Anti-Jamming Maritime Communications for LLM-Assisted Collaborative Perception

LIU Huanhuan; XIAO Liang; LIN Weiqi; ZHANG Pengli; CHEN Haoyu; CHEN Hongyi; TANG Zhongwei

doi:10.12263/DZXB.20250620

您当前的位置：

首页 >

文章列表页 >

Anti-Jamming Maritime Communications for LLM-Assisted Collaborative Perception

PAPERS | 更新时间：2026-06-16

- Anti-Jamming Maritime Communications for LLM-Assisted Collaborative Perception
- ACTA ELECTRONICA SINICA Vol. 54, Issue 3, Pages: 1252-1262(2026)
- 作者机构：
  
  1.厦门大学信息学院，福建厦门 361102
  2.厦门大学人工智能研究院，福建厦门 361105
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(U25A20388);Fundamental Research Funds for the Central Universities(20720250036);National Key Research and Development Program of China(2023YFB3107603)
- DOI：10.12263/DZXB.20250620
  CLC： TN92;
- Received：14 July 2025，
  
  Accepted：28 February 2026，
  
  Published：25 March 2026
- 稿件说明：
移动端阅览
刘欢欢, 肖亮, 林蔚祺, 等. 面向大语言模型协作感知的海域通信抗干扰[J]. 电子学报, 2026, 54(03): 1252-1262.

LIU Huanhuan, XIAO Liang, LIN Weiqi, et al. Anti-Jamming Maritime Communications for LLM-Assisted Collaborative Perception[J]. Acta Electronica Sinica, 2026, 54(03): 1252-1262.
刘欢欢, 肖亮, 林蔚祺, 等. 面向大语言模型协作感知的海域通信抗干扰[J]. 电子学报, 2026, 54(03): 1252-1262. DOI：10.12263/DZXB.20250620

LIU Huanhuan, XIAO Liang, LIN Weiqi, et al. Anti-Jamming Maritime Communications for LLM-Assisted Collaborative Perception[J]. Acta Electronica Sinica, 2026, 54(03): 1252-1262. DOI：10.12263/DZXB.20250620

摘要

海域恶劣无线信道和干扰攻击增加了感知数据和反馈信息的传输难度，进而降低了协作感知性能。为支撑基于大语言模型的协作感知和目标识别任务，海域通信需为文本、图像、视频和点云等多模态数据提供高可靠传输，满足多样化服务质量需求。为此，提出面向大语言模型协作感知的海域通信抗干扰技术，除了数据量、信道增益和历史性能外，进一步利用大语言模型根据接收到的多模态数据和文本提示词推断生成的通信环境，以及频谱感知获取的干扰特征，联合优化传输功率、信道和模型选择，在恶劣信道下防御干扰攻击和通信干扰，支撑基于多模态数据的协作感知。针对恶劣海域信道导致的反馈延迟或丢失问题，设计反馈恢复机制，提升多模态数据传输质量。建立海域通信抗干扰博弈模型，刻画海域终端和干扰机间的交互机理，分析纳什均衡点存在条件并给出通感性能界限，揭示模态数量和信道状态等参数对性能的影响。基于WaterScenes数据集和LLaVA等大语言模型的仿真结果表明，在防御基于Q学习的智能干扰机时，所提方案相较于对比方案可提升13.6%的感知精度，降低66.2%的通信能耗和21.7%的时延。

Abstract

Harsh maritime wireless channels and jamming attacks increase the transmission difficulty of sensing data and feedback information

thereby degrading collaborative perception performance. To support large language model (LLM)-assisted collaborative perception and object detection

maritime communications must provide highly reliable transmission for multi-modal data such as text

images

videos and point clouds to meet diverse quality-of-service requirements. In this paper

we propose an anti-jamming maritime communications scheme for LLM-assisted collaborative perception. Besides the data size

channel gain

and historical performance

the communication environment extracted by LLM based on received multi-modal data and prompt

as well as jamming features obtained via spectrum sensing

are further used to optimize the transmit power

channel and LLM selection for transmitting the multi-modal data to support collaborative perception against jamming and interference under harsh channel conditions. A feedback recovery mechanism is designed to address delayed feedback or loss caused by harsh maritime channels and improve the reliability of multi-modal data transmission. The interaction between maritime terminals and the jammer is formulated as the maritime anti-jamming game and the upper bound in terms of communication and perception performance is provided based on the Nash equilibrium to show the impact of the number of modalities and channel states. Simulation results based on the WaterScenes dataset and LLMs such as LLaVA show the performance gain of our proposed scheme with 13.6% higher perception accuracy

66.2% lower communication energy consumption and 21.7% less latency over benchmarks against the Q-learning-based smart jammer.

