DNO-RL: A Reinforcement-Learning-Based Approach to Dynamic Noise Optimization for Differential Privacy
Abstract
1. Introduction
- (1)
- Context-aware noise dynamic regulation mechanism: Through reinforcement learning, the intelligent body senses multi-dimensional states, such as vehicle density, positioning accuracy demand, and remaining privacy budget, in real time and dynamically optimizes the noise parameters (e.g., Laplacian noise scale) to realize the precise regulation of noise and adapt to the privacy protection demand under different environmental conditions.
- (2)
- Compliance action space modeling under multi-regulation constraints: An action space model based on regional privacy regulation constraints is designed to ensure that the noise policy can simultaneously comply with the compliance requirements of heterogeneous privacy regulations, such as GDPR and CCPA, while meeting privacy protection needs.
- (3)
- Lightweight online decision-making architecture: This combines a deep Q-network (DQN) with privacy budget constraints to realize low-latency noise-allocation policies for highly dynamic and real-time changing vehicular environments. This approach fills the research gap in dynamic differential privacy optimization for cross-border vehicle services and provides a scalable privacy-preserving solution for cross-border telematics applications that can adapt to the privacy regulatory requirements of different jurisdictions.
2. Related Work
2.1. Differential Privacy and Its Extensions
2.2. State-of-the-Art Research on Privacy Protection in Cross-Border Scenarios
2.3. Application of Reinforcement Learning to Privacy Budget Optimization
3. Related Definitions
3.1. Differential Privacy
3.2. Reinforcement Learning
4. Methodology
4.1. Problem Definition and Mathematical Modeling
4.2. Dynamic Noise Optimization Algorithm Framework
4.3. DNO-RL Algorithm Flow
Algorithm 1: Dynamic noise optimization reinforcement learning (DNO-RL) |
5. Experiments
5.1. Experimental Environment
5.1.1. Performance Evaluation Experiment
5.1.2. Noise Validation Experiment
5.1.3. Algorithm Comparison Experiments
5.2. Introduction of the Dataset
5.3. Comparison Methods and Evaluation Metrics
5.3.1. Baseline Methods
5.3.2. Evaluation Metrics
5.3.3. Experimental Parameterization
5.4. Analysis of Experimental Results
5.4.1. Analysis of Privacy–Utility Trade-Offs Under Different Privacy Budgets
5.4.2. Noise Scale Analysis for Different Areas and Densities
5.4.3. Performance Comparison of Different Algorithms in Cross-Border Scenarios
5.4.4. Comparison Experiments with Different Algorithms
6. Conclusions
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
Abbreviations
CCPA | California Consumer Privacy Act |
DQN | Deep Q-Network |
DDQN | Double Deep Q-Network |
DNO-RL | Dynamic Noise Optimization via Reinforcement Learning |
DP | Differential Privacy |
GDPR | General Data Protection Regulation |
GPS | Global Positioning System |
ITS | Intelligent Transportation Systems |
LDP | Local Differential Privacy |
MDP | Markov Decision Process |
R-DP | Rule-Based Dynamic Differential Privacy |
RL | Reinforcement Learning |
S-DP | Static Differential Privacy |
SGD | Stochastic Gradient Descent |
VANETs | Vehicular Ad Hoc Networks |
References
- Kousaridas, A.; Fallgren, M.; Fischer, E.; Moscatelli, F.; Vilalta, R.; Mühleisen, M.; Barmpounakis, S.; Vilajosana, X.; Euler, S.; Tossou, B.; et al. 5g vehicle-to-everything services in cross-border environments: Standardization and challenges. IEEE Commun. Stand. Mag. 2021, 5, 22–30. [Google Scholar] [CrossRef]
- Chiha, A.; Vannieuwenborg, F.; Denis, B.; Colle, D.; Verbrugge, S. Cooperative, connected and automated mobility (ccam) services provisioning in cross-border settings: Techno-economic analysis in the light of technical challenges. Transp. Policy 2023, 140, 68–84. [Google Scholar] [CrossRef]
- Tahir, S.; Tahir, W. Legal challenges in cross-border data transfers: Balancing security and privacy in a globalized world. Mayo RC J. Commun. Sustain. World 2024, 1, 1. [Google Scholar]
