Energy-Efficient Deep Learning Models for IoT-Based Smart Home Automation
doi.org/10.63339/FME20250905
Keywords:
Internet of Things (IoT), Smart Home Automation, Deep Learning, Energy Efficiency, Model CompressionAbstract
With the approval of the Internet of Things (IoT) and deep learning, smart home automation is changing quickly. Nevertheless, deep learning models are usually expensive and energy-consuming, which restricts their application on low-power IoT devices. This studysuggests a deep learning-based energy-efficient system to be used in a smart home. The solution suggested combines the lightweight deep learning models of MobileNetV3 and EfficientNet-Lite along with model compression techniques of quantization and pruning. Experimental assessment shows that the energy usage decreased by 38 percent with 93.1 percent accuracy in real-time home activity identification and automation of device control. The findings validate the belief that optimized deep learning models are able to resultinto high performance and at the same time can be practical in smart home applications by usage of the IoT.References
1] K. Kim, S. Park, and T. Lee, “Gesture-based smart home control using convolutional neural networks,” IEEE Internet of Things Journal, vol. 8, no. 14, pp. 11240–11249, 2021.
[2] S. Patel and A. Gupta, “Occupancy prediction in smart homes using long short-term memory networks,” Applied Soft Computing, vol. 118, pp. 108472, 2022.
[3] J. Chen, M. Zhao, and N. Li, “Transfer learning-driven voice command recognition for smart homes,” Sensors, vol. 23, no. 4, pp. 1–14, 2023.
[4] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “MobileNetV3: Searching for mobile architecture,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2019, pp. 1314–1324.
[5] M. Tan and Q. Le, “EfficientNet: Rethinking model scaling for convolutional neural networks,” in Int. Conf. Machine Learning (ICML), 2020, pp. 6105–6114.
[6] S. Han, H. Mao, and W. J. Dally, “Deep compression: Compressing deep neural networks with pruning, trained quantization, and Huffman coding,” in Proc. Int. Conf. Learning Representations (ICLR), 2016.
[7] G. Hinton, O. Vinyals, and J. Dean, “Distilling the knowledge in a neural network,” arXiv preprint arXiv:1503.02531, 2015.
[8] R. Singh and P. Sharma, “Energy-aware AI models for embedded IoT systems: A comparative study,” IEEE Access, vol. 11, pp. 84520–84533, 2023.
[9] A. Kumar, S. Rathore, and M. A. Javed, “Edge-based smart home automation using multimodal deep learning,” Journal of Ambient Intelligence and Smart Environments, vol. 15, no. 2, pp. 155–168, 2023.
[10] F. Al-Saidi and H. Al-Madani, “Power optimization techniques for Raspberry Pi-based embedded systems,” International Journal of Computer Applications, vol. 182, no. 31, pp. 22–29, 2021.
