The Future of WiFi Sensing AI

Since its inception in the late 1990s, Wi-Fi has had a single primary purpose: transmitting data. Every generation of Wi-Fi (from 802.11b to the latest Wi-Fi 7) has focused on increasing speed, reducing latency, and expanding bandwidth.

However, we are currently witnessing a major shift in how wireless networks are designed. Industry leaders and researchers have realized that WiFi signals can do much more than carry bits. They can also sense the physical world.

In the near future, your home router will not just connect your phone to the internet; it will also act as a central spatial radar, monitoring security, tracking occupancy, and checking the health of occupants. In this article, we will explore the upcoming IEEE 802.11bf standard, the rise of TinyML, and how open-source projects like RuView are leading this space.

Currently, WiFi sensing is highly fragmented. Different companies and open-source projects use custom firmware modifications and proprietary algorithms to extract and process CSI. This makes interoperability between devices from different manufacturers impossible.

To resolve this, the Institute of Electrical and Electronics Engineers (IEEE) is developing a new amendment: **IEEE 802.11bf**. This amendment formally defines WiFi sensing as a core, first-class capability of the Wi-Fi standard.

Under 802.11bf, Wi-Fi devices will support standard protocols to:

• **Request Sensing Sessions**: A smart home gateway can request a nearby smart plug to begin a sensing session on a specific frequency.
• **Standardize CSI Formats**: All hardware manufacturers (Broadcom, Qualcomm, Intel, Espressif) will output CSI data in a uniform format, making software development much simpler.
• **Optimize Channel Usage**: The standard defines mechanisms to share the radio spectrum between data transmission and sensing, preventing network slowdowns.

When 802.11bf is fully released, WiFi sensing will become standard on all consumer routers, opening up massive opportunities for software projects like RuView. To see the hardware you can use today, check out Best ESP32 Boards for CSI Projects.

Another major trend shaping the future of WiFi sensing is TinyML—the deployment of machine learning models on low-power microcontrollers.

Currently, RuView streams CSI data to an edge gateway for processing. While this works well, the future lies in running the entire neural network locally on the sensor node itself.

The latest generation of chips, such as the ESP32-S3 and future RISC-V microcontrollers, include dedicated Neural Processing Units (NPUs) and hardware accelerators. By optimizing our models using quantization and pruning, we can run inference directly on the chip using less than 100 Kilobytes of RAM. This eliminates the need for an external gateway, reducing latency and hardware costs even further.

To learn how this signal processing pipeline is structured, read How WiFi CSI Sensing Works.

As WiFi sensing moves into the mainstream, big tech companies are building proprietary, closed-source ecosystems. These commercial systems often upload raw spatial data to the cloud, raising significant privacy concerns and lock-in issues.

Open-source projects like **RuView** are essential to protecting user privacy. By building a transparent, community-audited system that runs locally at the edge, RuView ensures that users maintain complete control over their physical data.

Furthermore, the open-source community drives rapid innovation. Developers can adapt the code for custom research, integrate new hardware targets, and build custom home automations that commercial companies would never support.

Explore all the features RuView supports on our Features Page or dive into the implementation on the How It Works Page.

The transition from traditional, intrusive cameras to ambient WiFi sensing is inevitable. By getting involved with RuView today, you are learning the skills and building the tools that will define the next decade of spatial intelligence.

We invite you to join our community, flash your first ESP32 board, and contribute to the future of privacy-friendly sensing on GitHub. If you have questions, view our FAQ Page or get in Contact.