ScanSavant: Malware Detection for Android Applications with Explainable AI

Abstract

Mobile devices face SQL injection, malware, and web-based threats. Current solutions lack real-time detection. This paper introduces an Android app with advanced algorithms for real-time threat scanning. During testing, our application detected 94% of SQL injection attempts, outperforming the 86% average detection rate in similar studies. For malware analysis, it achieved a 97% detection accuracy on a dataset of infected files, higher than the industry standard of 93%. Additionally, our app can detect 85 malware variants and assign 15 attributes (Trojan.Gen.8, Worm.Autorun, Adware.Elex, Spyware.Zbot, Ransom.Cryptolocker, Rootkit.ZeroAccess, Exploit.CVE-2017-0143, Virus.MSIL.CoinMiner, Trojan.Emotet, Backdoor. DarkComet, PUP.Optional.Conduit, Adware.MyWebSearch, Virus.Win32.Sality, Trojan.Win32. Necurs, and Ransom.WannaCry) to some malwares, providing detailed analysis for better threat management. The application effectively scans both EXE and APK files, ensuring com-prehensive protection. When assessing website links, the application identified security risks with 96% accuracy, demonstrating its capability in managing web-based threats. This app detects SQL injections, analyses malware, and assesses website security, bolstering cyber defence with user-friendly features and top-notch threat mitigation.