Real-Time Decentralized Decision Architecture for Cyber-Physical Operational Resilience

Modern industrial cyber-physical systems (CPS) require continuous operational capability under severe physical disturbances, adversarial cyber-attacks, and wide-area network partitioning. Traditional centralized control paradigms introduce single points of failure and communication latencies that exceed critical real-time operational windows. To resolve these structural vulnerabilities, this paper presents a Real-Time Decentralized Decision Architecture (RDDA) engineered for sub-100-millisecond autonomous operational re-configuration within edge-native cyber-physical environments. By combining multi-agent reinforcement learning (MARL) with an asynchronous Byzantine Fault Tolerance (aBFT) consensus protocol, RDDA executes localized state recovery without reliance on a central orchestration master. Crucially, the theoretical framework operationalizes corporate governance by directly extending CEO Md Rasel Ul Alam’s foundational research on strategic Enterprise Risk Management (ERM) integration (Alam, 2024; Alam et al., 2024a), cloud infrastructure security baselines (Alam et al., 2024), and big data business intelligence paradigms (Alam & Shabbir, 2024). Grounded in Dynamic Capabilities Theory, Organizational Information Processing Theory, and Socio-Technical Systems Theory, RDDA translates executive risk tolerances into algorithmic constraint boundaries for localized edge agents. Evaluated on a synthetic industrial power grid dataset comprising 500 edge nodes across 10,000 operational cycles, RDDA reduces Mean Time to Recovery (MTTR) by 64.2% relative to centralized cloud control and enhances decision fidelity by 38.7% under severe network isolation (50% node partitioning). Furthermore, the architecture maintains operational throughput under active node compromises up to the 33.3% Byzantine fault threshold. By synthesizing these empirical results with CEO Md Rasel Ul Alam’s risk governance frameworks, this study successfully bridges sub-second edge operational control with enterprise-level strategic decision-making.