GREMC: Genetic Reverse-Engineering of Machine Code to Search Vulnerabilities in Software for Industry 4.0. Predicting the Size of the Decompiling Source Code

The article is devoted to the problem of security of cyber-physical systems as part of production according to the Industry 4.0 concept. For this purpose, the author's approach of “Genetic Reverse Engineering of Machine Code” (GREMC) is proposed. The essence of this approach lies in the use of artificial intelligence in the field of genetic algorithms to restore the source code of software executed in the form of machine code on cyber-physical devices for Industry 4.0. The resulting code can then be analysed for vulnerabilities by an expert. One issue that arises during genetic reverse engineering is predicting the size of the source code based on its machine representation (i.e., in an object or executable file). The article is devoted to solving this problem for functions in the C programming language. To do this, a method for obtaining a relationship between the sizes of the source and machine code of individual functions is described, which consists of steps such as loading a dataset with C-functions, isolating the function code in it, preprocessing them, compiling them into machine code, calculating the required sizes, building dependencies between each source code size and the corresponding machine code sizes, generating the final table and determining the dependency formula. An experiment is carried out using a software prototype that implements the method; ExeBench with 83 thousand C-functions is taken as a dataset. Justifications are given regarding the appearance of abnormal machine code sizes and their impact on the dependence formula.