A new deep learning algorithm developed by the National Institute of Standards and Technology (NIST) and recently published in IEEE Access harnesses the abilities of Wi-Fi routers for a new use: analyzing a person’s breathing and detecting when someone is struggling to breathe.
The BreatheSmart learning algorithm makes use of “channel state information,” or CSI—a set of signals that a device, like a cell phone or laptop, sends to a Wi-Fi router. CSI signals are distorted as they travel through a room when they bounce off of objects in their environment. Ordinarily, the Wi-Fi router simply analyzes the distortion in order to adjust and optimize the link; with the BreatheSmart system, CSI signals are examined more closely to monitor even minor changes to the environment—like an unusual breathing pattern of someone in the room. If a person’s chest is moving differently due to wheezing or coughing, the system is able to detect this.
Wi-Fi technology is at the center of this innovation, and many standards have guided Wi-Fi since the early days of its development to support its reliability and interoperability. IEEE 802.11, Information Technology – Telecommunications and Information Exchange Between Systems – Local and Metropolitan Area Networks – Specific Requirements, has long provided the basis for the architecture and specifications necessary for Wi-Fi systems. This standard was developed by IEEE, a member and accredited standards developer of the American National Standards Institute (ANSI).
In order to develop the technology, NIST scientists conducted experiments using a manikin that can simulate different breathing scenarios, situated in an anechoic chamber with commercial, off-the-shelf Wi-Fi equipment. Anechoic chambers stop reflections of sound or electromagnetic waves, offering “field-free” conditions for experiments, and their effective conditions and use are guided by standards such as ANSI/ASA S12.55-2012/ISO 3745:2012 (R2019), Acoustics – Determination of Sound Power Levels and Sound Energy Levels of Noise Sources Using Sound Pressure – Precision Methods for Anechoic Rooms and Hemi-Anechoic Room. This international standard was originally developed by the International Organization for Standardization (ISO) Technical Committee (TC) 43, Acoustics, Subcommittee 1, Noise. The Acoustical Society of America (ASA), an ANSI member and accredited standards developer, is the ANSI-accredited Technical Advisory Group (TAG) administrator to this TC and SC. The document has been nationally adopted by ASA as an American National Standard.
After the signals are received, the BreatheSmart system analyzes them to find unusual patterns. By simulating numerous respiratory scenarios in the experiment, scientists made use of deep learning—a type of machine learning that allows the system to “learn” as it receives more data, and improve its ability to recognize respiratory patterns. One standard that guides machine learning is ISO/IEC 23053, Framework for Artificial Intelligence (AI) Systems Using Machine Learning (ML), which describes the system components and their functions in the AI ecosystem using ML technology. It was developed by ISO/International Electrotechnical Commission (ISO/IEC) Joint Technical Committee (JTC) 1, Information technology, Subcommittee (SC) 42, Artificial intelligence. ANSI holds the Secretariat of JTC 1 and SC 42.
NIST scientists built on previous research that explored the use of Wi-Fi signals to sense people or movement, as part of ongoing efforts to help doctors fight the COVID-19 pandemic. The process presented in this experiment lays out how app and software developers may be able to create programs to remotely monitor breathing.
Learn more about BreatheSmart in the NIST article: Wi-Fi Could Help Identify When You’re Struggling to Breathe.