Detection of Java']JavaScript of Malware with un-readability using Mahalanobis-distance

Increasing of JavaScripts of malware requires automatic malware detection systems in these days. Our research takes note of the appearence probability and the state transition probability of first order Markov source for both of malware JavaScript and other JavaScript. The pre-experiments found statistical significance. We propose the malware detection method using Mahalanobis-distance with the probability variable of the appearance probability of characters and the probability variable of state transition probability of first order Markov source. As the results of experiments, the method of combining two probability variables proved the effectiveness compared to the method using single probability variable.