ANDROID MALWARE DETECTION USING GRAPHICAL TECHNIQUES

Abstract

Android is the most popular operating system for mobile devices which has dominated the smartphone industry. Traditional signature-based malware detection approaches have been in use for a long time. Yet their technology falls far short of being completely secure. Modern malware detection tools are used by major mobile application distributors, official stores, and marketplaces to analyze uploaded programs and eliminate any dangerous ones. Unfortunately, until they are taken off the market, malicious software has a long window of opportunity. In this paper, we studied different research papers based on graphical techniques and learned about static and dynamic malware detection approaches. We presented a new and unique method for detection of Android malware that uses apk function analysis and app-similarity graph in conjunction with ensemble techniques. This work uses graph neural networks and similarity graph, in which relationships are depicted as edges based on semantic similarities, while apps are represented as nodes. The study shows how well the suggested technique works to recognize malicious apps by comparing their functional and structural characteristics. The results improve the field of mobile app security and present an effective strategy of action to protect against threats that are constantly evolving within the Android app ecosystem. The proposed model provided reasonable accuracy and hence served to aid and maintain a user-safe environment.