynamic Action Mask enhanced QMIX algorithm (Q-SDAM) is proposed for multi-UAV deployment, which maximizes vehicle connectivity while minimizing multi-UAV energy consumption. Specifically, we design a score-based dynamic action mask mechanism to guide UAV agents in exploring large action spaces, accelerate the learning process and enhance optimization performance. The practicality of Q-SDAM is validated using real-world datasets. We show that Q-SDAM improves connectivity by 18.2% while reducing energy consumption by 66.6% compared with existing algorithms.

Abstract

Intelligent Transportation Systems

Explore similar work

Dynamic Mask Enhanced Intelligent Multi-UAV Deployment for Urban Vehicular Networks

TRUAV: Distributed Multi-Agent Reinforcement Learning for Trajectory Planning and Routing Enhancement in UAV-Aided IoT-Enabled VANETs

Spatio-Temporal Attention Enhanced Multi-Agent DRL for UAV-Assisted Wireless Networks with Limited Communications