Android Malware Detection Using Hybrid Meta-heuristic Feature Selection and Ensemble Learning Techniques

There is wide use of smart mobile phone in modern digital world which is generally operated using open-source software. Being open-source software, it becomes easier to intrude in the system using malicious code. Android malware gets installed into the smart mobile phone without the permission of user and causes harm to user’s personal and sensitive information. To detect this malware, various techniques are proposed by researchers. Existing malware detection techniques uses digital signature method which is unable to recognize unknown malware. Thus, this paper has a novel malware framework in which dynamic feature is exploited to detect android malware. In the proposed framework, we aim to select right subset of feature which can increase our performance. In the proposed framework, meta-heuristic feature selection (FS) method using Genetic Algorithm (GA), Gravitational Search Algorithm (GSA) and correlation is used which is named as Correlated Genetic Gravitational Search Algorithm (CGGSA). The optimized features are used by the Adaptive boosting and Extreme Gradient Boosting Classifiers to detect the malware. Performance analysis of the proposed framework is evaluated using real-time CICMalDroid-2020 dataset in terms of accuracy, precision, recall and f1-score. The proposed framework has achieved 95.3% of accuracy.