Security in Mobile Human-Device Interaction: Leveraging Mobile Keypad Input Patterns for Secure User Recognition and Fraud Detection

This study proposes a novel approach to enhancing mobile security by identifying key parameters of virtual keypad input for user authentication. In particular, the intention is to develop a behavior-based recognition system that uses user behavior during keypad interactions to minimize fraudulent activities. It is necessary to establish an automatic authentication mechanism to identify predictive functional correspondence based on user frailty and distinguish them from intruders. The contingency study design compared several authentication algorithms using a dataset from a company, including 792 users and 4,096 employees who left between 2016 and 2020. The metrics are an average precision of 1.0 and a recall of 1.0, as all models can detect fraud effectively. As for the CR-EPSB algorithm, the precision value was only 0.110, but the proposed TKIP-based algorithms indicated better precision values ranging from 0.516 to 0.723 and F1 scores ranging from 0.681 to 0.839, which contributed to the improvement of the authenticity rate. The outcomes of this work reveal the effectiveness of the proposed keypad-based authentication technique in improving mobile security systems and the security of the various financial transactions for businesses and individuals, as well as the security of their enterprises and personal devices. The study establishes the need to integrate behavioral biometrics as an extra security measure for portable products since other forms of identification might not be sufficient.