Implementasi Sistem Deteksi Intrusi Pintu Berbasis IoT dengan Sensor Getaran serta Integrasi Pemantauan Website dan Peringatan Telegram

Abstract

Keamanan fasilitas penyimpanan pasif memerlukan sistem pengawasan yang presisi, responsif, dan andal terhadap gangguan lingkungan maupun pemutusan daya. Penelitian ini mengimplementasikan sistem deteksi intrusi pintu gudang berbasis Internet of Things (IoT) dan komputasi tepi (edge computing) menggunakan ESP32-S3, sensor getaran MPU6050, magnetic reed switch, dan modul TP4056. Untuk mengurangi alarm palsu (false positive), sistem menerapkan metode Time-Windowed Thresholding yang membedakan gangguan mekanis sesaat dari pola hantaman destruktif berulang. Data anomali dikirim melalui MQTT ke website pemantauan berbasis Next.js dengan Role-Based Access Control (RBAC) serta bot Telegram untuk peringatan waktu nyata. Pengujian menunjukkan sistem berhasil mendeteksi seluruh aktivitas intrusi dalam ruang lingkup penelitian tanpa false positive maupun false negative. Sistem juga memiliki latensi end-to-end sebesar 457,0 ms (p50) hingga 799,7 ms (p95) serta mampu beralih otomatis ke baterai cadangan saat terjadi pemutusan listrik.

Passive storage facilities require security systems that are accurate, responsive, and resilient to environmental disturbances and power outages. This study implements an Internet of Things (IoT)-based warehouse door intrusion detection system using edge computing with an ESP32-S3 microcontroller, an MPU6050 vibration sensor, a magnetic reed switch, and a TP4056 power management module. To reduce false alarms, the system employs the Time-Windowed Thresholding method to distinguish transient mechanical disturbances from repeated destructive impacts. Validated anomaly data are transmitted via MQTT to a Next.js-based monitoring website with Role-Based Access Control (RBAC) and a Telegram bot for real-time notifications. Experimental results show that the system successfully detected all intrusion activities within the scope of this study without false positives or false negatives. The system achieved an end-to-end latency of 457.0 ms (p50) to 799.7 ms (p95) and automatically switched to the backup battery during power outages.