Deep learning-driven mobile crowd sensing for secure and efficient Hajj pilgrimage management

The annual Hajj pilgrimage attracts millions of people, necessitating effective crowd management strategies to ensure safety and enhance the experience for all participants. With the advent of digital technologies, machine learning techniques have become pivotal in analysing crowd behaviour and managing resources in real-time. This study explores the integration of deep learning methods in mobile crowd sensing frameworks, specifically within an IOTA-based decentralised environment, to address challenges such as data authenticity, privacy protection, and real-time anomaly detection. Using a logit-boosted convolutional neural network model, this research achieved an accuracy of 99.5% in identifying anomalous events, outperforming traditional machine learning models. The results demonstrate the model's capacity to provide a robust framework for detecting potential threats and ensuring secure, efficient information exchange during large-scale events. These findings underscore the potential of deep learning-enhanced crowd sensing in transforming the management of high-density gatherings. This approach not only enhances safety but also optimises resource allocation based on crowd density predictions. Future applications of this framework could extend to other large-scale gatherings, offering scalable solutions for various crowd management scenarios.