FedCAD: Federated Contrastive-Aware Decoupling for Heterogeneous Power System API Security Detection
Main Article Content
Abstract
Digital power grids increasingly rely on API-based interaction among dispatch automation systems, distribution management systems, advanced metering infrastructure, marketing platforms, energy management systems, distributed energy resource access platforms, and edge IoT gateways. These interfaces support remote monitoring, operation command delivery, load forecasting, meter data acquisition, fault isolation, and cross-domain business collaboration, but they also introduce new attack surfaces for unauthorized command invocation, abnormal parameter tampering, replay access, excessive service calling, and sensitive grid-data leakage. Building an accurate API security detection model for such scenarios is challenging because power-grid data cannot be freely centralized: dispatch logs, equipment identifiers, customer electricity information, station topology, and operation records are distributed across regional grids, voltage levels, business departments, and terminal types, and are subject to strict privacy and operational-security constraints. However, power-system API security data usually exhibits multiple forms of heterogeneity. Label distributions vary because different regions and business systems face different proportions of normal operation, abnormal metering access, dispatch-command abuse, distribution-terminal intrusion, and new-energy access anomalies. Feature distributions vary because API traffic is strongly coupled with grid operation modes, voltage levels, device types, communication protocols, seasonal load patterns, and local gateway configurations. In addition, missing timestamps, clock drift, incomplete gateway logs, noisy alarm labels, and irregular field-device communication may corrupt the observed data. These factors lead to inconsistent power API behavior representations, shifted risk decision boundaries, and local overfitting in traditional distributed learning methods. To address these challenges, this paper proposes FedCAD (Federated Contrastive-Aware Decoupling), a federated framework for heterogeneous power system API security detection. FedCAD decouples representation learning from classifier adaptation. A hierarchical contrastive learning module aligns power API behavior representations at both instance and risk-category levels, while preserving region-specific and business-specific operational characteristics. An adaptive classifier module further adjusts decision boundaries according to local power API risk distributions through logit adjustment and local fine-tuning. The proposed framework enables collaborative security modeling across dispatch, distribution, metering, marketing, and new-energy systems without sharing raw power-grid data, improving robustness and generalization under multi-source heterogeneous conditions.
Downloads
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. Agüera y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in AISTATS, 2017, pp. 1273–1282.
N. Rieke, J. Hancox, W. Li, et al., “The future of digital health with federated learning,” NPJ Digital Medicine, vol. 3, no. 1, pp. 1–7, 2020.
Q. Yang, Y. Liu, T. Chen, and Y. Tong, “Federated machine learning: Concept and applications,” ACM Transactions on Intelligent Systems and Technology, vol. 10, no. 2, pp. 1–19, 2019.
A. Hard, K. Rao, R. Mathews, S. Ramaswamy, F. Beaufays, S. Au-genstein, et al., “Federated learning for mobile keyboard prediction,” arXiv:1811.03604, 2018.
T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V. Smith, “Federated optimization in heterogeneous networks,” in MLSys, 2020, pp. 429–450.
Y. Zhao, M. Li, L. Lai, N. Suda, D. Civin, and V. Chandra, “Federated learning with non-IID data,” arXiv:1806.00582, 2018.
X. Li, M. Jiang, X. Zhang, M. Kamp, and Q. Dou, “FedBN: Federated learning on non-IID features via local batch normalization,” in ICLR, 2021.
S. P. Karimireddy, S. Kale, M. Mohri, S. J. Reddi, S. U. Stich, and A. T. Suresh, “SCAFFOLD: Stochastic controlled averaging for federated learning,” in ICML, 2020, pp. 5132–5143.
Z. Qu, X. Li, R. Duan, Y. Liu, B. Tang, and Z. Lu, “Generalized federated learning via sharpness aware minimization,” in ICML, 2022, pp. 18250– 18280.
R. Dai, X. Yang, Y. Sun, L. Shen, X. Tian, M. Wang, and Y. Zhang, “FedGAMMA: Federated learning with global sharpness-aware minimization,” IEEE Transactions on Neural Networks and Learning Systems, vol. 35, no. 12, pp. 17479–17492, 2024.
T. Yoon, S. Shin, S. J. Hwang, and E. Yang, “FedMix: Approximation of mixup under mean augmented federated learning,” in ICLR, 2021.
Y. Yan, H. Fu, Y. Li, et al., “A simple data augmentation for feature distribution skewed federated learning,” in CVPR, 2025, pp. 25749–25758.
T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in ICML, 2020, pp. 1597–1607.
K. He, H. Fan, Y. Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in CVPR, 2020, pp. 9729– 9738.
Q. Li, B. He, and D. Song, “Model-contrastive federated learning,” in CVPR, 2021, pp. 10713–10722.
Y. Tan, G. Long, J. Ma, L. Liu, T. Zhou, and J. Jiang, “Federated learning from pre-trained models: A contrastive learning approach,” in NeurIPS, 2022, pp. 19332–19344.
B. Kang, S. Xie, M. Rohrbach, Z. Yan, A. Gordo, J. Feng, and Y. Kalantidis, “Decoupling representation and classifier for long-tailed recognition,” in ICLR, 2019.
Y. Li, N. Wang, J. Shi, X. Hou, and J. Liu, “Adaptive batch normalization for practical domain adaptation,” Pattern Recognition, vol. 80, pp. 109– 117, 2018.
A. K. Menon, S. Jayasumana, A. S. Rawat, H. Jain, A. Veit, and S. Kumar, “Long-tail learning via logit adjustment,” in ICLR, 2020.
D. A. E. Acar, Y. Zhao, R. Matas, M. Mattina, P. Whatmough, and V. Saligrama, “Federated learning based on dynamic regularization,” in ICLR, 2021.