EdgeGuard-FL: A Privacy-Preserving Federated Learning Framework for Intrusion Detection in Smart Campus IoT Networks

EdgeGuard-FL proposes a privacy-conscious approach to collaborative intrusion detection in smart campus networks. Its central idea is to let edge nodes learn from one another's experience without sharing raw traffic. Local detection, clipped and noised updates, trust-weighted aggregation, drift monitoring, and a graded response policy work together to reduce privacy exposure and the effect of faulty or malicious participants. The framework also requires independent authorisation for enforcement actions, careful logging, source-separated evaluation, and explicit measurement of communication and latency cost. The approach is modular and suitable for a final-year CSE prototype that begins with replayed benchmark data and gradually adds federated defences. Because the framework has not yet been experimentally validated, its effectiveness remains a research question rather than a proven result. A completed implementation should use a well-labelled benchmark, report actual metrics, analyse errors, and state its limitations clearly. Secure collaborative detection requires layered controls and continuous evaluation; a learned detector is valuable, but it cannot replace network segmentation, patching, and access control.

[1]     H. B. McMahan et al., "Communication-Efficient Learning of Deep Networks from Decentralized Data," in Proc. AISTATS, PMLR vol. 54, pp. 1273–1282, 2017. [2]     T. Li et al., "Federated Optimization in Heterogeneous Networks," in Proc. Machine Learning and Systems (MLSys), 2020. [3]     P. Kairouz et al., "Advances and Open Problems in Federated Learning," Foundations and Trends in Machine Learning, vol. 14, no. 1–2, pp. 1–210, 2021. [4]     C. Dwork and A. Roth, "The Algorithmic Foundations of Differential Privacy," Foundations and Trends in Theoretical Computer Science, vol. 9, no. 3–4, pp. 211–407, 2014. [5]     K. Bonawitz et al., "Practical Secure Aggregation for Privacy-Preserving Machine Learning," in Proc. ACM CCS, pp. 1175–1191, 2017. [6]     M. Abadi et al., "Deep Learning with Differential Privacy," in Proc. ACM CCS, pp. 308–318, 2016. [7]     Y. Mirsky et al., "Kitsune: An Ensemble of Autoencoders for Online Network Intrusion Detection," in Proc. NDSS, 2018. [8]     I. Sharafaldin, A. H. Lashkari, and A. A. Ghorbani, "Toward Generating a New Intrusion Detection Dataset and Intrusion Traffic Characterization," in Proc. ICISSP, 2018. [9]     P. Blanchard et al., "Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent," in Advances in Neural Information Processing Systems, vol. 30, 2017. [10]   E. Bagdasaryan et al., "How To Backdoor Federated Learning," in Proc. AISTATS, PMLR vol. 108, 2020.