At the 15th International Conference on Dependable Systems, Services and Technologies (DESSERT’2025), held on December 19, 2025, in Athens, Greece, a research study by Professor Vitaliy Kobets in collaboration with PhD student Dmytro Klonov was presented in an online format.
The study, titled "AI-Driven Intellectual Supply Chain Resilience: Agent-Based Modeling with Machine Learning Integration in Dynamic Environments," was presented during the workshop "Securing the Future: AI-Based Protection and Strategies for Dependable Systems." This research addresses one of the most pressing challenges facing modern enterprises: building supply chains capable of withstanding global disruptions.
The research focuses on real-world issues affecting industries globally. Modern supply chains encounter severe disruptions due to pandemics, geopolitical conflicts, energy crises, market volatility, and logistical bottlenecks. Traditional approaches based on static planning and centralized decision-making are no longer sufficient in today’s volatile environment, leading to production delays, high costs, and diminished service levels.
The authors developed an innovative multi-agent framework for vertically integrated supply chains, where each participant—suppliers, manufacturers, distributors, and transport operators—functions as an autonomous agent. These agents make independent decisions while coordinating with others, creating a dynamic and adaptive system that responds to challenges in real-time.
The innovation lies in the integration of advanced Machine Learning (ML) technologies directly into the agents' decision-making processes. The framework utilizes demand forecasting and Reinforcement Learning (RL) to optimize the actions of supply chain participants. This ensures that decisions are data-driven rather than purely rule-based, allowing the system to learn from patterns and adapt to shifting conditions.
Furthermore, the study proposes the implementation of a Digital Twin—a virtual replica of the physical supply chain that is continuously synchronized with operational data. This enables real-time monitoring, predictive management, and scenario analysis, providing managers with powerful decision-support tools.
The research demonstrates significant practical value: an analytical model comparing uniform unified tariffs with tariff discrimination shows that while total output and market prices remain unchanged, tariff discrimination reduces profit disparities among manufacturers by approximately 50%, promoting a more equitable distribution of value across the supply chain network.
DESSERT’2025, held at the historic Hellenic Armed Forces Officers' Club near the Lyceum of Aristotle, gathered leading researchers and industry experts from around the world. The hybrid format facilitated both in-person and remote participation. The conference focused on dependable and trustworthy technologies in Artificial Intelligence, Internet of Things (IoT), Cloud Computing, Cybersecurity, and other critical technologies shaping our digital future.
Information prepared by Vitaliy Kobets, Professor of the Department of Computer Science and Software Engineering.
You can: Print version Send to friend