Android Malware Detection Using the Combination of Different Static Features.
Year: 2025
Type: dissertation
Abstract: ABSTRACT Recent smartphones offer a large number of services and with the increase of ser- vices security threat also increases. Use of personal data is not secure due to malware. Malware hacker can steal user personal information like messages, con-tacts, phone number, email accounts etc., and use this information for different criminal activities. Malware is basically a piece of code that is built with harmful intention. Malware attack the smartphone by using Android applications. Now- a- days large number of Android applications has been developed but a user does not sure about applications whether these applications are secure or not. For thedetection of malware many different approaches used some author use static and some use dynamic malware detections techniques. Both static and dynamic mal- ware detection approaches have some advantages and some limitations. In our research, we used static malware detection approach for the detection of malware. We aim to compare the performance of static analysis techniques by using three categories of features (permission, intents, and API call) for the detection of mal- ware. Although static malware detection has already been used but there is lack of research done that analyses all features of apk files on the single dataset. This approach examines the permissions tags and intent filter within the manifest file. Moreover, we analyze the source code of the Android application; this approach examines the Java code for Android applications. Our experiment results show remarkable results when we used info gain feature selection algorithm. We have obtained 97% accuracy rate in case of intents by using Decision tree. We have observed that overall performance of Decision Tree is good in case of info gain feature selection algorithm. Decision tree shows the accuracy value of 97% in caseof intents and 93% in case of permission and 95% for the combined feature set and 85% for API Calls by using info gain algorithm. The results showed that the individual feature set of API Calls are not good for Android malware detection.
Author: Sadak, Abdi Osman
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6949731544
Cites:
Cited by:
Related to: