Malicious Instructions Detection in Distribution Terminal Units - A Machine Learning Approach

With the continuous evolution of smart grids, embedded devices such as Distribution Terminal Units (DTUs) have become increasingly vital. However, it also makes DTU s an important target for attack. Among the most prevalent and critical security threats to DTU s are malicious instructions. Traditional anomaly detection techniques are ineffective due to the real-time operational requirements of DTUs. This paper presents a novel approach that utilizes machine learning techniques to identify malicious instructions by analyzing the CPU's power usage patterns. This non-invasive state monitoring approach is swift and highly accurate, achieving a detection accuracy of up to 93%.