Quantifying Systemic Contagion in Interbank Payment Networks: A Machine Learning Approach to Real-Time DDoS and Ransomware Mitigation

• Abey Litty

  Texas University

  Author

The digital transformation of financial services has rendered interbank payment networks increasingly vulnerable to systemic contagion triggered by cyber threats, particularly Distributed Denial-of-Service (DDoS) attacks and ransomware incidents. While existing risk management frameworks address individual bank security, they fail to model how localized cyber incidents propagate through interconnected payment networks, creating cascading liquidity pressures and settlement failures. This study addresses this gap by developing a machine learning framework that quantifies systemic contagion risks in interbank payment networks and enables real-time mitigation of cyber-propagated threats. The proposed methodology integrates graph neural network architecture with temporal transaction pattern analysis to model payment flows, coupled with a hybrid DDoS-ransomware propagation model calibrated against historical incident data. The framework was validated using Bank for International Settlements consolidated banking statistics (2000-2015) and synthetic interbank transaction networks aligned with FDIC regulatory data. Results demonstrate that the GNN-based model achieves 89.4% accuracy in predicting contagion propagation pathways, with early warning lead times of 11.5 days compared to 3.2 days for traditional threshold-based systems. Feature importance analysis identified network density (weight=0.34), institution centrality (weight=0.28), and anomalous payment velocity (weight=0.22) as the strongest predictors of contagion risk. The framework contributes to the literature by providing a validated, replicable methodology for systemic cyber-risk quantification, with practical implications for regulatory stress testing and real-time security operations deployment.

• PDF

Copyright (c) 2026 Abey Litty (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.