DFUN-KDF: A knowledge distillation-based decentralized federated learning framework for uav network optimization

As the use of Unmanned Aerial Vehicles (UAVs) expands across various fields, there is growing interest in leveraging Federated Learning (FL) to enhance the efficiency of UAV networks. However, existing methods face several challenges, including high energy consumption, the need for network homogeneity, and vulnerability to single points of failure.

                                                              

They published their study on 11 June 2025, in Communications in Transportation Research.

 

"We propose the DFUN-KDF framework, which utilizes federated knowledge distillation to enable UAVs to extract information from embedded data received from other nodes and update their local models. This approach reduces the data transmission to approximately 1% of the original amount and addresses the challenges of model heterogeneity in UAV networks. Additionally, the framework’s filtering mechanism eliminates biased embeddings, effectively mitigating interference from malicious UAVs or communication failures, thereby ensuring the stability and reliability of the system," says Wenyuan Yang, a researcher at the School of Cyber Science and Technology at Shenzhen Campus of Sun Yat-sen University.

 

Ensuring System Stability with Advanced Filtering Mechanism

A standout feature of the DFUN-KDF framework is its built-in filtering mechanism, designed to identify and remove biased or faulty data that could arise from malicious UAVs or communication failures. This mechanism ensures that the system remains stable and reliable, even in the face of external disruptions or data inconsistencies.

“Our filtering method enhances the robustness of the UAV network by eliminating erroneous or harmful data that could otherwise compromise the learning process,” explains Yang. “This makes DFUN-KDF particularly effective in environments where malicious nodes or network faults are a concern.”

 

Robust Performance in Static and Dynamic Environments

 

DFUN-KDF has demonstrated strong performance in both static and dynamic network environments through a series of rigorous experiments. The framework showed remarkable robustness in the face of node failures, communication interruptions, and malicious attacks, making it suitable for large-scale UAV deployments in real-world scenarios. Compared to traditional federated learning methods, DFUN-KDF offers significant advantages in terms of reducing communication energy consumption and improving adaptability, making it an ideal solution for diverse UAV network applications.

“The ability to handle node crashes and malicious attacks, while maintaining high accuracy, positions DFUN-KDF as a promising solution for UAV networks that require both security and efficiency,” notes Gege Jiang, a co-researcher in the study.

 

Future Research Directions

 

Looking ahead, the researchers plan to expand the coverage of the DFUN-KDF algorithm, further reducing computation time and improving the accuracy of network topology. Additionally, increasing the coverage of Federated Communication Data (FCD) will further enhance the algorithm’s effectiveness, making it more adaptable to complex transportation networks and varied traffic scenarios.

“We are excited about the potential of DFUN-KDF to transform UAV network operations,” says Yang. “With continued research and optimization, we believe that this framework can drive significant advancements in UAV intelligent systems, supporting a wide range of applications in urban management, logistics, and beyond.”

 

The above research is published in Communications in Transportation Research (COMMTR), which is a fully open access journal co-published by Tsinghua University Press and Elsevier. COMMTR publishes peer-reviewed high-quality research representing important advances of significance to emerging transport systems. COMMTR is also among the first transportation journals to make the Replication Package mandatory to facilitate researchers, practitioners, and the general public in understanding and advancing existing knowledge. At its discretion, Tsinghua University Press will pay the open access fee for all published papers from 2024 to 2025.

 

About Communications in Transportation Research

Communications in Transportation Research was launched in 2021, with academic support provided by Tsinghua University and China Intelligent Transportation Systems Association. The Editors-in-Chief are Professor Xiaobo Qu, a member of the Academia Europaea from Tsinghua University and Professor Shuai’an Wang from Hong Kong Polytechnic University. The journal mainly publishes high-quality, original research and review articles that are of significant importance to emerging transportation systems,aiming to serve as an international platform for showcasing and exchanging innovative achievements in transportation and related fields, fostering academic exchange and development between China and the global community.

It has been indexed in SCIE, SSCI, Ei Compendex, Scopus, DOAJ, TRID and other databases.

It was selected as Q1 Top Journal in the Engineering and Technology category of the Chinese Academy of Sciences (CAS) Journal Ranking List. In 2022, it was selected as a High-Starting-Point new journal project of the “China Science and Technology Journal Excellence Action Plan”. In 2024, it was selected as the Support the Development Project of “High-Level International Scientific and Technological Journals”. The same year, it was also chosen as an English Journal Tier Project of the “China Science and Technology Journal Excellence Action Plan PhaseⅡ”. The 2024 IF is 14.5, ranking in the Top1 (1/61, Q1) among all journals in "TRANSPORTATION" category. Tsinghua University Press will cover the open access fee for all published papers in 2025.

---

---