Wi-Fi sensing is often described as “using Wi-Fi to detect motion.” That description is technically correct – but it barely scratches the surface. Modern Wi-Fi sensing systems don’t simply detect that something moved. Instead they measure extraordinarily small changes in radio propagation, extract complex patterns from those measurements, and increasingly use AI models to interpret what those patterns actually mean.
Understanding how this works helps explain why Wi-Fi sensing has become one of the most exciting new applications for wireless infrastructure—and why recent advances in Wi-Fi 6, Wi-Fi 7, and AI have dramatically accelerated the technology.
Every Wi-Fi signal contains information beyond data
When we think about Wi-Fi, we normally think about digital communication.
- An access point sends packets.
- A client receives them.
- Data moves across the network.
But every transmission also carries something else: information about the physical environment the radio waves travelled through. Whenever a Wi-Fi packet travels between transmitter and receiver it experiences:
- Reflections
- Scattering
- Diffraction
- Attenuation
- Phase shifts
- Multipath propagation
Meaning that walls, furniture, doors—and especially people—modify the signal in measurable ways. A person walking across a room changes hundreds of radio paths simultaneously. Even tiny chest movements caused by breathing alter the received signal. The remarkable fact is that modern Wi-Fi hardware can measure these changes in RF signals with astonishing precision.
Channel State Information: the hidden data inside Wi-Fi
The key technology enabling Wi-Fi sensing is Channel State Information (CSI). Rather than simply measuring signal strength (RSSI), CSI measures how every individual OFDM subcarrier behaves as it propagates through the environment. Instead of one signal measurement, CSI provides dozens—or even hundreds—of measurements for every packet. For each subcarrier the receiver records values such as:
- Amplitude
- Phase
- Frequency response
- Timing characteristics
Together these measurements create an extremely detailed “fingerprint” of the radio environment. Whenever anything changes in the room, the fingerprint changes too.
Why RSSI is not enough
Many early research projects attempted sensing using simple RSSI values. RSSI works for coarse motion detection but suffers from major limitations:
- Only one measurement per packet
- Highly susceptible to interference
- Poor spatial resolution
- Limited repeatability
CSI solves these problems. Instead of asking: “Did the signal become weaker?” CSI effectively asks: “Exactly how did every frequency component of the radio channel change?”
That produces vastly richer information.
Multipath: Usually a problem – now an opportunity
Traditional wireless engineers spend their careers trying to overcome multipath. Signals bounce off:
- Walls
- Ceilings
- Floors
- Windows
- Furniture
- Metal objects
- People
Multiple reflected copies arrive at the receiver at different times and normally this creates interference. For sensing, however, multipath becomes an enormous advantage because each reflection effectively provides another “view” of the environment. Modern sensing algorithms combine all of these reflections into a surprisingly detailed understanding of movement. Instead of treating reflections as noise, Wi-Fi sensing treats them as valuable information.
Tiny movements create measurable changes
One of the most surprising aspects of Wi-Fi sensing is just how sensitive it can be. Researchers have repeatedly demonstrated detection of:
- Breathing
- Heartbeats
- Hand gestures
- Finger movements
- Posture changes
- Falls
- Sleep movements
How is this possible? Human bodies are excellent reflectors of 5 GHz and 6 GHz radio waves. Even a chest moving only a few millimetres during breathing slightly changes the phase of reflected signals. Those changes are extremely small – but entirely measurable. The more antennas available, the more accurately the system can isolate those tiny variations.
MIMO makes sensing dramatically better
Modern Wi-Fi uses multiple antennas. This technology – Multiple Input Multiple Output (MIMO) – was originally developed to increase throughput. It also happens to be ideal for sensing because instead of observing the environment from one perspective, multiple antennas observe it simultaneously from different positions. Each antenna pair measures a slightly different radio channel.
The result is much richer spatial information. As Wi-Fi evolved from 2×2 MIMO, to 4×4 MIMO, and even to 8×8 MIMO, the sensing capability improved dramatically as well. Wi-Fi 7 continues this trend with even more sophisticated multi-link and multi-band capabilities.
