<p>The Controller Area Network (CAN) plays a vital role in modern vehicles by facilitating real-time communication between key subsystems. Despite its widespread use, the association between raw CAN messages and actual vehicle signals remains undisclosed by Original Equipment Manufacturers (OEMs). These mappings differ significantly across brands, models, and production years, making it difficult to accurately interpret vehicle behavior from CAN data. This lack of standardization and openness creates major obstacles for research in fields such as automotive cybersecurity, data privacy, and fault analysis. This paper introduces a reverse engineering methodology to capture and preprocess CAN traffic, followed by the classification of CAN identifiers into types such as binary, CRC, counter, and constant signals. A similarity matrix is then used to cluster related CAN IDs, with the approach validated on datasets collected from multiple vehicle manufacturers. Key contributions include successful Electronic Control Unit (ECU) identification, consistent grouping of functionally similar signals across vehicles, and insights into signal behavior under different operational states–particularly notable changes in brake signal characteristics between idle and motion. The findings offer valuable tools for advancing in-vehicle network analysis and strengthening automotive security research.</p>