关键词

Keywords

references

Qiao Yuanyuan , Yin Jiaxin , Wang Wei , et al . Survey of deep learning for autonomous surface vehicles in marine environments [J ] . IEEE Transactions on Intelligent Transportation Systems , 2023 , 24 ( 4 ): 3678 - 3701 . DOI: 10.1109/tits.2023.3235911 http://dx.doi.org/10.1109/tits.2023.3235911

Wu Shengqiong , Fei Hao , Qu Leigang , et al . NExT-GPT: Any-to-any multimodal LLM [C ] // International Conference on Machine Learning . New York : ACM , 2024 : 1 - 32 .

张海君 , 陈安琪 , 李亚博 , 等 . 6G移动网络关键技术 [J ] . 通信学报 , 2022 , 43 ( 7 ): 189 - 202 .

Zhang Haijun , Chen Anqi , Li Yabo , et al . Key technologies of 6G mobile network [J ] . Journal on Communications , 2022 , 43 ( 7 ): 189 - 202 . (in Chinese)

Yang Helin , Lin Kailong , Xiao Liang , et al . Energy harvesting UAV-RIS-assisted maritime communications based on deep reinforcement learning against jamming [J ] . IEEE Transactions on Wireless Communications , 2024 , 23 ( 8 ): 9854 - 9868 . DOI: 10.1109/twc.2024.3367034 http://dx.doi.org/10.1109/twc.2024.3367034

Liu Chuhuan , Zhang Yi , Niu Guohang , et al . Towards reinforcement learning in UAV relay for anti-jamming maritime communications [J ] . Digital Communications and Networks , 2023 , 9 ( 6 ): 1477 - 1485 . DOI: 10.1016/j.dcan.2022.08.009 http://dx.doi.org/10.1016/j.dcan.2022.08.009

Li Xiangchen , Chen Jienan , Ling Xiang , et al . Deep reinforcement learning-based anti-jamming algorithm using dual action network [J ] . IEEE Transactions on Wireless Communications , 2023 , 22 ( 7 ): 4625 - 4637 . DOI: 10.1109/twc.2022.3227575 http://dx.doi.org/10.1109/twc.2022.3227575

Janiar S B , Wang Ping . Intelligent anti-jamming based on deep reinforcement learning and transfer learning [J ] . IEEE Transactions on Vehicular Technology , 2024 , 73 ( 6 ): 8825 - 8834 . DOI: 10.1109/tvt.2024.3359426 http://dx.doi.org/10.1109/tvt.2024.3359426

Li Yangyang , Xu Yuhua , Li Guoxin , et al . Dynamic spectrum anti-jamming access with fast convergence: A labeled deep reinforcement learning approach [J ] . IEEE Transactions on Information Forensics and Security , 2023 , 18 : 5447 - 5458 . DOI: 10.1109/tifs.2023.3307950 http://dx.doi.org/10.1109/tifs.2023.3307950

Yao Shanliang , Guan Runwei , Wu Zhaodong , et al . WaterScenes: A multi-task 4D radar-camera fusion dataset and benchmarks for autonomous driving on water surfaces [J ] . IEEE Transactions on Intelligent Transportation Systems , 2024 , 25 ( 11 ): 16584 - 16598 . DOI: 10.1109/tits.2024.3415772 http://dx.doi.org/10.1109/tits.2024.3415772

Liu Haotian , Li Chunyuan , Li Yuheng , et al . Improved baselines with visual instruction tuning [C ] // 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2024 : 26286 - 26296 . DOI: 10.1109/cvpr52733.2024.02484 http://dx.doi.org/10.1109/cvpr52733.2024.02484

