Federated Learning with Differential Privacy for Resilient Vehicular Cyber Physical Systems
Abstract
Vehicular cyber physical systems (VCPS) will play a vital role in the quest to develop intelligent transportation systems (ITS) and smart cities around the world. Consequently, researchers in academia, industry and government continue to leverage on emerging technologies like software defined networking (SDN), blockchain, cloud computing and machine learning (ML) to improve the overall efficiency of these intelligent systems. Recently, mobile edge computing (MEC) has been used to enhance content caching and efficient resource allocation and therefore advance the development of data-intensive and delay-constrained applications that improve the driving experience in VCPS. Security and privacy concerns that endanger the safety of lives and infrastructure necessitate the need to use federated learning (FL), a distributed ML algorithm that employs learning at the edge to ensure that data remains at the different vehicles and thus enhance greater efficiency. In this paper therefore, we propose the use of FL, together with differential privacy to improve the resiliency of VCPS to adversarial attacks in connected vehicles.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The resilience of FL to adversarial attacks in VCPS is enhanced through integration with other PPML techniques such as DP [104] and blockchain-leveraged methods [9].
“Furthermore, the resilience of FL to adversarial attacks in VCPS is enhanced through integration with other PPML techniques such as DP [104] and blockchain-leveraged methods [9].”
From Privacy-preserving Machine Learning in Internet of Vehicle Applications: Fundamentals, Recent Advances, and Future Direction (2025) · cites this paper as [104]
There are hybrid frameworks combining blockchain, FL, and DP to enhance data security and resilience in VCPS [104].
“There are hybrid frameworks combining blockchain, FL, and DP to enhance data security and resilience in VCPS [104].”
From Privacy-preserving Machine Learning in Internet of Vehicle Applications: Fundamentals, Recent Advances, and Future Direction (2025) · cites this paper as [104]
The system integrates FL with DP and applies layer-wise relevance propagation (LRP) to regulate perturbation values.
“The system integrates FL with DP and applies layer-wise relevance propagation (LRP) to regulate perturbation values.”
From Privacy-preserving Machine Learning in Internet of Vehicle Applications: Fundamentals, Recent Advances, and Future Direction (2025) · cites this paper as [104]
Authors in [104] propose a privacy-preserving VCPS to further defend against adversarial attacks exploiting vehicle speed and location.
“To further defend against adversarial attacks exploiting vehicle speed and location, authors in [104] propose a privacy-preserving VCPS.”
From Privacy-preserving Machine Learning in Internet of Vehicle Applications: Fundamentals, Recent Advances, and Future Direction (2025) · cites this paper as [104]
Cites
Federated Learning With Differential Privacy: Algorithms and Performance Analysis
Wei, K., Li, J., Ding, M., Ma, C., Yang, H. H., Farokhi, F., Jin, S., Quek, T. Q. S., & Vincent Poor, H. (2020) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2020.2988575 · Cited by 2,400 · see citing works
Federated Learning: A Survey on Enabling Technologies, Protocols, and Applications
Aledhari, M., Razzak, R., Parizi, R. M., & Saeed, F. (2020) · IEEE Access · doi:10.1109/access.2020.3013541 · Cited by 733 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Ferrag, M. A., Friha, O., Maglaras, L., Janicke, H., & Shu, L. (2021) · IEEE Access · doi:10.1109/access.2021.3118642 · Cited by 342 · see citing works
Islam, N., & Zulkernine, M. (2025) · arXiv · doi:10.48550/arxiv.2503.01089 · Cited by 1 · see citing works
“Additionally, managing cumulative privacy loss over multiple queries or iterations (the privacy budget) is crucial to maintaining robust privacy guarantees over time.”
- Authors
- Olowononi, F. O., Rawat, D. B., & Liu, C.
- Affiliations
- Howard University (US)
- Venue
- 2021 IEEE 18th Annual Consumer Communications & Networking Conference (CCNC)
- Publisher
- IEEE
- Published
- 2021-01-09
- Pages
- 1-5
- ISSN
- 2331-9852 (ISSN-L)
- OpenAlex
- W3139227553
- Cited by
- 47
- References
- 2
- Citations by year
- 2026: 4 · 2025: 8 · 2024: 12 · 2023: 13 · 2022: 7 · 2021: 3
- FWCI
- 4.04
- Type
- conference paper
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE Consumer Communications and Networking Conference (publishedVersion)
- Funders
- U.S. NSF (CNS-1650831,CCF-0939370,HRD-1828811)
- Howard University (W911NF-20-2-0277)
- U.S. Army Research Laboratory
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Vehicular Ad Hoc Networks (VANETs)
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Keywords
- federated learning, differential privacy, vehicular cyber-physical systems, connected vehicles, adversarial attacks, attack resilience, mobile edge computing, intelligent transportation systems, privacy preservation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
