Wireless networks are typically associated with internet access in corporate networks or entertainment services like Netflix. Yet, WiFi’s application extends far beyond just streaming data to electronics. Now that a common household owns about 10 smart devices on average, it has set up WiFi sensing to take the stage.
WiFi sensing is a type of short-range passive radar technology, and it’s surprisingly accurate. It can easily pick up an object’s movement from room to room and zero in on gestures for activity classification. For large events, a sensor could be placed at the entrance to count visitors. Hospitals and elderly care facilities can use WiFi sensors to monitor patient movement and biometric data like heartbeats, breathing, and limb movements.
Simply put, WiFi sensing measures how WiFi signals interact with movement. By pinging the environment, WiFi sensing systems can easily track locations and movement based on how the signals are reflected and deflected.
WiFi sensing systems communicate in either infrastructure mode or ad-hoc mode. In infrastructure mode, each node in the sensing system communicates with a central access point (AP). In ad-hoc mode, each of the nodes communicates with one another directly.
Topologies aside, WiFi sensing can be active or passive. An active system sends a WiFi packet dedicated to sensing purposes. Conversely, a passive WiFi sensing system appends WiFi sensing data to existing WiFi traffic.
Since a passive system doesn’t send extra packets, it requires minimal processing overhead. Although active systems need higher computational power, it also has greater control over the transmission rate, bandwidth, beamforming and other environmental measurements.
Preliminary testing shows that WiFi sensing performance is correlated to channel bandwidth. The larger the bandwidth, the higher the resolution. Channel bandwidth in the 2.4GHz spectrum is 20MHz, 5GHz is 160MHz, and 60GHz is 2GHz.
Motion sensing can be achieved using infrared and radar sensors. Patient monitoring can be driven by cameras plus AI, and smartphones can already detect gestures by amalgamating a time of flight sensor (ToF) with a standard camera. These existing solutions naturally beg the question: where does WiFi sensing fit in all this?
The answer is clear cut; WiFi sensing has an advantage over existing solutions in that for most applications, it does not need any extra hardware. Active radar systems require dedicated antennas and transceivers that are complex and costly. On the other hand, WiFi sensing uses existing devices like cell phones, PCs, and mesh WiFi systems. The user would only need to install the required software to transform their setup into a WiFi sensing.
“There are about 15 billion WiFi clients devices out there,” said Taj Manku, CEO of Cognitive Systems. “With this [WiFi sensing], you can now enable all these devices, which are never meant to be motion sensors, to now be motion sensors. And that can then provide the user with other capabilities going forward, whether that’s home monitoring…or IoT integration for smart homes. You are doing this simply by software.”
WiFi also penetrates through walls, enabling out of line-of-sight (LOS) operations, an important consideration for security monitoring applications. And because it doesn’t rely on image data, it retains a degree of privacy.
For home applications, WiFi sensing can be installed on virtually any WiFi device. Manku noted that in terms of motion sense, service quality doesn’t degrade much with the quality of the device.
“A lot of the very cheap devices, like the smart plugs, for example, they are just as good as a complicated device like an Alexa or Google Home,” said Manku.
Still, Manku noted that the software solution would evaluate every device to see if it has the necessary performance, but the baseline requirement is very low.
While the idea is promising, WiFi sensing isn’t without challenges. WiFi signals, like any wireless transmission, are vulnerable to interference that decreases their accuracy. And if the WiFi equipment acting as sensors come under heavy traffic, the depleted resources could reduce service quality.
Coverage and signal strength is another consideration for out of LOS applications. As previously mentioned, WiFi sensing works best with high-frequency, high-bandwidth transmission. But high frequencies have trouble penetrating walls. Thus, solution designers need to balance bandwidth and accuracy, rely on more sense nodes, or consider the sensor’s proximity to the target.
In enterprise scenarios like healthcare, the high resolution demands set more stringent hardware requirements. In addition to frequency and bandwidth criteria, the devices need higher processing power for active systems with large performance overheads.
Because the WiFi standard was developed with interoperability and backwards compatibility in mind, it makes it easier to layer extra functionalities on top. With that said, WiFi equipment manufacturers need to enable lower-level access and chipset firmware access to control data flow. Similarly, the operating system may also need lower-level access to network gear to allow a standardized application interaction.
“The first hurdle is that you have to be able to work with the WiFi chipset vendors,” Manku weighed in. “And there are many different chipset vendors: there’s Qualcomm, there’s Broadcom, there’s a bunch of them. When you start, you may start working with one, but then eventually, you have to start working with all of them.”
Manku said that Cognitive Systems is working with 17 chipsets today.
Equally important is how these solutions are tested and verified. Manku commented that while Cognitive Systems has its own testing facilities, other manufacturers may not have the same luxury. Thus, independent third-parties need to have a standardized testing method, and the industry needs a strong push to establish them.
When will it arrive?
Although WiFi sensing is just beginning to gain traction, solutions built around it are already here. Cognitive Systems already have both software and hardware products that help capture motion sensing. It hopes to work with major internet service providers in Canada to help to differentiate their service packages.
Another example comes from the School of Electrical Engineering & Computer Science (SEECS) at the National University of Sciences & Technology in Islamabad, Pakistan. The study, titled Wireless Health Monitoring using Passive WiFi Sensing published in 2017, explored the potential of using WiFi sensing to track tremors, falls, and breathing rates of the elderly. The study concluded that the system, developed by the university, had an 87 per cent accuracy in measuring breathing rate, 98 per cent accuracy in detecting falls, and 93 per cent accuracy in classifying tremor. Moreover, the study argued that the WiFi sensing solution is low cost and is far less “cumbersome or even demeaning” than wearing monitoring bracelets, which is even more challenging for dementia patients.
Correction: All instances of WiFi sense in this story have been replaced with WiFi Sensing. WiFi Sense generally refers to a Microsoft technology that lets multiple devices more easily connect to a WiFi network, while WiFi sensing refers to the motion detection system. Also, a previous version of this article said that Cognitive Systems is working with 17 chipset vendors. However, Cognitive Systems is actually working with 17 chipsets, not chipset vendors.