Imagine a future where your home security system is constantly vigilant, but doesn’t require expensive installation or invasive modifications to your home. This is the vision behind research by Eyuphan Bulut, Ph.D. at Virginia Commonwealth University.
Current home security systems are expensive and require installation of new sensors in the house. Dr. Bulut proposes a low-cost alternative that leverages the existing Wi-Fi network and its signals to detect intrusions. When Wi-Fi signals bounce off objects and people in the house, they are altered. By analyzing these changes, the system can determine if a door or window has been opened, or if someone is walking around inside the house. The system can even distinguish between a package being delivered and a package being stolen from a porch.
Dr. Bulut’s research addresses a major pain point for homeowners: the high cost and complexity of traditional security systems. Over 92% of houses in the US already have a Wi-Fi network, and Dr. Bulut’s system leverages this existing infrastructure to provide a low-cost security solution.
Here’s how it works: Wi-Fi signals can penetrate walls and travel through the house, making them suitable for sensing activities even behind walls. The system extracts Channel State Information (CSI) from Wi-Fi signals, which contains information about the signal’s strength and phase. By analyzing CSI data using machine learning models, the system can identify different activities happening in the house, such as door opening, window opening, and even package stealing on the porch.
Dr. Bulut’s research has achieved high accuracy in identifying these activities, even in rooms that the system hasn’t been exposed to before. This is a significant advantage over traditional sensor-based systems, which require sensors to be installed in every location of interest.
However, there are also challenges to address. Collecting CSI data efficiently and processing it for real-time applications requires further development. The system also needs to be able to differentiate between the actions of an intruder and those of the homeowner or their pets. Dr. Bulut acknowledges these challenges and proposes future research directions to address them, including using low-rate Wi-Fi packets for sensing, developing methods to defend against adversarial attacks, and improving the system’s ability to distinguish between different types of activities.
Overall, Dr. Bulut’s research presents a promising approach for developing a low-cost, easy-to-deploy, and privacy-preserving alternative to current home security systems. By leveraging existing Wi-Fi infrastructure and utilizing machine learning techniques, this system has the potential to revolutionize home security and make it more accessible to everyone.
Watch Eyuphan Bulut Ph.D.’s talk on wireless home security here.
