SHADE : Privacy-preserving multivariate anomaly detection in industrial IoT using CKKS homomorphic encryption
*G. Hemanth KumarCorresponding authorDepartment of Computer Science and Engineering (Cyber Security)School of EngineeringDayananda Sagar UniversityBangalore, Karnataka, 562112, IndiaView full profile → , Dilip Kumar Jang Bahadur Sainidilipsaini@gmail.comDepartment of Computer Science and Engineering (Cyber Security)School of EngineeringDayananda Sagar UniversityBangalore, Karnataka, 562112, IndiaView full profile → , P. V. Bhaskar Reddybhaskarreddy.pv@reva.edu.inDepartment of Computer Science and EngineeringSchool of Computer Science and EngineeringREVA UniversityBengaluru, Karnataka, 560064, IndiaView full profile → , C. Ravindra Murthyravins.ch@gmail.comDepartment of Electronics & Communication EngineeringK. S. R. M. College of EngineeringKadapa, Andhra Pradesh, 516005, IndiaView full profile → , Yudhishthir Rautyudhishthir.raut@sitpune.edu.inDepartment of Robotics & AutomationSymbiosis Institute of Technology (SIT)Lavle, Pune, Maharashtra, 412115, IndiaView full profile → , Tiruveedula Gopikrishnatiruveedula-ds@dsu.edu.inDepartment of Computer Science and Engineering (Data Science)School of EngineeringDayananda Sagar UniversityBangalore, Karnataka, 562112, IndiaView full profile →
* Corresponding author · click or hover a name for details
- Received:
- 01 Mar 2026
- Published Online:
- 31 Aug 2026
- Article type:
- Research Article
- Language:
- EN
- Article no.:
- JIOS-2359
- Pages:
- 2977–2986
Abstract
The vast amount of sensor data provided by Industrial Internet of Things (IIoT) systems needs to be processed in real time process. Besides the technical challenge, there are also compliance issues that need to be taken into account. Recently, a large number of regulations have been introduced in order to protect the privacy of citizens, e.g. GDPR and IEC 62443. Existing works on anomaly detection in the IIoT environment have to decrypt the sensor data in order to apply the anomaly detection algorithm, exposing sensitive operational information. In this paper, the proposed framework is based on the CKKS homomorphic encryption scheme, is proposed. The framework enables the application of an anomaly detection algorithm directly in the encrypted data. With SHADE, the multivariate Mahalanobis distance metric can be calculated in the ciphertext domain, and the encrypted comparison result can be evaluated using a Chebyshev polynomial approximation for the threshold computation. The performance of SHADE is evaluated using synthetic IIoT datasets. The experimental results show that the proposed SHADE framework is able to achieve an Area under Curve (AUC) of 0.99 or even higher than its plaintext counterparts with only less than (ΔAUC < 0.0001) loss in AUC, while maintaining a very low false positive rate and enabling real-time processing.
Keywords
Subject Classifications
References
[1] J. H. Cheon, A. Kim, M. Kim, and Y. Song, “Homomorphic encryption for arithmetic of approximate numbers,” in Advances in Cryptology—ASIACRYPT 2017, Lecture Notes in Computer Science, vol. 10624, Cham, Switzerland: Springer, pp. 409–437 (2017), doi: 10.1007/978-3-319-70694-8_15.
[2] A. Al Badawi, J. Bates, F. Bergamaschi, D. B. Cousins, S. Erabelli, N. Genise, S. Halevi, H. Hunt, A. Kim, Y. Lee, and Z. Liu, “OpenFHE: Open-source fully homomorphic encryption library,” in Proc. 10th Workshop Encrypted Comput. Appl. Homomorphic Cryptogr. (WAHC), pp. 53–63 (Nov. 2022).
[3] J. Lee, P. N. Duong, and H. Lee, “Configurable encryption and decryption architectures for CKKS-based homomorphic encryption,” Sensors, vol. 23, no. 17, Art. no. 7389 (2023).
