OMARF: An Orchestrated Multi-Agent Resilience Framework for Autonomic Software-Defined Offshore Wind Industrial Networks

Abstract

Software-defined industrial networks face growing challenges from component failures, network congestion, thermal instability, and cyberattacks, requiring resilience mechanisms that can respond and adapt in real time. This paper presents the Orchestrated Multi-Agent Resilience Framework (OMARF), an autonomic framework that coordinates self-healing and self-defense agents across the control, data, and knowledge planes of software-defined industrial networks. Unlike conventional approaches that treat resilience and security separately, OMARF integrates cloud-native controller recovery, reinforcement-learning-based traffic adaptation, and adaptive cyber defense within a unified orchestration framework. The framework continuously monitors multidimensional network states, detects policy violations, and activates the most appropriate mitigation agents in real time. Simulations conducted in an offshore wind power plant network demonstrate that OMARF reduces path latency by up to 28.3% while improving throughput across diverse fault and attack scenarios. In degraded operating conditions, the framework maintained stable adaptive behavior, highlighting its potential to improve the reliability, scalability, and cyber resilience of next-generation industrial software-defined networks.