### Abstract: Enhancing Smart Home Security with AIoT and WiFi In recent years, the fusion of Artificial Intelligence (AI) and the Internet of Things (IoT) has given rise to the Artificial Intelligence of Things (AIoT), a technology that combines the data collection and processing capabilities of IoT with the decision-making prowess of AI. Unlike traditional IoT setups that primarily focus on data transfer and remote processing, AIoT devices process data locally and in real-time, enabling them to make intelligent decisions. This technology has seen significant applications in various sectors, including intelligent manufacturing, health care monitoring, and smart home security. A critical component of smart home AIoT technology is accurate human activity recognition (HAR), which allows smart devices to identify and respond to different user activities. For instance, HAR can help adjust lighting or switch music based on whether a user is cooking or exercising, enhancing user experience and ensuring energy efficiency. Among the various methods of HAR, WiFi-based motion recognition stands out due to its ubiquity, privacy, and cost-effectiveness. In a groundbreaking study published in the IEEE Internet of Things Journal, a team of researchers led by Professor Gwanggil Jeon from the College of Information Technology at Incheon National University, South Korea, has developed a new AIoT framework called the Multiple Spectrogram Fusion Network (MSF-Net) for WiFi-based human activity recognition. This framework aims to address the issue of unstable performance in WiFi-based recognition, often caused by environmental interference. MSF-Net is a robust deep learning framework that utilizes channel state information (CSI) for both coarse and fine activity recognition. The framework consists of three main components: 1. **Dual-Stream Structure**: This structure employs short-time Fourier transform (STFT) and discrete wavelet transform (DWT) to pinpoint abnormal information in CSI, enhancing the detection of subtle movements. 2. **Transformer**: The transformer module efficiently extracts high-level features from the data, improving the accuracy of activity recognition. 3. **Attention-Based Fusion Branch**: This component boosts cross-model fusion, integrating the insights from both STFT and DWT to provide a more comprehensive and accurate recognition of activities. The researchers conducted extensive experiments to validate the performance of MSF-Net. They tested the framework on four datasets: SignFi, Widar3.0, UT-HAR, and NTU-HAR. The results were impressive, with MSF-Net achieving Cohen's Kappa scores of 91.82%, 69.76%, 85.91%, and 75.66% on these datasets, respectively. These scores indicate that MSF-Net outperforms existing state-of-the-art techniques in WiFi data-based coarse and fine activity recognition. Professor Jeon highlighted the potential applications of MSF-Net, stating, "The multimodal frequency fusion technique has significantly improved accuracy and efficiency compared to existing technologies, increasing the possibility of practical applications. This research can be used in various fields such as smart homes, rehabilitation medicine, and care for the elderly. For example, it can prevent falls by analyzing the user's movements and contribute to improving the quality of life by establishing a non-face-to-face health monitoring system." The development of MSF-Net is expected to revolutionize smart home security and convenience. By enabling more accurate and reliable activity recognition, this technology can enhance safety measures, such as fall detection, and improve the overall user experience through automated adjustments to home settings. Additionally, the framework's potential in health care and elderly care underscores its broader impact, contributing to the well-being and independence of users. Overall, the convergence of IoT and AI through WiFi-based activity recognition, as proposed in this research, holds the promise of significantly enhancing people's lives through everyday convenience and safety. The robust performance of MSF-Net and its potential applications in various fields make it a notable advancement in the realm of AIoT technology.