Advancing Mobile Sensor Data Authentication: Application of Deep Machine Learning Models
Tanvir Ahmed, Sydul Arefin, Rezwanul Parvez, Fariha Jahin, FNU Sumaiya, Munjur Hasan · 2024
The authentication of sensor data is a must-need when we talk about the domain of mobile security. This paper explores the efficacy of deep learning models known as Convolutional Neural Network (CNN), Long Short-Term Memory Network (LSTM), and Transformer by analyzing a comprehensive mobile sensor dataset. While different models demonstrate considerable accuracy—CNN at 81.51% and LSTM at 85.69%—the Transformer model lags slightly at 77.69%. To address these disparities and further advance the state of the art, we introduce a novel deep-learning model specifically architected for mobile sensor data. This proposed model not only captures the temporal and spatial dependencies inherent in sensor data more effectively but also achieves a notable accuracy of 87.14%. Our results indicate that the proposed model offers a substantial improvement in mobile sensor data authentication, paving the way for more secure mobile computing environments.