Falling of older people living on their own has been regarded as a major public health worry that can even lead to death. But with Wi-Fi sensing, caregivers or family members can make use of the revolutionary technology by which they can potentially save the lives of the elderly without installing any additional hardware or involving any wearable sensors. The next big revolution in sensing technology is Wi-Fi sensing which is being standardized by IEEE 802.11bf Task Group and is expected to be completed by 2024.
Existing sensing technologies require a device such as a sensor or a camera that can detect an object’s physical, chemical, or biological parameters and then convert the information into an electrical signal. However, Wi-Fi sensing makes use of existing devices and Wi-Fi signals present in the Wi-Fi network, which can be used for applications such as detection (human presence and activity such as falling, walking detection), recognition (activity recognition, gesture recognition, and human/user identification/authentication) and estimation (estimation of breathing rate, heart rate, etc.). Since no specific monitoring devices such as sensors or cameras are required in Wi-Fi sensing technology, it enables a low-cost sensing solution.
Input
The raw CSI measurements are received from multiple devices and are fed into the signal processing module for noise reduction, signal transformation, and/or signal extraction.
Signal Processing & Algorithm
The raw CSI measurements contain noises and outliers that could significantly reduce Wi-Fi sensing performance. Hence, Phase offset and Outlier removal techniques are applied for noise reduction. Also, the Fast Fourier Transform technique is used for signal transformation to find the distinct dominant frequencies, which can be combined with a Low Pass Filter to remove high-frequency noises. Further, Signal extraction is used for extracting target signals from raw or pre-processed CSI measurements. It requires thresholding, filtering, or signal compression to remove unrelated or redundant signals.
After Signal processing, multiple AI/Machine Learning Algorithms are applied for more specific Wi-Fi sensing applications. Two types of algorithms are used, i.e., modeling-based and learning-based algorithms for Wi-Fi sensing. Modeling-based algorithms are based on physical theories like the Fresnel Zone model, or statistical models like the Rician fading model. Binary and multi-class classification applications usually use learning-based algorithms. These algorithms try to learn the mapping function using training samples of CSI measurements and the corresponding ground truth labels.
Applications
The signal processing techniques and AI/Machine Learning algorithms in combination result in specific Wi-Fi sensing applications are summarized below:
Major Industry Players
The below provides an overview of major players contributing to developing the Wi-Fi sensing standard (IEEE 802.11bf). LG Electronics leads the contribution, followed by Intel Corporation and Ericsson.
Patent Filling Trends
The below provides an overview of the patent filing trend of the Wi-Fi sensing standard (IEEE 802.11bf). Most of the innovation started in the years 2020 and 2021.
Below is an overview of the geographical distribution of the patents filed in the Wi-Fi sensing standard (IEEE 802.11bf). The United States of America is the primary source of innovation.
Wi-Fi Sensing Standards
The IEEE 802.11bf standard brings amendments to Medium Access Control (MAC) layer, Directional Multi-Gigabit (DMG), and enhanced DMG (EDMG) PHY layer of Wireless Local Area Network (WLAN) to enable Wi-Fi sensing operation. With these amendments, stations (Access Point or non-AP STA-like devices) can inform other stations of their Wi-Fi sensing capabilities, request and set up Wi-Fi measurements, exchange Wi-Fi sensing feedback and information, and provide a MAC service interface for layers above the MAC to request Wi-Fi sensing measurements. Wi-Fi sensing uses frequency bands between 1 GHz and 7.125 GHz and above 45 GHz. Below is the timeline of IEEE 802.11bf standard development.
The Future Challenges
Existing Wi-Fi sensing mainly focuses on humans. Future Wi-Fi sensing could be in other domains, such as detecting, recognizing, and estimating the surrounding environments, animals, and objects.
The main challenges of existing Wi-Fi sensing technologies are:
Robustness: Wi-Fi signals are susceptible to network settings, environments, objects, humans, geometry and mobility situations, etc., which makes it crucial and challenging for Wi-Fi sensing to be robust in different real-world settings.
Privacy and Security: One of the advantages of Wi-Fi sensing is that it is non-intrusive and non-obtrusive, which introduces many privacy and security issues.
Coexistence of Wi-Fi Sensing and Networking: Wi-Fi is designed for wireless communications but not sensing applications. When a Wi-Fi device is used for sensing, it could influence the network performance and be impacted by network settings.
Possible Solution
The future trends of Wi-Fi sensing technologies to overcome the above challenges are:
Cross-Layer Wi-Fi Sensing: CSI (Channel State Information) can be integrated with upper-layer information for cross-layer Wi-Fi sensing, which could help develop new sensing applications or enhance existing Wi-Fi sensing applications.
Cross-Device Wi-Fi Sensing: Some Wi-Fi-based localization and tracking applications use CSIs from multiple Wi-Fi devices. Other Wi-Fi sensing applications can combine multi-device CSIs for higher performance and efficiency.
Cross-Sensor Wi-Fi Sensing: Some sensing applications use the fusion of CSIs with other signals, such as videos and audio, as the input. Hence, for cross-sensor Wi-Fi sensing, CSIs can be combined with other sensor sources, e.g., Bluetooth, 5G, ZigBee, GPS, microphones, image/video cameras, motion sensors, etc.
Due to considerable progress in the development of sensing applications, market interest in Wi-Fi sensing technology is growing. Wi-Fi sensing standards will enable further innovation by defining key features such as interoperability, reliability, lower overhead, and new application areas.
Author
Abhijai Sahai
Assistant Vice President at Lumenci
Abhijai is a Senior Consultant at Lumenci. He holds a bachelor’s degree in Electronics and Communication from Punjab Technical University. He has extensive experience in Standard Essential Patents (SEP) work related to LTE/5G, Wi-Fi 6 and Electric Vehicle Charging technology.
