Study of various approaches for malware detection in android

Static, Dynamic and Behavioral malware detection approaches

Download .zip Download .tar.gz View on GitHub

Welcome to the Malware Detection Approaches in Android by Rajesh!

Day-by-day the number of android users are increasing and people have started to store really valuable data in their ‘smartphones’. Report of International Data Corporation states that in the third quarter of 2013, 83% out of the total mobiles shipped were Android smartphones. This made android very tempting target for the attackers and hackers. Notion behind this is that even one malicious program could get them large user base to victimize. Security of Android phones has become even more difficult due to the limitations of mobile devices such as less battery, low processing power, limited memory, and size. Malware analysis involves studying applications for their malicious intent and stopping them in advance. In this report, we discuss three different approaches for malware detection in android environment and asses each of them for their strengths and weaknesses.

Our first approach is static analysis in which we study various methods and discuss classification and data mining approaches in specific. We cover signature based analysis and how classification is performed in static analysis. In our second approach, dynamic analysis, we run the application in simulated controlled environment to generate log of all activities like network messages, SMS, phone call etc. in order to detect hidden suspicious nature of the application. In general, dynamic analysis is performed over cloud due to the limitations of smartphones mentioned before. Our third approach, behavioral analysis uses different attributes to build the android environment by collecting and storing them in the form of vectors and then employing machine learning or classification algorithms, support vector machines and Naïve Bayesian techniques to analyze the applications. To read full report please click here.