[4] M. Boura, C. Carabelli, A. Dautbegovic, V. Laithier, and M. Zuber, “Approximate homomorphic encryption based privacy-preserving machine learning: A survey,” Artif. Intell. Rev., vol. 58, no. 1, Art. no. 25 (2025).
[5] H. Li, W. Mo, C. Shen, and L. Pang, “EvalComp: Bootstrapping based on homomorphic comparison function for CKKS,” IEEE Trans. Inf. Forensics Security, vol. 20, pp. 1349–1361 (2025).
[6] Y. B. Wiryen, N. W. A. Vigny, M. J. Ngono, and F. L. Aimé, “Leveraging TenSEAL: A comparative study of BFV and CKKS schemes for training ML models on encrypted IoT data,” Int. J. Inf. Security Privacy, vol. 18, no. 1, pp. 1–17 (2024).
[7] J. H. Cheon, M. Kang, T. Kim, J. Jung, and Y. Yeo, “Batch inference on deep convolutional neural networks with fully homomorphic encryption using channel-by-channel convolutions,” IEEE Trans. Dependable Secure Comput., vol. 22, no. 2, pp. 1674–1685 (2024).
[8] M. Arazzi, S. Nicolazzo, and A. Nocera, “A fully privacy-preserving solution for anomaly detection in IoT using federated learning and homomorphic encryption,” Inf. Syst. Front., vol. 27, pp. 367–390 (2025).
[9] A. Hangan, D. Lazea, and T. Cioara, “Privacy preserving anomaly detection on homomorphic encrypted data from IoT sensors,” arXiv preprint arXiv:2403.09322 (2024).
[10] H. Qin, D. He, Q. Feng, M. K. Khan, M. Luo, and K.-K. R. Choo, “Cryptographic primitives in privacy-preserving machine learning: A survey,” IEEE Trans. Knowl. Data Eng., vol. 36, no. 5, pp. 1919–1934 (2023).
[11] D. Torre, A. Chennamaneni, J. Jo, G. Vyas, and B. Sabrsula, “Toward enhancing privacy preservation of a federated learning CNN intrusion detection system in IoT: Method and empirical study,” ACM Trans. Softw. Eng. Methodol., vol. 34, no. 2, pp. 1–48 (2025).
[12] B. Chen, L. Wu, and Y. Liu, “Privacy-preserving machine learning based on cryptography: A survey,” ACM Trans. Knowl. Discovery Data, vol. 19, no. 6, pp. 1–46 (2025).
[13] A. Abusitta, G. H. de Carvalho, O. A. Wahab, T. Halabi, B. C. M. Fung, and S. Al Mamoori, “Deep learning-enabled anomaly detection for IoT systems,” Internet Things, vol. 21, Art. no. 100656 (2023).
[14] Z. Chen, Z. Li, J. Huang, S. Liu, and H. Long, “An effective method for anomaly detection in industrial Internet of Things using XGBoost and LSTM,” Sci. Rep., vol. 14, Art. no. 24209 (2024).
[15] A. Alqazzaz, “SecuFL-IoT: An adaptive privacy-preserving federated learning framework for anomaly detection in smart industrial networks,” Sci. Rep., vol. 16, Art. no. 4107 (2026), doi: 10.1038/s41598-025-11883-1.
[16] R. Joshi, P. Mathur, A. K. Gupta, S. Singh, V. Paliwal, and S. Nayar, “Mathematical modeling of intelligent system for predicting effectiveness of premenstrual syndrome,” J. Interdiscip. Math., vol. 26, no. 3, pp. 551–562 (2023).
[17] A. Noonia, D. Thakral, P. Mathur, F. Sheth, H. Shaikh, and A. K. Gupta, “A discrete mathematical model and cryptography for secure medical image analysis: Encrypted chest X-ray classification,” J. Discrete Math. Sci. Cryptogr., vol. 28, no. 5-A, pp. 1473–1486 (2025).
