FLBEC: A Unified Framework Integrating Federated Learning, Blockchain, and Edge Computing for Privacy-Preserving Cybersecurity Governance in Next-Generation Networks

Abstract

Across healthcare, telecommunications, and critical infrastructure, organizations are sitting on threat intelligence that could protect their peers and sharing almost none of it. The reason is not selfishness; it is a genuine legal and competitive constraint. Data that reveals one network's vulnerabilities also reveals its patients, its subscribers, or its operational secrets. This paper begins with that everyday reality and asks a practical question: can we build a system that enables organizations to learn from each other's threats without ever exposing the underlying data? We propose FLBEC, a framework that binds together Federated Learning (FL), Blockchain governance (BC), and Edge Computing intelligence (EC) to deliver threat detection that is simultaneously private, auditable, real-time, and ready standards. We evaluate FLBEC across five benchmark datasets, NSL-KDD, CICIDS-2017, CTU-13, UNSW-NB15, and a custom healthcare IoT dataset, measuring detection accuracy, false positive rate, AUC-ROC, inference latency, and communication overhead. We also run a systematic ablation study to isolate what each of the three components actually contributes. FLBEC reaches 96.3% detection accuracy, 3.7% false positive rate, and AUC-ROC of 0.963, clearing every single-paradigm and pairwise baseline. End-to-end response latency sits at 42 ms, and communication overhead drops to 1.8 GB per federated round. Every component passes the ablation test for independent statistical significance at p < 0.05. The results confirm that privacy, governance, and real-time performance are not a trade-off triangle, they become mutually reinforcing when the three paradigms are co-designed from the start. FLBEC maps directly to IEEE 802.1X, 3GPP Release 18, and NIST CSF 2.0, making it deployable in regulated industries without modification.