Lightweight Semantic Communication System Driven by User Personalization in UAV Networks

WEI Yuxuan; CHEN Xiao; CHEN Qiuyu; JIANG Hao; YANG Zhaohui

doi:10.11999/JEIT260370

Article Contents

Article Navigation > Journal of Electronics & Information Technology > 2026 >

WEI Yuxuan, CHEN Xiao, CHEN Qiuyu, JIANG Hao, YANG Zhaohui. Lightweight Semantic Communication System Driven by User Personalization in UAV Networks[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT260370

Citation:

WEI Yuxuan, CHEN Xiao, CHEN Qiuyu, JIANG Hao, YANG Zhaohui. Lightweight Semantic Communication System Driven by User Personalization in UAV Networks[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT260370

Citation:

PDF( 2082 KB)

Lightweight Semantic Communication System Driven by User Personalization in UAV Networks

doi: 10.11999/JEIT260370 cstr: 32379.14.JEIT260370

1.
School of Artificial Intelligence/School of Future Technology, Nanjing University of Information Science and Technology, Nanjing 210044, China
2.
School of Information Science and Electronic Engineering, Zhejiang University, Zhejiang 310058, China

Funds: General Program of the National Natural Science Foundation of China (62471238), the Young Elite Scientists Sponsorship Program by CAST (2023QNRC001)

Received Date: 2026-03-31
Accepted Date: 2026-05-15
Rev Recd Date: 2026-05-15

Available Online: 2026-06-02

Abstract

Abstract

Objective With the rapid development of the low-altitude economy and 6G intelligent networks, Unmanned Aerial Vehicle (UAV) image communication shows great promise in scenarios such as target reconnaissance, emergency communications, and intelligent inspection. However, constrained by the limited bandwidth, payload capacity, and onboard computational resources of UAVs, conventional pixel-level transmission fails to meet the demands of efficient, low-latency, and intelligent communications. Semantic Communication (SC), which transmits only task-relevant information, offers an effective solution to enhance communication efficiency in such resource-constrained scenarios. However, existing research on UAV image SC faces several challenges. First, fixed network architectures apply unified semantic encoding and transmission strategies for all users, failing to accommodate diverse personalized requirements. Second, new user onboarding typically requires interest pre-training or model fine-tuning, leading to high deployment overhead. Third, the computational complexity of models is generally high. To address these issues, this paper proposes a lightweight personalized UAV SC system, LPUSC, aimed at achieving a balanced trade-off among computation, bandwidth, and personalized demands. The system enables personalized transmission via low-cost semantic index interaction and a lightweight semantic extraction module, without pre-training for new users. Additionally, a dual-branch end-to-end network is designed, where the semantic index transmission network collaborates with the semantic image transmission network trained with a weighted hybrid loss strategy to ensure high-precision and high-quality transmission of personalized semantic images. Methods The proposed LPUSC system adopts a dual-branch architecture to enable accurate task-driven semantic content transmission. First, in the semantic index interaction branch, the lightweight object detection model YOLO11s is employed to perform semantic perception on UAV-captured visual scenes, compressing complex image information into low-dimensional semantic index vectors to reduce transmission redundancy and communication overhead. On this basis, an end-to-end semantic index transmission network is designed to enhance the robustness of semantic index transmission under complex wireless channel conditions. Through the semantic index interaction mechanism, the system is capable of accurately identifying targets of user interest, providing prior guidance for subsequent semantic content extraction. Second, in the semantic image transmission branch, the lightweight yet high-precision MobileSAM model is adopted for semantic region extraction. This branch receives the interest target bounding boxes returned by the semantic index interaction branch as heuristic prompt inputs, enabling pixel-level accurate segmentation and extraction of specific semantic targets. Third, to further enhance the reconstruction quality of semantic images, a weighted hybrid loss function is designed. This loss function integrates Mean Squared Error (MSE), L1 norm, Structural Similarity Index Measure (SSIM), gradient, perceptual, and background suppression losses to jointly optimize the network across pixel-level accuracy, structural preservation, and fine detail restoration. Through the joint constraint of multiple loss terms, the proposed system effectively enhances the reconstruction capability of semantic regions, thereby achieving high-quality semantic image transmission. Results and Discussions Simulation results validate the proposed LPUSC system in terms of semantic extraction and end-to-end transmission. In terms of semantic extraction, three schemes are compared, including YOLO11s-seg, “YOLO11s + SAM”, and “YOLO11s + MobileSAM” (Fig. 4). The results show that the detection-segmentation decoupled architecture achieves superior semantic boundary localization accuracy. Combined with the quantitative analysis in Table 1, the “YOLO11s + MobileSAM” scheme significantly reduces resource consumption while maintaining high extraction accuracy, confirming its suitability for resource-constrained UAV platforms. In terms of end-to-end transmission, the semantic index vector transmission results (Fig. 5) show that the Bit Error Rate (BER) decreases monotonically with increasing Signal-to-Noise Ratio (SNR) across all three channel environments, with rural environments achieving the best performance, followed by suburban and urban environments. The performance differences are primarily attributed to variations in scatterer density and link blockages across environments. The proposed transmission network maintains stable BER under different Doppler frequencies, demonstrating its robustness in dynamic channel conditions. In terms of semantic image transmission, the proposed weighted hybrid loss function demonstrates good training stability (Fig. 6), and LPUSC consistently outperforms the DeepJSCC and “JPEG + LDPC” baselines across the full SNR range (Fig. 7). Specifically, LPUSC achieves SSIM and Peak Signal-to-Noise Ratio (PSNR) gains of 1.3% and 4.8% over DeepJSCC, and 43% and 79.5% over JPEG, respectively. The results indicate that the proposed personalized semantic image transmission network achieves high-quality reconstruction with robustness to channel variations. Conclusions To improve the efficiency and flexibility of UAV image communication, this paper proposes a lightweight personalized SC system called LPUSC. The system employs a dual-branch transmission architecture that integrates a lightweight, high-precision object detection model and a semantic segmentation model, enabling personalized content transmission without interest pre-training. This design meets personalized user requirements while maintaining low computational and communication overhead. Simulation results demonstrate that the LPUSC system achieves stable and reliable semantic index interaction, and significantly outperforms DeepJSCC and JPEG baselines in semantic region reconstruction. The proposed system offers a valuable reference for efficient UAV image SC in 6G low-altitude intelligent networking.
- Semantic communication,
- UAV communication,
- Semantic image transmission,
- User personalization

