Enhancing A New Classification for IDN Homograph Attack Detection

Web security issues have become more critical for web users because the development of information technologies makes many web threats much sophisticated. Internationalized Domain Name (IDN) homograph attack is one of the common threats which deceives web users by fabricating homologous domain names of their accessed websites by exploiting the fact that many characters look alike, for instance, go0gle.com is a homograph targeting to the brand domain google.com by replacing the third character `o' to zero. Since IDN homograph attacks can lead users to web phishing or privacy interception, its detection has been an important challenge. In this paper, we propose an IDN homograph detection method by multi-feature classification based on Mean Squared Error (MSE), Peak Signal-to-noise Ratio (PSNR), and Structural Similarity Index (SSIM). Compared to the state-of-art existing classification method, our accuracy is increased from 95.07% to 97.04%, and false positive rate (FPR) is reduced from 3.92% to 2.31%.