AI plus Wi-Fi makes smart homes smarter

A new Artificial Intelligence of Things (AIoT) framework, developed by scientists from South Korea’s Incheon National University, promises to make smart technology much more sensitive to human activity.

AIoT combines the features of artificial intelligence and Internet of Things (IoT) technologies. Unlike typical IoT setups, where devices collect and transfer data for processing at some other location, AIoT devices acquire data locally and in real time, enabling them to make smart decisions. The technology has found applications in intelligent manufacturing, smart home security and healthcare monitoring.

In smart home AIoT technology, accurate human activity recognition is crucial, helping smart devices to identify various tasks such as cooking and exercising. Using this information, the AIoT system can tweak lighting or switch music automatically, thus improving user experience while also maintaining energy efficiency. WiFi-based motion recognition is a promising candidate for this type of application, as Wi-Fi devices are ubiquitous, can ensure privacy and tend to be cost-effective.

The Incheon National University research team, led by Professor Gwanggil Jeon from the College of Information Technology, has devised an AIoT framework called multiple spectrogram fusion network (MSF-Net) for WiFi-based human activity recognition. The team’s findings have been published in the IEEE Internet of Things Journal.

“As a typical AIoT application, WiFi-based human activity recognition is becoming increasingly popular in smart homes. However, WiFi-based recognition often has unstable performance due to environmental interference. Our goal was to overcome this problem,” Jeon explained.

MSF-Net is a deep learning framework that achieves coarse as well as fine activity recognition via channel state information (CSI). It has three main components: a dual-stream structure comprising short-time Fourier transform along with discrete wavelet transform; a transformer; and an attention-based fusion branch. While the dual-stream structure pinpoints abnormal information in CSI, the transformer efficiently extracts high-level features from the data. The fusion branch then boosts cross-model fusion.

“The multimodal frequency fusion technique has significantly improved accuracy and efficiency compared to existing technologies, increasing the possibility of practical applications,” Jeon said.

“This research can be used in various fields such as smart homes, rehabilitation medicine, and care for the elderly. For instance, it can prevent falls by analysing the user’s movements and contribute to improving the quality of life by establishing a non-face-to-face health monitoring system.”

Image credit: iStock.com/onurdongel