The Secure Multi-Execution (SME) framework introduces a novel approach to enhancing C++ application security by dynamically detecting and addressing vulnerabilities. By utilizing a unique method of executing multiple instances of a program at different security levels, SME ensures that outputs conform to their respective security classifications and also effectively isolates and mitigating security risks. In the report, it outlines the framework's design, implementation, and its efficacy in identifying common security flaws such as buffer overflows, insecure data handling, and concurrency issues through comprehensive testing on vulnerable C++ code. The results demonstrate in the report shows SME's potential to significantly improve software security by enforcing strict non-interference and robust data flow management across various operational contexts.
This project was further modified and implemented LLM models for better vulnerability detection.