Federated-SDN Hybridization for Proactive and Explainable Cyber Defense in Mega-Event Infrastructures
Year: 2025
Type: preprint
Abstract: Mega-events such as international summits, global sports competitions, and political gatherings face heightened cybersecurity risks, particularly from sophisticated distributed denial of service (DDoS) and multi-vector attacks that can disrupt accreditation, streaming, and emergency communication services. Traditional approaches-threshold-based anomaly detection, signature-driven intrusion detection, and centralized scrubbing-do not satisfy the stringent latency, scalability, privacy, and accountability requirements of these environments. This paper presents a Federated-SDN Hybridization framework that integrates federated learning for distributed, privacy-preserving anomaly detection; software-defined networking for real-time orchestration; predictive modules for short-horizon traffic forecasting; and explainable AI for transparent, compliance-ready justifications. Experimental validation on benchmark datasets (CICIDS2017, ToN-IoT, HL-IoT) and a COP30-scale emulation shows that the framework achieves high detection accuracy (F1 ≈ 0.95), subsecond mitigation for most attacks, modest resource overhead, and increased analyst trust through auditable explanations. These results indicate the framework's potential as a next-generation cyber-defense paradigm for securing critical digital infrastructures in mega-events.
Authors: Wallace Cruz, Willian Rodrigo, Allan Douglas Bento da Costa
Institution: Universidade Federal Rural da Amazônia
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4414365975
Cites:
Cited by:
Related to: