MODELING CYBER THREATS IN AUTONOMOUS GUIDED VEHICLES USING MEAN FIELD MODELS

This paper presents a novel analytical framework for modeling cyber threats targeting Autonomous Guided Vehicles (AGVs) in logistics scenarios, with a particular focus on large-scale port environments. We leverage a Markovian Agent Model (MAM) supported by mean field theory to capture the dynamic interplay among AGVs, attackers, control centers, and security systems. The originality of the work lies in its ability to formally characterize cyber-physical interactions in AGV ecosystems through scalable, differential equation-based approximations, which remain computationally tractable even for large populations of agents. By modeling various states-such as compromised, detected, and mitigated-across interacting agents, the study reveals how attack propagation, detection delays, and countermeasures impact system stability over time. Results demonstrate the model's effectiveness in forecasting AGV losses, assessing control recovery efforts, and quantifying the timing and efficiency of security responses.