A Data-driven Approach for Reverse Engineering Electric Power Protocols

Electric power protocol is a typical kind of industrial protocols, and is widely-used in electric power systems. Since most electric power protocols are private and have no public protocol specification, it poses a great challenge for security analysis and vulnerability discovery. Protocol reverse engineering makes it possible to analyze unknown or private protocols. However, previous reverse engineering methods which are proposed to analyze private protocols are not suitable for reversing engineering electric power protocols, because electric power protocols have many unique features and have more compact structures. To address this issue, we present a novel data-driven approach to infer the fields of electric power protocols. The approach leverages clustering technique to reverse-engineer the structure information of electric power protocols and a new metric is proposed to measure the distance between adjacent fields and merge fields recurrently. We use Precision, Recall and F1-measure as the evaluation metrics. Results show that our methods can infer most protocol fields of three commonly-used electric power protocols correctly. We also compare our approach with some state-of-the-art approaches, and results show that our approach performs better.