Wu Lei , Wang Wei , Ji Zengshuan , et al . UAV-assisted maritime legitimate surveillance: Joint trajectory design and power allocation [J ] . IEEE Transactions on Vehicular Technology , 2023 , 72 ( 10 ): 13701 - 13705 . DOI: 10.1109/tvt.2023.3276323 http://dx.doi.org/10.1109/tvt.2023.3276323

苏新 , 孟蕾蕾 , 周一青 , 等 . 基于深度强化学习的海洋移动边缘计算卸载方法 [J ] . 通信学报 , 2022 , 43 ( 10 ): 133 - 145 . DOI: 10.11959/j.issn.1000-436x.2022197 http://dx.doi.org/10.11959/j.issn.1000-436x.2022197

Su Xin , Meng Leilei , Zhou Yiqing , et al . Maritime mobile edge computing offloading method based on deep reinforcement learning [J ] . Journal on Communications , 2022 , 43 ( 10 ): 133 - 145 . (in Chinese) . DOI: 10.11959/j.issn.1000-436x.2022197 http://dx.doi.org/10.11959/j.issn.1000-436x.2022197

Yang Tingting , Jiang Zhi , Sun Ruijin , et al . Maritime search and rescue based on group mobile computing for unmanned aerial vehicles and unmanned surface vehicles [J ] . IEEE Transactions on Industrial Informatics , 2020 , 16 ( 12 ): 7700 - 7708 . DOI: 10.1109/tii.2020.2974047 http://dx.doi.org/10.1109/tii.2020.2974047

周权 , 牛英滔 . 基于相似性样本生成的深度强化学习快速抗干扰算法 [J ] . 通信学报 , 2024 , 45 ( 7 ): 117 - 126 .

Zhou Quan , Niu Yingtao . Fast deep reinforcement learning anti-jamming algorithm based on similar sample generation [J ] . Journal on Communications , 2024 , 45 ( 7 ): 117 - 126 . (in Chinese)

Yang Tingting , Li Jiabo , Feng Hailong , et al . A novel transmission scheduling based on deep reinforcement learning in software-defined maritime communication networks [J ] . IEEE Transactions on Cognitive Communications and Networking , 2019 , 5 ( 4 ): 1155 - 1166 . DOI: 10.1109/tccn.2019.2939813 http://dx.doi.org/10.1109/tccn.2019.2939813

辜方林 , 黄育侦 , 赵莺 , 等 . 基于单载波频域均衡的直接序列扩频自适应抗干扰技术 [J ] . 通信学报 , 2022 , 43 ( 11 ): 26 - 34 .

Gu Fanglin , Huang Yuzhen , Zhao Ying , et al . Adaptive anti-jamming technologies based on direct sequence spread spectrum and single carrier frequency domain equalization [J ] . Journal on Communications , 2022 , 43 ( 11 ): 26 - 34 . (in Chinese)

Shang Yuzhang , Cai Mu , Xu Bingxin , et al . LLaVA-PruMerge: Adaptive token reduction for efficient large multimodal models [C ] // IEEE Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2025 : 22857 - 22867 . DOI: 10.1109/iccv51701.2025.02122 http://dx.doi.org/10.1109/iccv51701.2025.02122

Lin Bin , Tang Zhenyu , Ye Yang , et al . MoE-LLaVA: Mixture of experts for large vision-language models [J ] . IEEE Transactions on Multimedia , 2026 : 1 - 14 . DOI: 10.1109/tmm.2026.3654458 http://dx.doi.org/10.1109/tmm.2026.3654458

Zhu Yichen , Zhu Minjie , Liu Ning , et al . LLaVA-phi: Efficient multi-modal assistant with small language model [C ] // Proceedings of the 1st International Workshop on Efficient Multimedia Computing under Limited . New York : ACM , 2024 : 18 - 22 . DOI: 10.1145/3688863.3689575 http://dx.doi.org/10.1145/3688863.3689575

Zhou Baichuan , Hu Ying , Weng Xi , et al . TinyLLaVA: A framework of small-scale large multimodal models [PP/OL ] . V1. arXiv ( 2024-02-22 )[ 2025-07-14 ] . https://doi.org/10.48550/arXiv.2402.14289 https://doi.org/10.48550/arXiv.2402.14289 .

