AI-Driven Cyber-attack Detection and Mitigation Algorithm for the MG-FARM Project

Abstract

The integration of Internet of Things (IoT), Artificial Intelligence (AI), and renewable-energy micro-grids within
the MG-FARM platform offers a transformative approach to enhancing agricultural productivity, energy
efficiency, and sustainability in remote regions. While this digital-agricultural convergence enables optimized
management of energy, water, and crop systems, it also increases exposure to cyber threats capable of disrupting
critical operations such as irrigation scheduling, water pumping, and micro-grid control. Addressing these
challenges, this paper introduces a novel AI-driven intrusion detection and mitigation algorithm specifically
designed for MG-FARM. The proposed solution combines machine learning-based anomaly detection with
resilient control strategies, incorporating real-time sensor data validation, network traffic monitoring, and
adaptive response mechanisms. These features ensure service continuity and operational stability even under
malicious attack scenarios. Extensive simulations and case studies validate the approach, demonstrating its
capacity to detect and neutralize false data injection (FDI) and denial-of-service (DoS) attacks. Results indicate a
32% improvement in system resilience compared to conventional rule-based methods, underscoring the critical
role of ontology-aware, AI-enhanced cyber security frameworks in securing next-generation smart farming
infrastructures.