A Classifier of Malicious Android Applications

Malware for smartphones is rapidly spreading out. Thispaper proposes a method for detecting malware based on threemetrics, which evaluate: the occurrences of a specific subset ofsystem calls, a weighted sum of a subset of permissions that theapplication required, and a set of combinations of permissions.The experimentation carried out suggests that these metrics arepromising in detecting malware, but further improvements are neededto increase the quality of detection.