- Brambilla, M.; Nicoli, M.; Soatti, G.; Deflorio, F. Augmenting vehicle localization by cooperative sensing of the driving environment: Insight on data association in urban traffic scenarios. IEEE Trans. Intell. Transp. Syst. 2020, 21, 1646–1663. [Google Scholar] [CrossRef]
- Chen, C.; Hu, X.; Li, Y.; Tang, Q. Optimization of privacy budget allocation in differential privacy-based public transit trajectory data publishing for smart mobility applications. IEEE Trans. Intell. Transp. Syst. 2023, 24, 15158–15168. [Google Scholar] [CrossRef]
- Song, S.; Chaudhuri, K.; Sarwate, A.D. Stochastic gradient descent with differentially private updates. In Proceedings of the 2013 IEEE Global Conference on Signal and Information Processing, Austin, TX, USA, 3–5 December 2013; pp. 245–248. [Google Scholar]
- Arif, M.; Chen, J.; Wang, G.; Geman, O.; Balas, V.E. Privacy preserving and data publication for vehicular trajectories with differential privacy. Measurement 2021, 173, 108675. [Google Scholar] [CrossRef]
- Yang, M.; Wu, Y.; Chen, Y. A k-anonymity optimization algorithm under attack model. In Proceedings of the 2022 IEEE International Conferences on Internet of Things (iThings) and IEEE Green Computing Communications (GreenCom) and IEEE Cyber, Physical Social Computing (CPSCom) and IEEE Smart Data (SmartData) and IEEE Congress on Cybermatics (Cybermatics), Espoo, Finland, 22–25 August 2022; pp. 357–362. [Google Scholar]
- Florin, R.; Olariu, S. Real-time traffic density estimation: Putting on-coming traffic to work. IEEE Trans. Intell. Transp. Syst. 2023, 24, 1374–1383. [Google Scholar] [CrossRef]
- Wong, R.Y.; Chong, A.; Aspegren, R.C. Privacy legislation as business risks: How gdpr and ccpa are represented in technology companies’ investment risk disclosures. Proc. ACM Hum.-Comput. Interact. 2023, 7, 1–26. [Google Scholar] [CrossRef]
- Al-Hussaeni, K.; Fung, B.C.M.; Iqbal, F.; Dagher, G.G.; Park, E.G. Safepath: Differentially-private publishing of passenger trajectories in transportation systems. Comput. Netw. 2018, 143, 126–139. [Google Scholar] [CrossRef]
- Shen, Y.; Shepherd, C.; Ahmed, C.M.; Shen, S.; Wu, X.; Ke, W.; Yu, S. Game-theoretic analytics for privacy preservation in internet of things networks: A survey. Eng. Appl. Artif. Intell. 2024, 133, 108449. [Google Scholar] [CrossRef]
- Seeman, J.; Susser, D. Between privacy and utility: On differential privacy in theory and practice. ACM J. Responsib. Comput. 2024, 1, 1–18. [Google Scholar] [CrossRef]
- Mesbah, W. Dp with auxiliary information: Gaussian mechanism versus laplacian mechanism. IEEE Open J. Commun. Soc. 2025, 6, 143–153. [Google Scholar] [CrossRef]
- Sharma, J.; Kim, D.; Lee, A.; Seo, D. On differential privacy-based framework for enhancing user data privacy in mobile edge computing environment. IEEE Access 2021, 9, 38107–38118. [Google Scholar] [CrossRef]
- Erlingsson, Ú.; Pihur, V.; Korolova, A. Rappor: Randomized aggregatable privacy-preserving ordinal response. In Proceedings of the 2014 ACM SIGSAC Conference on Computer and Communications Security, Scottsdale, AZ, USA, 3–7 November 2014; pp. 1054–1067. [Google Scholar]
- Xu, C.; Luo, L.; Ding, Y.; Zhao, G.; Yu, S. Personalized location privacy protection for location-based services in vehicular networks. IEEE Wirel. Commun. Lett. 2020, 9, 1633–1637. [Google Scholar] [CrossRef]
- Dwork, C.; Roth, A. The algorithmic foundations of differential privacy. Found. Trends Theor. Comput. Sci. 2014, 9, 211–407. [Google Scholar] [CrossRef]
- Wang, H.; Li, Y.; Gao, C.; Wang, G.; Tao, X.; Jin, D. Anonymization and de-anonymization of mobility trajectories: Dissecting the gaps between theory and practice. IEEE Trans. Mob. Comput. 2021, 20, 796–815. [Google Scholar] [CrossRef]
- Wang, Y.; Yang, J.; Zhang, J. Differential privacy for weighted network based on probability model. IEEE Access 2020, 8, 80792–80800. [Google Scholar] [CrossRef]
- Wu, Y.; Liu, C.; Xie, Z. A privacy preserving incentive mechanism for intelligent transportation systems. In Proceedings of the 2024 Cross Strait Radio Science and Wireless Technology Conference (CSRSWTC), Taipei, Taiwan, 11–14 November 2024; pp. 1–3. [Google Scholar]
