HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments

Yiming Chen; Celimuge Wu; Yangfei Lin; Zhaoyang  Du; Jianhang  Tang; Soufiene Djahel; Lei Zhong

doi:10.1109/INFOCOM59046.2026.11571476

HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments

Yiming Chen
, Celimuge Wu
, Yangfei Lin
, Zhaoyang Du
, Jianhang Tang
, Soufiene Djahel
, Lei Zhong

Centre for Future Transport and Cities

The University of Electro-Communications
GuiZhou University
Toyota Motor Corporation

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding › peer-review

1 Citation (Scopus)

2 Downloads (Pure)

Abstract

Traditional federated learning (FL) approaches face significant challenges when applied to dynamic and heterogeneous Internet of Vehicles (IoV) environments, which are characterized by frequent node mobility, unstable communication links, and highly non-independent and identically distributed (Non-IID) data. In particular, decentralized network topologies exacerbate the difficulty of maintaining model consistency, thereby impairing overall learning performance. To address these challenges, we propose a new hierarchical decentralized federated learning (HDFL) framework. This framework combines the advantages of centralization and decentralization, builds a three-layer collaborative structure, and improves communication flexibility through an asynchronous model exchange mechanism between the edge and the client. Simultaneously, HDFL introduces a local fine-tuning strategy based on knowledge distillation to enhance the generalization ability and stability of the model. Experimental results using an urban traffic simulation platform show that HDFL consistently outperforms representative decentralized FL methods in terms of the achieved accuracy and convergence speed under heterogeneous IoV environments.

Original language	English
Title of host publication	IEEE INFOCOM 2026 - IEEE Conference on Computer Communications
Publisher	IEEE
Pages	(In-Press)
Number of pages	6
ISBN (Electronic)	979-8-3315-4961-9
ISBN (Print)	979-8-3315-4962-6
DOIs	https://doi.org/10.1109/INFOCOM59046.2026.11571476
Publication status	E-pub ahead of print - 29 Jun 2026
Event	INFOCOM 2026 International Workshop on Fusion of Data, Operation, Information, and Communication Technology for Industry 4.0 and Society 5.0 (DOICT-IndSoc) - Tokyo, Japan Duration: 18 May 2026 → 21 May 2026 https://infocom2026.ieee-infocom.org/ieee-infocom-2026-81/pages/international-workshop-fusion-data-operation-information-and

Publication series

Name	Proceedings - IEEE INFOCOM
ISSN (Print)	0743-166X

Workshop

Workshop	INFOCOM 2026 International Workshop on Fusion of Data, Operation, Information, and Communication Technology for Industry 4.0 and Society 5.0 (DOICT-IndSoc)
Country/Territory	Japan
City	Tokyo
Period	18/05/26 → 21/05/26
Internet address	https://infocom2026.ieee-infocom.org/ieee-infocom-2026-81/pages/international-workshop-fusion-data-operation-information-and

Bibliographical note

© Copyright 2026 IEEE - All rights reserved, including rights for text and data mining and training of artificial intelligence and similar technologies.
Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Copyright © and Moral Rights are retained by the author(s) and/ or other copyright owners. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders.

This document is the author’s post-print version, incorporating any revisions agreed during the peer-review process. Some differences between the published version and this version may remain and you are advised to consult the published version if you wish to cite from it.

Keywords

decentralized federated learning
hierarchical federated learning
Internet of Vehicles
knowledge distillation

Access to Document

10.1109/INFOCOM59046.2026.11571476

Djahel2026AAMAccepted author manuscript, 971 KBLicence: CC BY-NC

Cite this

Chen, Y., Wu, C., Lin, Y., Du, Z., Tang, J., Djahel, S., & Zhong, L. (2026). HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments. In IEEE INFOCOM 2026 - IEEE Conference on Computer Communications (pp. (In-Press)). (Proceedings - IEEE INFOCOM). IEEE. Advance online publication. https://doi.org/10.1109/INFOCOM59046.2026.11571476

HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments. / Chen, Yiming; Wu, Celimuge; Lin, Yangfei et al.
IEEE INFOCOM 2026 - IEEE Conference on Computer Communications. IEEE, 2026. p. (In-Press) (Proceedings - IEEE INFOCOM).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding › peer-review

Chen, Y, Wu, C, Lin, Y, Du, Z, Tang, J, Djahel, S & Zhong, L 2026, HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments. in IEEE INFOCOM 2026 - IEEE Conference on Computer Communications. Proceedings - IEEE INFOCOM, IEEE, pp. (In-Press), INFOCOM 2026 International Workshop on Fusion of Data, Operation, Information, and Communication Technology for Industry 4.0 and Society 5.0 (DOICT-IndSoc), Tokyo, Japan, 18/05/26. https://doi.org/10.1109/INFOCOM59046.2026.11571476

