A GenAI-Powered Authentication Protocol for Secure Monitoring of Hajj/Umrah Pilgrims Using Cinematographic Drones
Abstract
Maintaining safety in crowded events with more than 4.5 million Hajj and Umrah pilgrims becomes a very challenging issue in terms of crowd security. While occlusion occurs in traditional closed-circuit television (CCTV) systems, drone-based crowd surveillance is still susceptible to threats like spoofing, command injection, and video replay attacks. This article proposes generative AI (GenAI)-powered authentication protocol, which provides both mutual authentication and an intelligent threat detection module based on GenAI between three entities. The proposed scheme provides security under the random oracle model (ROM), offers perfect forward secrecy, and resists a replay attack using SHA-256, xor, and concatenation. The key revocation and desynchronization attack countermeasures have been proposed and verified formally through ProVerif and Burrows–Abadi–Needham (BAN) logic proof techniques. Furthermore, the vision–language model (VLM) detects any stampede hazard, fire outbreak, or medical emergency based on the analysis of authenticated video frames with a 91.7% success rate. The computational cost for Core i7, Galaxy A07, and Raspberry Pi 4 processors at $40~{^{\circ }}$ C is 3.8, 13.1, and 10.5 ms, respectively, with 1984-bit overhead and 32.4 mJ of energy expenditure.