FullText(HTML)

References(22)

References

[1]	PEI Jiaming, DAI Minghui, AL-DULAIMI A, et al. Task-oriented communication and optimization framework for 6G non-terrestrial networks: Challenges and solutions[J]. IEEE Communications Magazine, 2025, 63(11): 138–144. doi: 10.1109/MCOM.001.2500162.
[2]	LIN Yueshan, FENG Wei, CHEN Yunfei, et al. Latency-constrained resource synergization for mission-oriented 6G non-terrestrial networks[J]. IEEE Internet of Things Journal, 2026. (查阅网上资料,未找到卷期页码信息,请确认). doi: 10.1109/JIOT.2026.3676035.
[3]	CHEN Wanshi, LIU Lingjia, LIU Xiaofeng, et al. Toward standardization of 6G and nextG: Key technologies to enable fundamental enhancements[J]. IEEE Journal on Selected Areas in Communications, 2026, 44: 4333–4365. doi: 10.1109/JSAC.2026.3671338.
[4]	王云涛, 苏洲, 高源, 等. 低空智联网架构、安全与优化关键技术[J]. 电子与信息学报, 2026, 48(3): 889–913. doi: 10.11999/JEIT250947. WANG Yuntao, SU Zhou, GAO Yuan, et al. Key technologies for low-altitude intelligent networks: Architecture, security, and optimization[J]. Journal of Electronics & Information Technology, 2026, 48(3): 889–913. doi: 10.11999/JEIT250947.
[5]	KALEEM Z, ORAKZAI F A, ISHAQ W, et al. Emerging trends in UAVs: From placement, semantic communications to generative AI for mission-critical networks[J]. IEEE Transactions on Consumer Electronics, 2025, 71(3): 7412–7438. doi: 10.1109/TCE.2024.3434971.
[6]	MENG Siqi, WU Shaohua, ZHANG Jiaming, et al. Semantics-empowered space-air-ground-sea integrated network: New paradigm, frameworks, and challenges[J]. IEEE Communications Surveys & Tutorials, 2025, 27(1): 140–183. doi: 10.1109/COMST.2024.3416309.
[7]	钱志鸿, 王义君. 低空经济赋能者: 智能无人机技术体系综述与展望[J]. 电子与信息学报, 2026, 48(1): 1–33. doi: 10.11999/JEIT251246. QIAN Zhihong and WANG Yijun. Intelligent unmanned aerial vehicles for low-altitude economy: A review of the technology framework and future prospects[J]. Journal of Electronics & Information Technology, 2026, 48(1): 1–33. doi: 10.11999/JEIT251246.
[8]	KANG Xu, SONG Bin, GUO Jie, et al. Task-oriented image transmission for scene classification in unmanned aerial systems[J]. IEEE Transactions on Communications, 2022, 70(8): 5181–5192. doi: 10.1109/TCOMM.2022.3182325.
[9]	LIU Sai, ZHANG Zhenjiang, and HAN Guangjie. Semantic task offloading and resource allocation in UAV-assisted mobile edge computing network for disaster rescue[J]. IEEE Transactions on Vehicular Technology, 2026. (查阅网上资料,未找卷期页码信息,请确认). doi: 10.1109/TVT.2026.3665191.
[10]	CHACCOUR C, SAAD W, DEBBAH M, et al. Less data, more knowledge: Building next-generation semantic communication networks[J]. IEEE Communications Surveys & Tutorials, 2025, 27(1): 37–76. doi: 10.1109/COMST.2024.3412852.
[11]	YANG Zhaohui, CHEN Mingzhe, ZHANG Zaoyang, et al. Energy efficient semantic communication over wireless networks with rate splitting[J]. IEEE Journal on Selected Areas in Communications, 2023, 41(5): 1484–1495. doi: 10.1109/JSAC.2023.3240713.
