This study designs and implements an IoT-based fire alarm activation system that enhances accessibility, particularly for vulnerable populations in high-density residential environments. In modern apartment complexes, early fire response is critical d...
This study designs and implements an IoT-based fire alarm activation system that enhances accessibility, particularly for vulnerable populations in high-density residential environments. In modern apartment complexes, early fire response is critical due to rapid smoke spread and limited evacuation time, yet conventional manual fire alarm devices require physical force and proximity, making them difficult for children, the elderly, or mobility-impaired individuals to operate. To address this limitation, we propose a smart fire-alarm architecture integrating an AI vision sensor (HuskyLens), a deterministic preprocessing module based on the Arduino Uno R3, and an ESP8266 Wi-Fi communication module. The HuskyLens sensor performs real-time flame detection using an on-device CNN inference model, while the Arduino collects and preprocesses event data with deterministic timing. Fire event messages are transmitted through an MQTT lightweight messaging protocol and immediately delivered to a remote alarm server or user devices via existing Wi-Fi infrastructure. The implemented system was evaluated in terms of flame recognition accuracy, alarm activation latency, and wireless transmission reliability. Results show improved response speed and reduced false alarms compared to conventional manual activation methods, while significantly enhancing accessibility by enabling vulnerable users to trigger alarms with minimal physical interaction. The significance of this work lies in demonstrating that IoT, embedded AI, and wireless communication technologies can be effectively applied to fire safety domains, contributing to the development of next-generation smart fire alarm and emergency response systems.