- Ghane, S.; Jolfaei, A.; Kulik, L.; Ramamohanarao, K.; Puthal, D. Preserving privacy in the internet of connected vehicles. IEEE Trans. Intell. Transp. Syst. 2021, 22, 5018–5027. [Google Scholar] [CrossRef]
- Xu, J.; Liu, L.; Zhang, R.; Xie, J.; Duan, Q.; Shi, L. Ifts: A location privacy protection method based on initial and final trajectory segments. IEEE Access 2021, 9, 18112–18122. [Google Scholar] [CrossRef]
- Chen, H.; Li, S.; Zhang, Z. A differential privacy based (κ-ψ)-anonymity method for trajectory data publishing. Comput. Mater. Contin. 2020, 65, 2665–2685. [Google Scholar] [CrossRef]
- Min, M.; Zhu, H.; Li, S.; Zhang, H.; Xiao, L.; Pan, M.; Han, Z. Semantic adaptive geo-indistinguishability for location privacy protection in mobile networks. IEEE Trans. Veh. Technol. 2024, 73, 9193–9198. [Google Scholar] [CrossRef]
- Han, Y.; Wang, M.; Leclercq, L. Leveraging reinforcement learning for dynamic traffic control: A survey and challenges for field implementation. Commun. Transp. Res. 2023, 3, 100104. [Google Scholar] [CrossRef]
- Sutton, R.S.; Barto, A.G. Reinforcement learning: An introduction. IEEE Trans. Neural Netw. 1998, 9, 1054. [Google Scholar] [CrossRef]
- Li, J.; Zhang, F.; Guo, Y.; Li, S.; Wu, G.; Li, D.; Zhu, H. A privacy-preserving online deep learning algorithm based on differential privacy. In Proceedings of the 2023 26th International Conference on Computer Supported Cooperative Work in Design (CSCWD), Rio de Janeiro, Brazil, 24–26 May 2023; pp. 559–564. [Google Scholar]
- Chen, X.; Zhang, T.; Shen, S.; Zhu, T.; Xiong, P. An optimized differential privacy scheme with reinforcement learning in vanet. Comput. Secur. 2021, 110, 102446. [Google Scholar] [CrossRef]
- Duan, Z.; Mahmood, J.; Yang, Y.; Berwo, M.A.; Yassin, A.A.K.A.; Mumtaz Bhutta, M.N.; Chaudhry, S.A. TFPPASV: A three-factor privacy preserving authentication scheme for VANETs. Secur. Commun. Netw. 2022, 2022, 8259927. [Google Scholar] [CrossRef]
- Erdemir, E.; Dragotti, P.L.; Gündüz, D. Privacy-aware location sharing with deep reinforcement learning. In Proceedings of the 2019 IEEE International Workshop on Information Forensics and Security (WIFS), Delft, The Netherlands, 9–12 December 2019; pp. 1–6. [Google Scholar]
- Zhang, L.; Yan, Y.; Hu, Y. Dynamic flexible scheduling with transportation constraints by multi-agent reinforcement learning. Eng. Appl. Artif. Intell. 2024, 134, 108699. [Google Scholar] [CrossRef]
- Ouadrhiri, A.E.; Abdelhadi, A. Differential privacy for deep and federated learning: A survey. IEEE Access 2022, 10, 22359–22380. [Google Scholar] [CrossRef]
- Sathish Kumar, G.; Premalatha, K.; Uma Maheshwari, G.; Rajesh Kanna, P.; Vijaya, G.; Nivaashini, M. Differential privacy scheme using laplace mechanism and statistical method computation in deep neural network for privacy preservation. Eng. Appl. Artif. Intell. 2024, 128, 107399. [Google Scholar] [CrossRef]
- Gutiérrez, N.; Otero, B.; Rodríguez, E.; Utrera, G.; Mus, S.; Canal, R. A differential privacy protection-based federated deep learning framework to fog-embedded architectures. Eng. Appl. Artif. Intell. 2024, 130, 107689. [Google Scholar] [CrossRef]
- Dwork, C. Differential privacy. In Automata, Languages and Programming; Bugliesi, M., Preneel, B., Sassone, V., Wegener, I., Eds.; Springer: Berlin/Heidelberg, Germany, 2006; pp. 1–12. [Google Scholar]
- McSherry, F.; Talwar, K. Mechanism design via differential privacy. In Proceedings of the 48th Annual IEEE Symposium on Foundations of Computer Science (FOCS’07), Providence, RI, USA, 20–23 October 2007; pp. 94–103. [Google Scholar]
- Laplace, P.S. Memoir on the probability of the causes of events. Stat. Sci. 1986, 1, 364–378. [Google Scholar] [CrossRef]
- Naeem, M.; Rizvi, S.T.H.; Coronato, A. A gentle introduction to reinforcement learning and its application in different fields. IEEE Access 2020, 8, 209320–209344. [Google Scholar] [CrossRef]
- Wang, X.; Wang, S.; Liang, X.; Zhao, D.; Huang, J.; Xu, X.; Dai, B.; Miao, Q. Deep reinforcement learning: A survey. IEEE Trans. Neural Netw. Learn. Syst. 2022, 35, 5064–5078. [Google Scholar] [CrossRef]
