Graph-Theoretic Approaches to Resilience: Strengthening Al Systems Against Coordinated Cyberattacks

This study proposes a multilayered graph-theoretic framework to improve the resilience of interconnected infrastructure, such as IoT infrastructure, autonomous vehicles, and smart cities, against cyber threats.By harnessing Artificial Intelligence techniques with graph-theoretic models, a solution enables real-time adaptation to changes in attack patterns.Based on the seminal work of Pirani and Mitra, adaptive algorithms optimize the response of the system to a cyber threat as these threats evolve.Using real-time traffic data with four years of archive data from San Francisco Bay Area Traffic Sensors, the model was validated against various cyber-attacks by simulation, changing metrics used for evaluation, namely the attack impact score, vulnerability index, resilience score, and adaptability.The results indicated a large improvement in resilience, with the attack impact being reduced to 0.10 from 0.70, the vulnerability index dropping from 0.85 to 0.30, and the resilience index increasing from 0.60 to 0.90 after implementing real-time adjustments.The adaptability metric changed from low to high after optimization and adjustment in the real-time phases.These results show that an AI and graph-theoretic paradigm, such as this, has advantages over classical methods and provides a scalable solution for strengthening critical infrastructure while assuring real-time mitigation plans for safeguarding interconnected systems.