Blended Ensemble Learning for Robust MITM Attack Detection and Classification in Smart Grid

The advent of smart grids and Intelligent Control Systems (ICS) has significantly enhanced efficiency and quality in power generation and distribution. Despite the large-scale benefits of these innovations, their interconnected nature exposes power systems to increasing cyber threats, including complex and persistent Man-in-the-Middle (MITM) attacks, which can inflict severe disruption in the power supply chain. To this extent, this study proposes a Blended Ensemble learning model, utilizing tree-based algorithms to detect and classify intricate MITM attacks effectively. Using the unique characteristics of its base classifiers, the model enhances performance, minimizes overfitting, and reduces latency. Evaluated on a real-world power system attack dataset, the model demonstrates superior accuracy in classifying MITM attacks and detecting subtle variants compared to traditional tree-based algorithms, highlighting its robustness and practical applicability in mitigating advanced cyberattacks.