- Liu, X.; Ren, M.; Yang, Z.; Yan, G.; Guo, Y.; Cheng, L.; Wu, C. A multi-step predictive deep reinforcement learning algorithm for HVAC control systems in smart buildings. Energy 2022, 259, 124857. [Google Scholar] [CrossRef]
- Lovejoy, W.S. A survey of algorithmic methods for partially observed Markov decision processes. Ann. Oper. Res. 1991, 28, 47–65. [Google Scholar] [CrossRef]
- Feng, T.; Zhang, Z.; Wong, W.-C.; Sun, S.; Sikdar, B. A framework for tradeoff between location privacy preservation and quality of experience in location based services. IEEE Open J. Veh. Technol. 2024, 5, 428–439. [Google Scholar] [CrossRef]
- Lu, Z.; Shen, H. A new lower bound of privacy budget for distributed differential privacy. In Proceedings of the 2017 18th International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT), Taipei, Taiwan, 18–20 December 2017; pp. 25–32. [Google Scholar]
- Shen, X.; Jiang, H.; Chen, Y.; Wang, B.; Gao, L. Pldp-fl: Federated learning with personalized local differential privacy. Entropy 2023, 25, 485. [Google Scholar] [CrossRef]
- Zhu, T.; Li, J.; Hu, X.; Xiong, P.; Zhou, W. The dynamic privacy-preserving mechanisms for online dynamic social networks. IEEE Trans. Knowl. Data Eng. 2022, 34, 2962–2974. [Google Scholar] [CrossRef]
- Xu, H.; Fan, Z.; Liu, X. Application of personalized federated learning methods to environmental sound classification: A comparative study. Eng. Appl. Artif. Intell. 2024, 135, 108760. [Google Scholar] [CrossRef]
- Alebouyeh, Z.; Bidgoly, A.J. Privacy-preserving federated learning compatible with robust aggregators. Eng. Appl. Artif. Intell. 2025, 143, 110078. [Google Scholar] [CrossRef]
State Variable | Symbol | Explanation |
---|---|---|
Vehicle density | Current vehicle density in the area (vehicles/km2) | |
Average speed | Average driving speed of vehicles within the area (km/h) | |
Temporal characteristics | Number of hours in a day, reflecting time patterns | |
Sensitive area identification | Sensitive area identification (1 for sensitive area, 0 for regular area) | |
Privacy level | Current differential privacy parameter value | |
Utility score | Current data utility evaluation score |
Parameters | Define |
---|---|
Privacy budget | |
Sensitivity | |
Discount factor | |
Privacy parameter tuning volume | |
Sensitivity modulus | |
Noise level | |
Vehicle density | |
Average speed | |
Time-specificity | |
Regional sensitivities | |
Privacy level | |
Utility score |
Parameter Name | Symbol | Default Value | Description |
---|---|---|---|
Learning Rate | 0.001 | Adam optimizer learning rate | |
Discount Factor | 0.99 | Future reward discount | |
Initial Exploration Rate | 1.0 | Initial -greedy exploration rate | |
Final Exploration Rate | 0.01 | Minimum exploration rate | |
Exploration Decay Rate | 0.995 | Exploration rate decay | |
Batch Size | B | 64 | Training batch size |
Total Privacy Budget | 150.0 | Global privacy budget | |
Budget Decay Rate | 0.98 | Budget allocation decay rate |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Wang, G.; Liu, X.; Zheng, Y.; Zhang, Z.; Cai, Z. DNO-RL: A Reinforcement-Learning-Based Approach to Dynamic Noise Optimization for Differential Privacy. Electronics 2025, 14, 3122. https://doi.org/10.3390/electronics14153122
Wang G, Liu X, Zheng Y, Zhang Z, Cai Z. DNO-RL: A Reinforcement-Learning-Based Approach to Dynamic Noise Optimization for Differential Privacy. Electronics. 2025; 14(15):3122. https://doi.org/10.3390/electronics14153122
Chicago/Turabian StyleWang, Guixin, Xiangfei Liu, Yukun Zheng, Zeyu Zhang, and Zhiming Cai. 2025. "DNO-RL: A Reinforcement-Learning-Based Approach to Dynamic Noise Optimization for Differential Privacy" Electronics 14, no. 15: 3122. https://doi.org/10.3390/electronics14153122
APA StyleWang, G., Liu, X., Zheng, Y., Zhang, Z., & Cai, Z. (2025). DNO-RL: A Reinforcement-Learning-Based Approach to Dynamic Noise Optimization for Differential Privacy. Electronics, 14(15), 3122. https://doi.org/10.3390/electronics14153122