Abstract

Multi-modal transportation networks in the Middle East are embedded within one of the world's most operationally volatile logistical environments, where compound uncertainty from geopolitical instability, extreme climatic events, infrastructure heterogeneity, and demand unpredictability routinely disrupts conventional static routing approaches. Despite growing research in dynamic routing and supply chain resilience globally, a critical gap persists in the development of regionally calibrated, real-time adaptive logistics models capable of managing multi-modal interdependencies under simultaneous, compounding uncertainty sources specific to the Middle East. This study addresses this gap by proposing a hybrid dynamic routing framework that integrates stochastic dynamic programming, deep reinforcement learning, and live IoT sensor data to enable real-time route optimization across road, rail, air, and maritime modalities. The research objective is to design, evaluate, and validate a computationally tractable adaptive logistics system that maximizes service levels and supply chain resilience while minimizing cost and delay under uncertainty. Simulation experiments calibrated on Gulf Cooperation Council (GCC) logistics corridors demonstrate that the proposed dynamic routing model reduces average delivery delays by 28–34% and improves composite supply chain resilience metrics by 41% relative to deterministic routing baselines. Furthermore, the framework achieves a 19% reduction in total logistics cost under moderate disruption scenarios. Findings confirm that integrating machine learning-driven re-routing with stochastic modeling produces significant operational gains, and that adaptive logistics systems represent a viable and scalable path toward robust supply chain resilience in the Middle East. Implications extend to policymakers, logistics operators, and infrastructure planners navigating the region's evolving transportation landscape.