Enhancing Attack Detection in Android Environment Through Software Based on Machine Learning Methods

With the growing use of mobile devices, it is projected that nearly 70% of mobile phone users own an Android smartphone. Due to the completely open-source nature of Android, the Android operating system is vulnerable to various attacks. The smartphone’s data has a sensitive nature, making it important to protect from these attacks. Machine learning (ML) approaches have proven to be an efficient methods of detecting these assaults since they can create a classifier from a set of training instances; therefore, it does not need an existing database of harmful signatures. This review paper aims to cover different characteristics involved in Android attack detection systems, such as the Android operating system environment, feature extraction, feature selection, performance measures, supervised, and unsupervised models. They are described based on previous works that employ machine learning for detecting Android malware. Furthermore, this paper intends to help researchers gain expertise in the subject of Android attack detection.