A Method for Classifying Polymorphic Viruses Based on Fuzzy Logical Inference

A paper presents a comprehensive study on the modeling and classification of polymorphic viruses based on their structural complexity and levels of polymorphism. Six distinct classes of polymorphic viruses are developed, ranging from basic forms that utilize constant values in decryptors to advanced forms employing highly sophisticated code obfuscation and internal code mixing techniques. A novel fuzzy classification method is proposed, which includes the identification of virus characteristics, construction of a logical inference tree, definition of linguistic variables, assignment of membership functions, and formation of a fuzzy inference system knowledge base. This approach enables probabilistic classification of polymorphic viruses across six complexity levels and allows for nuanced assessments using fuzzy linguistic terms such as low, below average, average, above average, and high. The proposed methodology enhances the accuracy and reliability of classification outcomes, supporting the development of more effective countermeasures against polymorphic malware threats.