@inproceedings{b9b9709ea892408d80d9e127e890b2d5,

title = "HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments",

abstract = "Traditional federated learning (FL) approaches face significant challenges when applied to dynamic and heterogeneous Internet of Vehicles (IoV) environments, which are characterized by frequent node mobility, unstable communication links, and highly non-independent and identically distributed (Non-IID) data. In particular, decentralized network topologies exacerbate the difficulty of maintaining model consistency, thereby impairing overall learning performance. To address these challenges, we propose a new hierarchical decentralized federated learning (HDFL) framework. This framework combines the advantages of centralization and decentralization, builds a three-layer collaborative structure, and improves communication flexibility through an asynchronous model exchange mechanism between the edge and the client. Simultaneously, HDFL introduces a local fine-tuning strategy based on knowledge distillation to enhance the generalization ability and stability of the model. Experimental results using an urban traffic simulation platform show that HDFL consistently outperforms representative decentralized FL methods in terms of the achieved accuracy and convergence speed under heterogeneous IoV environments.",

keywords = "decentralized federated learning, hierarchical federated learning, Internet of Vehicles, knowledge distillation",

author = "Yiming Chen and Celimuge Wu and Yangfei Lin and Zhaoyang Du and Jianhang Tang and Soufiene Djahel and Lei Zhong",

note = "{\textcopyright} Copyright 2026 IEEE - All rights reserved, including rights for text and data mining and training of artificial intelligence and similar technologies. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Copyright {\textcopyright} and Moral Rights are retained by the author(s) and/ or other copyright owners. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders. This document is the author{\textquoteright}s post-print version, incorporating any revisions agreed during the peer-review process. Some differences between the published version and this version may remain and you are advised to consult the published version if you wish to cite from it. ; INFOCOM 2026 International Workshop on Fusion of Data, Operation, Information, and Communication Technology for Industry 4.0 and Society 5.0 (DOICT-IndSoc) ; Conference date: 18-05-2026 Through 21-05-2026",

year = "2026",

month = jun,

day = "29",

doi = "10.1109/INFOCOM59046.2026.11571476",

language = "English",

isbn = "979-8-3315-4962-6",

series = "Proceedings - IEEE INFOCOM",

publisher = "IEEE",

pages = "(In--Press)",

booktitle = "IEEE INFOCOM 2026 - IEEE Conference on Computer Communications",

address = "United States",

url = "https://infocom2026.ieee-infocom.org/ieee-infocom-2026-81/pages/international-workshop-fusion-data-operation-information-and",

}

TY - GEN

T1 - HDFL

T2 - INFOCOM 2026 International Workshop on Fusion of Data, Operation, Information, and Communication Technology for Industry 4.0 and Society 5.0 (DOICT-IndSoc)

AU - Chen, Yiming

AU - Wu, Celimuge

AU - Lin, Yangfei

AU - Du, Zhaoyang

AU - Tang, Jianhang

AU - Djahel, Soufiene

AU - Zhong, Lei

N1 - © Copyright 2026 IEEE - All rights reserved, including rights for text and data mining and training of artificial intelligence and similar technologies. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Copyright © and Moral Rights are retained by the author(s) and/ or other copyright owners. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders. This document is the author’s post-print version, incorporating any revisions agreed during the peer-review process. Some differences between the published version and this version may remain and you are advised to consult the published version if you wish to cite from it.

PY - 2026/6/29

Y1 - 2026/6/29

N2 - Traditional federated learning (FL) approaches face significant challenges when applied to dynamic and heterogeneous Internet of Vehicles (IoV) environments, which are characterized by frequent node mobility, unstable communication links, and highly non-independent and identically distributed (Non-IID) data. In particular, decentralized network topologies exacerbate the difficulty of maintaining model consistency, thereby impairing overall learning performance. To address these challenges, we propose a new hierarchical decentralized federated learning (HDFL) framework. This framework combines the advantages of centralization and decentralization, builds a three-layer collaborative structure, and improves communication flexibility through an asynchronous model exchange mechanism between the edge and the client. Simultaneously, HDFL introduces a local fine-tuning strategy based on knowledge distillation to enhance the generalization ability and stability of the model. Experimental results using an urban traffic simulation platform show that HDFL consistently outperforms representative decentralized FL methods in terms of the achieved accuracy and convergence speed under heterogeneous IoV environments.

AB - Traditional federated learning (FL) approaches face significant challenges when applied to dynamic and heterogeneous Internet of Vehicles (IoV) environments, which are characterized by frequent node mobility, unstable communication links, and highly non-independent and identically distributed (Non-IID) data. In particular, decentralized network topologies exacerbate the difficulty of maintaining model consistency, thereby impairing overall learning performance. To address these challenges, we propose a new hierarchical decentralized federated learning (HDFL) framework. This framework combines the advantages of centralization and decentralization, builds a three-layer collaborative structure, and improves communication flexibility through an asynchronous model exchange mechanism between the edge and the client. Simultaneously, HDFL introduces a local fine-tuning strategy based on knowledge distillation to enhance the generalization ability and stability of the model. Experimental results using an urban traffic simulation platform show that HDFL consistently outperforms representative decentralized FL methods in terms of the achieved accuracy and convergence speed under heterogeneous IoV environments.

KW - decentralized federated learning

KW - hierarchical federated learning

KW - Internet of Vehicles

KW - knowledge distillation

UR - https://www.scopus.com/pages/publications/105044561563

U2 - 10.1109/INFOCOM59046.2026.11571476

DO - 10.1109/INFOCOM59046.2026.11571476

M3 - Conference proceeding

SN - 979-8-3315-4962-6

T3 - Proceedings - IEEE INFOCOM

SP - (In-Press)

BT - IEEE INFOCOM 2026 - IEEE Conference on Computer Communications

PB - IEEE

Y2 - 18 May 2026 through 21 May 2026

ER -

HDFL: A Hierarchical Decentralized Federated Learning Framework for Dynamic and Heterogeneous IoV Environments

Abstract

Publication series

Workshop

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this