PERMISSIONS RANKING WITH STATISTICAL TECHNIQUES FOR ANDROID MALWARE DETECTION

Abstract

In today’s world, smartphones play a vital role in our lives as the control of, everything we do, has been taken over by digital platforms. At present, it is impossible to live without smartphones. There is more than one operating system for smartphones amongst which the Android operating system can be considered one of the most popular ones. Android is an open-source operating system which means it is available freely. Anyone can develop their android application. Furthermore, being an open-source system cybercriminals also use it to develop their malicious applications. That is why Android has become a key for attackers to invade the privacy of users and sabotage using the same weapon. Further, many malicious applications perform venomous activities to breach the privacy of the users. So, it is mandatory to detect these infringe or malicious applications. In this paper, we have proposed two statistical-based ranking techniques by taking some references from the Spearman Ranking test and Mann-Whitney U test to rank the permissions in order to check whether the permission is significant or not. Such a ranking helps us to eliminate irrelevant permissions. Further, we apply machine learning algorithms to the dataset by eliminating the lower-ranked permissions from the dataset. The experimental results demonstrate that we obtain the highest accuracy of 99.4% and 99.25% using both ranking techniques respectively.