Pengembangan Alat dan Analisis LSTM Kekuatan Sinyal RSSI Pada Smart Meter Gas Berbasis Narrowband IoT

Abstract

Pencatatan gas manual sering kali menyebabkan kesalahan dan inefisiensi. Penelitian ini bertujuan mengembangkan Sistem Smart meter gas berbasis Narrowband IoT (NB-IoT) untuk pemantauan konsumsi gas otomatis dan real-time. Penelitian ini juga menganalisis kekuatan sinyal RSSI dengan algoritma Long Short-Term Memory (LSTM). Metode meliputi perancangan perangkat keras menggunakan modul NB-IoT SIM7000C, sensor Hall effect, dan mikrokontroler Arduino Pro Mini, serta perangkat lunak berbasis Laravel untuk antarmuka website dan analisis LSTM. Pengujian dilakukan pada berbagai jarak dan kondisi lingkungan untuk mengevaluasi transmisi data dan stabilitas sinyal. Hasilnya, sistem mengirimkan data RSSI real-time (-83,67 hingga -56 dBm) sesuai standar NB-IoT. Antarmuka website menampilkan kekuatan sinyal, waktu pengiriman, dan statistik RSSI. Prediksi RSSI dengan LSTM mencapai akurasi 88,07% (MAE 3,30 dBm), mendukung stabilitas di wilayah dengan jaringan terbatas. Sistem ini stabil dan berpotensi untuk aplikasi perkotaan dan pedesaan.

Manual gas recording often leads to errors and inefficiencies. This research aims to develop a Smart meter gas system based on Narrowband Internet of Things (NB-IoT) for automated and real-time gas consumption monitoring. This research also analyzing Received Signal Strength Indicator (RSSI) using the Long Short Term Memory (LSTM) algorithm. The methodology includes designing hardware with the SIM7000C NB-IoT module, Hall-effect sensor, and Arduino Pro Mini microcontroller, alongside developing Laravel-based software for the website interface and LSTM analysis. Testing was conducted across various distances and environmental conditions to evaluate data transmission and signal stability. The system successfully transmits RSSI data in real-time (-83.67 to -56 dBm), compliant with NB-IoT standards. The website interface displays signal strength, transmission time, and RSSI statistics. LSTM-based RSSI prediction achieves 88.07% accuracy (MAE 3.30 dBm), supporting system stability in areas with limited network coverage. The system is stable and has potential for urban and rural applications.