The evolution of networked systems, driven by innovations in software-defined networking, network function virtualization, open radio access networks (O-RAN) and cloud-native architectures, is redefining both the operational landscape and the threat surface of critical infrastructures. This paper offers an in-depth, interdisciplinary examination of how resilience must be reconceptualized and re-engineered to address the multifaceted challenges posed by these transformations.
The paper is structured as a cohesive framework that progresses from characterization of the threat landscape to formal definitions and metrics of resilience, followed by theoretical foundations rooted in control theory, game theory, learning and network science. Building on these foundations, the work develops methodologies for quantitative risk assessment, including probabilistic modeling, network analytics and emerging agentic AI approaches leveraging large language models (LLMs).
The final part of the paper focuses on resilient system design paradigms in NextG networks, with emphasis on risk-aware resource orchestration, multi-agent reinforcement learning for slice management and LLM-driven adaptive control in O-RAN environments. The overarching objective is to establish a principled connection between threat modeling, risk quantification and control design, enabling the development of networked systems that can anticipate, withstand and adapt to evolving adversarial conditions.
