A Cross Version Binary Code Similarity Detection Based on Deep Learning Approach

Abstract

According to a recent study, the prevalence of dangerous software, or malware, is increasing at an alarming rate. Some malware can hide itself in the system by using various obfuscation techniques. To safeguard computer systems and the Internet against malware, it must be identified before infecting a large number of computers. Recently, there have been numerous studies on malware detection technologies. Nevertheless, malware detection remains a hurdle. In contrast, Open Source Software (OSS) code is commonly reused in software development. However, reusing some specific OSS versions results in one-day vulnerabilities, the details of which are made public. These vulnerabilities could be abused and result in major- security problems. The most advanced OSS reuse detection methods now in use struggle to pinpoint the precise versions of OSS that are being reused. The matching scores are only based on resemblance, and the criteria they chose are not distinctive enough for version detection. Current techniques, such as behaviour-based, model checking-based, and deep learning-based approaches, work effectively for complex and unknown malware and can be used to detect some known and unknown malware. However, no approach can detect every piece of malware out there. This emphasizes how difficult it is to establish an efficient malware detection tool, as well as the potential for fresh research and approaches. Cross Version Binary Code Similarity Detector (SIMCODE-NET) is a well-optimized version Detection Method for Open Source Software (OSS) in commercial off-the-shelf (COTS) software, which is presented in this work. First, we go over five different types of version-specific programming elements that can be tracked in instances of both source and binary code. Based on the two levels of code features, we propose a two-stage version identification approach and classify these features into program-level and function-level features. In addition, SIMCODE-NET classifies various forms of OSS versions...