张青龙 , 韩锐 , 刘驰 . 云边协同大模型块粒度重训方法 [J ] . 电子学报 , 2025 , 53 ( 2 ): 287 - 300 .

Zhang Qinglong , Han Rui , Liu Chi . Cloud-edge collaborative retraining of foundation models at the block granularity [J ] . Acta Electronica Sinica , 2025 , 53 ( 2 ): 287 - 300 . (in Chinese)

Wang Xiaowei , Wang Xue , Gao Jie , et al . A denoising method for cable partial discharge signals based on image information entropy and multivariate variational mode decomposition [J ] . IEEE Transactions on Instrumentation and Measurement , 2024 , 73 : 3500415 . DOI: 10.1109/tim.2023.3334357 http://dx.doi.org/10.1109/tim.2023.3334357

Frantar E , Alistarh D . SparseGPT: Massive language models can be accurately pruned in one-shot [C ] // International Conference on Machine Learning . New York : ACM , 2023 : 10323 - 10337 .

Sheng Ying , Zheng Lianmin , Yuan Binhang , et al . Flexgen: High-throughput generative inference of large language models with a single GPU [C ] // International Conference on Machine Learning . New York : ACM , 2023 : 31094 - 31116 .

Khanmohammadi R , Miahi E , Mardikoraem M , et al . Calibrating LLM confidence by probing perturbed representation stability [C ] // Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing . Stroudsburg : ACL , 2025 : 10459 - 10525 . DOI: 10.18653/v1/2025.emnlp-main.530 http://dx.doi.org/10.18653/v1/2025.emnlp-main.530

Shankar S , Zamfirescu-Pereira J D , Hartmann B , et al . Who validates the validators aligning LLM-assisted evaluation of LLM outputs with human preferences [C ] // Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology . New York : ACM , 2024 : 3676450 . DOI: 10.1145/3654777.3676450 http://dx.doi.org/10.1145/3654777.3676450

Huang Guangbin . Learning capability and storage capacity of two-hidden-layer feedforward networks [J ] . IEEE Transactions on Neural Networks , 2003 , 14 ( 2 ): 274 - 281 . DOI: 10.1109/tnn.2003.809401 http://dx.doi.org/10.1109/tnn.2003.809401

Nikookar S , Sakharkar P , Somasunder S , et al . Cooperative route planning framework for multiple distributed assets in maritime applications [C ] // Proceedings of the 2022 International Conference on Management of Data . New York : ACM , 2022 : 1518 - 1527 . DOI: 10.1145/3514221.3526131 http://dx.doi.org/10.1145/3514221.3526131

Wang Yuzhen , Zhou Ting , Feng Wei , et al . When maritime wireless communications meet evaporation ducts: A three-ray path loss modeling perspective [J ] . IEEE Wireless Communications Letters , 2025 , 14 ( 2 ): 470 - 474 . DOI: 10.1109/lwc.2024.3509740 http://dx.doi.org/10.1109/lwc.2024.3509740

Zeng Hui , Su Zhou , Xu Qichao , et al . USV fleet-assisted collaborative computation offloading for smart maritime services: An energy-efficient design [J ] . IEEE Transactions on Vehicular Technology , 2024 , 73 ( 10 ): 14718 - 14733 . DOI: 10.1109/tvt.2024.3359310 http://dx.doi.org/10.1109/tvt.2024.3359310

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Identification and Migration of Silent Security Patches in Blockchain Systems via Information Fusion

Multimodal Smart Contract Vulnerability Detection Enhanced by Large Language Model Semantics

Testing Text Input Components of Web Applications Based on Large Language Models

Fusing Large Language Models with Cross-Domain Graphs for Multimodal Emotion Recognition in Dialogue

Selective Learning for Incomplete Medical Data

Related Author

HAO Sicheng

WEI Guipeng

XIAO Yuming

NAN Yuhong

ZHENG Peilin

ZHENG Zibin

KANG Haiyan

FAN Ruiyang

Related Institution

Guangdong Engineering Technology Research Center of Blockchain

School of Software Engineering, Sun Yat-sen University

College of Computer Science, Beijing Information Science and Technology University

College of Information Science and Engineering, Henan University of Technology

Hubei Province Project of Key Research Institute of Humanities and Social Sciences at Universities

⁰