HAMS: A Hybrid Approach to Malware Detection on Smartphones
We aim at developing a novel methodology to design and implement secure mobile devices by offering a resource-optimized approach that combines efficient malware detection on the device with high precision detection algorithms on cloud servers. Methods currently used in the field of failure prediction, such as feature extraction and selection, will be investigated.
The HAMS project is conceived as preparatory for a larger project. Therefore, in HAMS we propose to perform the first steps of this larger project that are: the creation of a simulation infrastructure for malware and the classification of malware in families (i.e., groups of malware with similar behavior). Collected malware samples will be run in a simulation environment and related features, both at application and at operating system level, will be extracted. The features will be analyzed, by using statistical methods, to study their correlation with a malicious behavior. Relevant features will be used for classifying malware samples.