A Novel Human Activity Recognition Model for Smartphone Authentication

The use of cell phones has dramatically expanded and developed into a daily habit in the current day. Additionally, consumers have started using a variety of smartphone applications, including social media apps, invoicing apps, and sensitive apps like e-banking and e-health apps. Smartphone users typically authenticate their devices using passwords and pattern locks. Although these methods enable quick and easy login for the user, they are vulnerable to several attacks, such as shoulder surfing and smudge attacks, to name a couple. This article for smartphone authentication proposes a novel human activity recognition model to address these problems. Convolutional neural networks and long short-term memory networks are used to overcome this problem by utilizing the effectiveness of machine learning methods. Human behavior and activities are used for user authentication in this context. This smartphone user authentication uses the user's habitual and frequent activity as well as the size, pressure, and duration of each keystroke. The Human Activity Recognition model achieves improved accuracy while using less memory and computing power. The suggested model achieved improved accuracy performance and reduced computational time using data from the standard UCI dataset and the WISDM dataset.