Android Malware Detection Through Application Permission: An Analysis On Feature Selection And Classification Algorithm

Mohd Shahrulazam, Samsudin (2015) Android Malware Detection Through Application Permission: An Analysis On Feature Selection And Classification Algorithm. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Android Malware Detection Through Application Permission An Analysis On Feature Selection And Classification Algorithm 24 Pages.pdf - Submitted Version Download (442kB)

Abstract

In this technology era Internet of Thing(IoT), where everything or gadget are connected to the internet with no exception of android smartphone which had become vital accessory for everyone. As the number of android user grows up every year, more user are vulnerable to malware application on their smartphone which can widespread in the network within a seconds to perform malicious activity. Regardless of the new advancement technology shown in every new model introduce every year, the malware also evolve to achieve it target. Thus, to overcome this issue this research analysis will investigate the use of different feature selection through filter method and different classification algorithm to optimize the android permission to indicate malware application. These feature selection were evaluated through performance parameter to produce the highest accuracy of classification algorithm. Over 500 android application which consist of 250 malware and benign application each were tested with different feature length, feature selection and classification algorithm. For the purpose of studying the android manifest permission, the malware can be identifying by using data mining technique. In testing and validating the parameter of malware permission, this project have 3 types of performance parameter which are accuracy, true positive rate (TPR) and true negative rate (TNR). The result to be achieve the accuracy higher than 80% and lower of true negative rate.

Actions (login required)

View Item