MH-ASO Based Deep NNET Classifier Scheme for Effective Android Malware Recognition & Classification Strategy

Abstract

With the emergence of smartphone technology and mobile applications, mobile phones become a most vibrant tool for accessing internet to get several services in just a single click. At the same time, susceptibilities of application are considered as a hazard for the Android device security. Because of this weakness, an attacker could hack the privacy of mobile phone data more easily. The malware application thus performs fraud actions automatically on mobile devices without the knowledge of user. Hence, these attacks are regarded as a major threat to the mobile device security. For detecting the malicious applications that are installed on Android smartphones, a work is proposed to detect Android malware using deep learning-based classification approach. At first, the input android malware dataset is considered as input and the redundant data is removed. The features are extracted and optimal features are selected using Meta-Heuristic Aquila Swarm Optimization strategy (MH-ASO). An Opti Deep Nnet classifier is employed so as to classify the malwares effectively. The proposed classifier is responsible for detecting and classifying android malwares as Adware, Scareware, SMS Malware, and Ransomware. The blowfish-based encoder-decoder is employed so as to protect the data from attackers. By this, the privacy of android device is maintained effectively. Finally, the performance analysis is carried, tested and verified over CICInvesAndMal2019 dataset and the outcomes are compared with traditional methods in terms of accuracy, F-score, precision, recall, ROC plot, and TPR (True positive rate). The analysis shows that the proposed model is effective than others.