An Immunity Based Computer Virus Detection Method with GA-RVNS

Most of the current virus detection approaches, such as antivirus software, require precognition of virus signatures for detection, but they are difficult to detect firstly unknown virus. A novel virus detection method inspired by immune theory with GA-RVNS (Genetic Algorithm based real-valued negative selection) is proposed. Feature vectors of program codes are mapped into high dimension real-valued space. The architecture of this model, the formal definitions of self, non-self, antigen, antibody, and gene library are given. And the process of generation of detectors by GA-RVNS in real-valued space is discussed in detail. The experimental results show that the method can detect obfuscated and firstly unknown virus more effectively than traditional model.