[12]	SONG Xi, ZHOU Fuhui, DING Rui, et al. UAV cognitive semantic communications enabled by knowledge graph for robust object detection[J]. IEEE Transactions on Communications, 2025, 73(8): 6052–6067. doi: 10.1109/TCOMM.2025.3538850.
[13]	王文远, 周明宇, 王朝炜, 等. 面向空天地网络的弹性语义通信[J]. 电子与信息学报, 2025, 47(10): 3646–3657. doi: 10.11999/JEIT250077. WANG Wenyuan, ZHOU Mingyu, WANG Chaowei, et al. Resilient semantic communication for space-air-ground networks[J]. Journal of Electronics & Information Technology, 2025, 47(10): 3646–3657. doi: 10.11999/JEIT250077.
[14]	LIU Shai, YANG Helin, XIE Wanchen, et al. Intelligent semantic communication scheme integrating ISAC for low-altitude intelligent networks[J]. IEEE Transactions on Communications, 2026, 74: 3018–3033. doi: 10.1109/TCOMM.2025.3649924.
[15]	FAN Jian, REN Pengfei, CHEN Jianrui, et al. Diffusion-based semantic-communication-assisted low-altitude intelligent service for IoT[J]. IEEE Internet of Things Journal, 2025, 12(10): 13568–13580. doi: 10.1109/JIOT.2025.3530462.
[16]	HU Han, ZHU Xingwu, ZHOU Fuhui, et al. Resource allocation for multi-modal semantic communication in UAV collaborative networks[J]. IEEE Transactions on Communications, 2025, 73(9): 7599–7616. doi: 10.1109/TCOMM.2025.3552303.
[17]	KANG Jiawen, DU Hongyang, LI Zonghang, et al. Personalized saliency in task-oriented semantic communications: Image transmission and performance analysis[J]. IEEE Journal on Selected Areas in Communications, 2023, 41(1): 186–201. doi: 10.1109/JSAC.2022.3221990.
[18]	DU Baoxia, DU Hongyang, LIU Haifeng, et al. YOLO-based semantic communication with generative AI-aided resource allocation for digital twins construction[J]. IEEE Internet of Things Journal, 2024, 11(5): 7664–7678. doi: 10.1109/JIOT.2023.3317629.
[19]	SI Peiyuan, ZHAO Jun, LAM K Y, et al. UAV-assisted semantic communication with hybrid action reinforcement learning[C]. Proceedings of the GLOBECOM 2023–2023 IEEE Global Communications Conference, Kuala Lumpur, Malaysia, 2023: 3801–3806. doi: 10.1109/GLOBECOM54140.2023.10437643.
[20]	KHUWAJA A A, CHEN Yunfei, ZHAO Nan, et al. A survey of channel modeling for UAV communications[J]. IEEE Communications Surveys & Tutorials, 2018, 20(4): 2804–2821. doi: 10.1109/COMST.2018.2856587.
[21]	ZHANG Chaoning, HAN Dongshen, QIAO Yu, et al. Faster segment anything: Towards lightweight SAM for mobile applications[Z]. arXiv: 2306.14289, 2023. doi: 10.48550/arXiv.2306.14289. (查阅网上资料,不确定本条文献的格式和类型,请确认).
[22]	BOURTSOULATZE E, KURKA D B, and GÜNDÜZ D. Deep joint source-channel coding for wireless image transmission[J]. IEEE Transactions on Cognitive Communications and Networking, 2019, 5(3): 567–579. doi: 10.1109/TCCN.2019.2919300.