Behavior Analysis-Based IoT Services For Crowd Management.

This article focuses on a behavior analysis-based Internet of Things (IoT) services architecture designed to improve crowd management, particularly in highly crowded locations. The proposed approach segments video footage from surveillance cameras into spatio-temporal flow-blocks, classifying each as normal or abnormal using a generative Hidden Markov Model (HMM) to analyze crowd behavior at both individual and global levels. A real-world case study involving Muslim pilgrimages (Hajj and Umrah) in Saudi Arabia was used to validate the system, demonstrating high accuracy (up to 97.3%) in recognizing crowd behaviors in real time. The architecture integrates multiple layers—including IoT services, mobile networks, crowd management, and cloud services—to support decision-making and timely invocation of IoT services such as smart gates and emergency responses. Results indicate that this method compares favorably with existing literature, with future plans to deploy camera networks and IoT services on-site for further experimentation.