TrustAgg: A Privacy-Preserving Data Aggregation Framework for Industrial IoT Based on TEE, Homomorphic Encryption, and Differential Privacy

Authors

  • Jian Zhang School of Computer Science and Software Engineering, Southwest Petroleum University, Chengdu, China Author
  • Fucheng Wang School of Computer Science and Software Engineering, Southwest Petroleum University, Chengdu, China Author
  • Cunrui Ying School of Civil Engineering and Geomatics, Southwest Petroleum University, Chengdu, China Author

DOI:

https://doi.org/10.70088/h7aax866

Keywords:

data aggregation, internet of things, trusted computing, homomorphic encryption, differential privacy

Abstract

The proliferation of Industrial Internet of Things (IIoT) devices has intensified the tension between data utility and privacy protection. Existing privacy-preserving data aggregation schemes typically rely on either cryptographic techniques or differential privacy alone, leaving vulnerabilities in trust assumptions and adversary models. This paper proposes TrustAgg, a novel three-layer privacy-preserving data aggregation framework that synergistically integrates Trusted Execution Environment (TEE), Paillier homomorphic encryption, and two-stage differential privacy. The hierarchical architecture ensures robust confidentiality even against malicious adversaries: IIoT devices encrypt measurements with Paillier before local differential privacy perturbation; edge gateways perform secure aggregation within isolated TEE enclaves; and the cloud server injects final differential privacy noise while coordinating threshold-based distributed decryption. Formal security analysis under the universal composability framework demonstrates semantic security, strict epsilon-differential privacy, and resilience against malicious adversaries. Experimental evaluation on a real-world oilfield production dataset and public benchmarks shows that TrustAgg completes aggregation for 796 devices in 4.12 seconds with a 2.9% relative error at ε = 1.0, while achieving a membership inference success rate of 52.3%, which is close to random guessing. Compared to state-of-the-art schemes, TrustAgg provides significantly stronger security guarantees with comparable computational efficiency, thereby offering a highly practical and secure solution for large-scale industrial data analytics. Furthermore, the proposed framework effectively mitigates data leakage risks during the transmission and processing phases, ensuring compliance with stringent industrial data protection regulations.

References

J. Fan and F. Vercauteren, “Somewhat practical fully homomorphic encryption,” Cryptology ePrint Archive, Report 2012/144, 2012.

Z. Brakerski, C. Gentry, and V. Vaikuntanathan, “(Leveled) fully homomorphic encryption without bootstrapping,” in Proc. 3rd Innovations in Theoretical Computer Science Conf. (ITCS), Cambridge, MA, USA, pp. 309–325, 2012.

P. Paillier, “Public-key cryptosystems based on composite degree residuosity classes,” in Advances in Cryptology – EUROCRYPT ’99, pp. 223–238, 1999.

S. Kadhe et al., “FastSecAgg: Scalable secure aggregation for privacy-preserving federated learning,” arXiv preprint arXiv:2009.11248, 2020.

C. Dwork and A. Roth, “The algorithmic foundations of differential privacy,” Found. Trends Theor. Comput. Sci., vol. 9, no. 3–4, pp. 211–407, 2014.

K. Bonawitz et al., “Practical secure aggregation for privacy-preserving machine learning,” in Proc. 2017 ACM SIGSAC Conf. Computer and Communications Security (CCS), Dallas, TX, USA, pp. 1175–1191, 2017.

S. P. Kasiviswanathan et al., “What can we learn privately?,” SIAM J. Comput., vol. 40, no. 3, pp. 793–826, 2011.

ARM Limited, “TrustZone technology for ARMv8-M architecture,” ARM Technical White Paper, 2016.

J. W. Kim, D. H. Kim, and B. Jang, “Application of local differential privacy to indoor positioning data,” IEEE Access, vol. 6, pp. 51555–51565, 2018.

Intel Corporation, “Intel Software Guard Extensions Programming Reference,” Document No. 329298-002US, 2015.

O. Babalola, J. Cano, N. H. Tran et al., “Survey of privacy preserving federated tree-based learning,” Artificial Intelligence Review, vol. 59, no. 7, art. no. 123, 2026. (to appear)

O. Ohrimenko et al., “Oblivious multi-party machine learning on trusted processors,” in Proc. 25th USENIX Security Symp. (USENIX Security 16), Austin, TX, USA, pp. 619–636, 2016.

N. Papernot et al., “Scalable private learning with PATE,” in Proc. 6th Int. Conf. Learning Representations (ICLR), Vancouver, BC, Canada, 2018.

S. H. Hashemi et al., “Secure genomic data analysis using Intel SGX and homomorphic encryption,” BMC Bioinformatics, vol. 21, no. 1, art. no. 449, 2020.

R. Canetti, “Universally composable security: A new paradigm for cryptographic protocols,” in Proc. 42nd IEEE Symp. Foundations of Computer Science (FOCS), pp. 136–145, 2001.

F. Loukil, C. Ghedira-Guegan, K. Boukadi, and A.-N. Benharkat, “Privacy-preserving IoT data aggregation based on blockchain and homomorphic encryption,” Sensors, vol. 21, no. 7, art. no. 2452, 2021.

R. Krajewski, J. Bock, L. Kloeker, and L. Eckstein, “The highD dataset: A drone dataset of naturalistic vehicle trajectories on German highways for validation of highly automated driving systems,” in Proc. IEEE Intell. Vehicles Symp. (IV), pp. 2118–2125, 2018.

F. Brasser, U. Müller, A. Dmitrienko, K. Kostiainen, S. Čapkun, and A.-R. Sadeghi, “Software grand exposure: SGX cache attacks are practical,” in Proc. 11th USENIX Workshop Offensive Technol. (WOOT), Vancouver, BC, Canada, 2017.

A. Nejadhossein Qasemabadi, S. Mozaffari, M. Ahmadi, and S. Alirezaee, “Privacy-preserving lane change prediction using recurrent neural network with secure multiparty computation,” Int. J. Intell. Transp. Syst. Res., vol. 23, pp. 893–907, 2025.

Downloads

Published

18 November 2026

Issue

Section

Article

How to Cite

Zhang, J., Wang, F., & Ying, C. (2026). TrustAgg: A Privacy-Preserving Data Aggregation Framework for Industrial IoT Based on TEE, Homomorphic Encryption, and Differential Privacy. Artificial Intelligence and Digital Technology, 3(6), 41-59. https://doi.org/10.70088/h7aax866