Reducing Interception Latency in High-Security Facilities via AO* Optimization Algorithm

Abstract

Traditional physical security systems are largely reactive, relying on localized alarms that trigger only after a perimeter breach has occurred. In complex, high-value environment – such as data centers or nuclear facilities – this latency often allows intruders to reach their objectives before an interceptor can arrive. This thesis proposes a proactive security framework that models facility logic as an AND-OR graph, representing potential “attack recipes” as a series of interdependent security states. By employing the AO* optimization algorithm, the system dynamically identifies the most probable paths an intruder might take based on real-time Anomaly detection from distributed sensor networks. To ensure the integrity of the decision-making process against digital tampering, the framework integrates Elliptic Curve Digital Signature Algorithms (ECDSA) to authenticate every state change within the graph. The core of this research investigates whether AO* optimization can significantly reduce the response time of autonomous mobile interceptors (drones or robots) by positioning them at “logic bottlenecks” – points in the AND-OR graph that an intruder must satisfy to complete their objective. Preliminary simulation suggests that this predictive allocation outperforms traditional patrolling methods, providing a robust, self-healing, and mathematically verifiable approach to modern autonomous defense.