Penerapan Edge AI untuk Smart Home Deteksi Aktivitas Penghuni Berbasis IoT
DOI:
10.33395/jmp.v14i2.15658Keywords:
Internet of Things, Edge Computing, Edge AI, Smart Home, Activity RecognitionAbstract
Perkembangan Internet of Things (IoT) dan kemampuan komputasi di tepi jaringan (edge computing) membuka peluang untuk membangun sistem smart home yang responsif, hemat bandwith, dan lebih menjaga privasi (Atzori et al., 2010). Penelitian ini merancang dan menguji sistem deteksi aktivitas penghuni berbasis IoT yang melakukan inferensi di edge node (Raspberry Pi 4) menggunakan model lightweight CNN (TensorFlow Lite) untuk data visual dan Random Forest untuk data sensor non-visual(Google, 2019; Warden & Situnayake, 2019). Data simulasi ilmiah disusun dari 5.000 gambar (ESP32-CAM) dan 20.000 sampel sensor (PIR, accelerometer). Pengujian menunjukkan bahwa Edge AI menurunkan latensi rata-rata dari 510 ms (cloud) menjadi 178 ms (edge) dan ensemble model mencapai akurasi ~94.4% pada lima kelas aktivitas utama. Hasil ini menunjukkan Edge AI layak diimplementasikan untuk use-case smart home yang memerlukan respon cepat dan privasi(Huang & Li, 2021; Shi et al., 2016).
References
Atzori, L., Iera, A., & Morabito, G. (2010). The Internet of Things: A survey. Computer Networks, 54(15), 2787–2805.
Chen, M., & Hao, Y. (2021). A survey on edge AI: Concepts, architectures, and applications. Journal of Systems Architecture.
Espressif Systems. (2017). ESP32-CAM Technical Datasheet.
Google. (2019). TensorFlow Lite Documentation.
Howard, A. G. (2017). MobileNets: Efficient convolutional neural networks for mobile vision applications. https://arxiv.org/abs/1704.04861
Huang, X., & Li, Y. (2021). Edge AI for next-generation IoT applications. IEEE IoT Magazine, 4(3), 22–30.
InvenSense, T. D. K. (2015). MPU-6050 Product Specification. TDK InvenSense. https://invensense.tdk.com/products/motion-tracking/6-axis/mpu-6050/
Lane, N. D., Bhattacharya, S., & Mathur, A. (2015). DeepX: A software accelerator for low-power deep learning on mobile devices.
Liu, J., & Chang, Y. (2020). Activity recognition for smart homes using IoT sensors. Sensors, 20(12).
Patel, S., Park, H., & Bonato, P. (2016). A review of wearable sensors and systems for rehabilitation. Journal of NeuroEngineering and Rehabilitation, 13.
Rahman, M., & Zhou, Y. (2020). Privacy issues in IoT: A survey. Journal of Internet Services and Applications, 11.
Redmon, J., & Farhadi, A. (2018). YOLOv3: An incremental improvement. https://arxiv.org/abs/1804.02767
Shi, W., Cao, J., Zhang, Q., Li, Y., & Xu, L. (2016). Edge computing: Vision and challenges. IEEE Internet of Things Journal, 3(5), 637–646.
Sucipto, A., & Nugroho, H. (2021). Implementasi smart home berbasis IoT dengan Raspberry Pi. Jurnal Teknologi & Informasi.
Tang, S. (2018). Real-time vision inference on edge devices. International Journal of Computer Vision Applications.
Wang, J., Chen, Y., Hao, S., Peng, X., & Hu, L. (2019). Deep learning for sensor-based activity recognition: A survey. Pattern Recognition Letters, 119, 3–11.
Warden, P., & Situnayake, D. (2019). TinyML: Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers. O’Reilly Media.
Zhang, K., & Zhu, Y. (2019). Edge intelligence: Paving the last mile of artificial intelligence with edge computing. Proceedings of the IEEE.
Zhao, L., & Chen, M. (2022). Low-latency AI inference at the edge. IEEE Access, 10, 3340–3351.
Zhou, Z., & Chen, X. (2020). Human activity recognition using deep learning: A review. IEEE Sensors Journal.
Downloads
How to Cite
Issue
Section
License
Copyright (c) 2025 Jimmy Jimmy, Johan Johan, Albert Suwandhi, Benny Benny
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