OFDM creates hundreds of sensing points
Another reason Wi-Fi sensing works so well is OFDM (Orthogonal Frequency Division Multiplexing). Instead of transmitting on one carrier frequency, Wi-Fi divides the channel into many subcarriers. For example:
- Wi-Fi 6 uses hundreds of OFDM subcarriers
- Each subcarrier experiences the environment slightly differently.
- CSI records measurements for each one.
Think of it as replacing one camera with hundreds of tiny cameras – all observing the room through slightly different radio paths.
Beamforming adds another dimension
Wi-Fi 6 and Wi-Fi 7 also make extensive use of beamforming. Normally beamforming focuses radio energy toward a client device. For sensing it provides another benefit: Changing beam directions effectively allows the system to observe the environment from multiple radio perspectives. Future sensing systems are expected to combine:
- Beam steering
- CSI
- Multiple antennas
- Multiple frequency bands to improve localisation accuracy significantly.
Why AI changed everything
Collecting CSI is only the first step. Interpreting it is the real challenge. Raw CSI data resembles a gigantic stream of complex numbers that human engineers cannot realistically analyse directly. Machine learning changed that and now neural networks learn to recognise patterns associated with:
- Walking
- Sitting
- Falling
- Sleeping
- Entering rooms
- Unusual behaviour
- Occupancy changes
Instead of programming rules manually, AI models learn from thousands—or millions—of examples. This dramatically improves both accuracy and robustness.
From motion detection to activity recognition
First-generation sensing systems answered simple questions: Is somebody present? Did someone move? Today’s AI-powered systems answer much richer questions. For example:
- Is someone cooking?
- Has an elderly person fallen?
- Is a meeting room occupied?
- How many people are present?
- Is a patient breathing normally?
- Is unusual behaviour occurring?
The industry increasingly refers to this as context awareness rather than simple motion detection.
Edge AI or Cloud AI?
One important architectural decision concerns where AI processing occurs. Some vendors perform inference directly inside access points, gateways, or edge processors. Advantages include:
- Lower latency
- Improved privacy
- Reduced bandwidth
Other systems upload CSI data to cloud services where larger AI models perform analysis while many commercial products now use hybrid approaches where simple event detection occurs locally and more advanced analytics happen in the cloud.
Standardisation begins: IEEE 802.11bf
Until recently, every sensing implementation was proprietary meaning different vendors used different CSI extraction methods, different APIs, different algorithms, and usually different hardware. IEEE recognised the need for standardisation. The result is IEEE 802.11bf, the first Wi-Fi amendment dedicated specifically to wireless sensing. Its objectives include:
- Standard CSI reporting
- Interoperable sensing measurements
- Improved timing precision
- Sensing-specific signalling
- Coexistence with normal Wi-Fi communications
Importantly, 802.11bf does not standardise sensing algorithms and instead standardises how measurements are collected and exchanged. This allows vendors to innovate while ensuring interoperability at the radio level.
Why Wi-Fi 7 matters
Although sensing works with earlier Wi-Fi generations, Wi-Fi 7 provides significant advantages. These include:
- Wider channels
- More bandwidth
- Improved timing precision
- Multi-Link Operation (MLO)
- Higher packet rates
- More sophisticated beamforming
- Lower latency
All of these improvements also improve sensing performance, which is why many experts expect Wi-Fi 7 deployments to become the foundation for the first large-scale commercial sensing ecosystems.
WiFi NOW’s take: Wi-Fi sensing is no longer just an interesting research topic. It is becoming a sophisticated radio intelligence platform built on advances in CSI, MIMO, OFDM, beamforming, and AI. The key insight is that the same infrastructure already deployed to deliver connectivity can also become a rich source of environmental awareness with little or no additional hardware.
As IEEE 802.11bf matures and Wi-Fi 7 deployments expand, the focus will increasingly shift from how radio waves are measured to what valuable insights can be extracted from them. That transition—from signal processing to intelligence—is likely to define the next phase of the Wi-Fi sensing industry.
Coming in Part 3: The Wi-Fi Sensing Ecosystem – Industry Leaders, Standards, Applications, and the Road to Commercial Adoption.