Authors:
Alexandre Dey
;
Loic Beheshti
and
Marie-Kerguelen Sido
Affiliation:
ESIEA, France
Keyword(s):
Android, Malware, Machine Learning, Crawler, Neural Network, Static Analysis.
Abstract:
Android has, to this day, more than 80% of the mobile OS market share. Android users also have access to
more than 2 million applications via the Google Playstore. The Playstore being an official market, users tend
to trust the applications they find in it, and therefore, the store is an interesting platform to spread malware. We
want to provide a health state of this store by finding the proportion of malware that managed to get published
in it. In this paper, we explain how we developed the crawler that massively downloads the application
directly from the Playstore. Then we describe what features we extract from the applications and how we
classified them with the help of an Artfificial Neural Network. Our study confirms that there are malicious
applications on the Playstore. The proportion of them is around 2%, which corresponds to about 40,000
officially downloadable malware.