What Is WiFi CSI and Why Does It Matter?

WiFi CSI (Channel State Information) is the low-level signal data your 802.11 adapter captures from every wireless frame. When a human body moves through a room, it disturbs the multipath propagation of WiFi signals — the radio waves bounce off walls, furniture, and bodies differently than through empty air.

Traditional WiFi networks use this data to optimize throughput. Researchers and engineers have figured out how to use it for something far more interesting: detecting people, tracking movement, recognizing activities, and even monitoring breathing — through walls, without any cameras, without any device worn by the subject.

This is not a future capability. It is shipping now. And the implications for personal safety, defense applications, and spatial intelligence platforms are enormous.

WiFi CSI vs Traditional Computer Vision: The Real Comparison

The standard narrative pits cameras against privacy. WiFi CSI breaks that frame entirely. Here is how the two approaches actually compare:

The takeaway: For indoor presence detection, WiFi CSI matches or beats cameras in almost every dimension that matters for safety applications — and it does it without a single pixel of visual data.

How WiFi Sensing Works: The Physics

Standard 802.11n/ac/ax APs transmit using OFDM — Orthogonal Frequency Division Multiplexing. Each OFDM subcarrier carries phase and amplitude data that reflects how the signal traveled from transmitter to receiver. This per-subcarrier snapshot is the Channel State Information matrix.

When a human moves through a room, they create micro-Doppler shifts in the CSI readings. The amplitude and phase of individual subcarriers fluctuates in patterns that machine learning models can decode into:

• Presence/absence: Is someone in the room?
• Location: Where in the room are they (1-3m accuracy)?
• Movement: Walking, running, falling, sitting still
• Breathing: Respiration rate via 0.1-1 Hz amplitude oscillations in still subjects
• Gestures: Emerging research — hand movements, wake gestures

The hardware constraint that limited commercial deployment for years was access to raw CSI data. Consumer routers exposed only RSSI (received signal strength indicator) — a single number, not the full matrix. Modern platforms including ESP32, Intel 5300, Nexmon-patched Raspberry Pi firmware, and commercial SDKs now expose full CSI matrices.

The Practical Sensitivity Gap Nobody Talks About

Here is what the research papers gloss over: the difference between detection and reliable detection in real-world conditions is enormous.

Academic results often come from controlled environments with a single subject, cleared rooms, and purpose-built CSI extraction hardware. Residential and commercial environments introduce:

• Multipath clutter — furniture, HVAC ducts, and appliances all create static multipath that resembles human presence
• Environmental drift — temperature and humidity changes shift baseline CSI over hours
• Concurrent WiFi traffic — other devices’ transmissions corrupt the CSI stream
• Multi-occupant ambiguity — distinguishing two people vs one person with a raised arm remains an open problem

The platforms solving these problems in production — like the system we have been testing with our own wireless infrastructure — combine temporal filtering, background subtraction, and edge-compute classifiers running on the access point itself. No cloud. No latency. No data leaving the premises.

Defense and Tactical Applications: Why This Changes the Perimeter Game

For defense and tactical applications, WiFi CSI offers something cameras and IR cannot: passive detection that leverages existing infrastructure with zero optical signature.

A forward operating base with existing tactical mesh connectivity can repurpose those same mesh nodes for through-wall occupancy detection without deploying additional sensors. The mesh radio that handles communications also builds a live presence map of every room in a structure — before entry.

This is the integration path that makes spatial intelligence platforms built on mesh networking fundamentally different from camera-based surveillance: the same hardware stack handles communications, position, and sensing simultaneously. No additional sensor payload. No separate data stream to secure.

GPS-TDMA synchronized mesh — the architecture we have been developing at Edge Orbital — provides the precise timing substrate that makes multi-point CSI correlation across mesh nodes possible. When every node shares a common GPS time reference, cross-node CSI fusion becomes tractable. That is when through-wall detection scales from a single-room trick to a building-wide spatial intelligence layer.

Personal Safety Applications: The No-Camera Advantage

The personal safety space has been waiting for a sensing technology that works without compromising privacy. WiFi CSI delivers it.

A wearable device with a low-power WiFi CSI radio can detect when another person enters a room — without capturing any image, without requiring the other person to carry a device, and without any visible indication that sensing is occurring. For someone concerned about stalking, unauthorized entry, or unsafe environments, this is a qualitatively different threat detection capability than any camera-based system.

The hardware-software stack we are building integrates mesh networking (peer-to-peer, no cell required) with WiFi CSI presence detection in a wearable form factor. The result: a personal safety platform that works when cameras are inappropriate, when cell service fails, and when you need detection before visual confirmation.

If you want early access or are an investor evaluating this space, our full technical brief is available through the Edge Orbital data room.

The Near-Term Trajectory: Where WiFi Sensing Is Going in 2026-2027

Three developments are accelerating the WiFi CSI adoption curve faster than most analysts expect:

• 802.11bf (WiFi Sensing standard): The IEEE task group published its draft standard for standardized WiFi sensing in 2024. When this ratifies, every compliant AP ships with a standardized sensing API. Deployment goes from niche to default infrastructure.
• On-device ML acceleration: The latest AP SoCs (Qualcomm, Broadcom, MediaTek) include NPUs capable of running CSI classifiers at the edge. Zero cloud dependency, sub-100ms latency.
• Mesh network integration: As mesh WiFi systems proliferate in homes and enterprise, multi-node CSI fusion becomes a software update. The sensing infrastructure is already deployed — it just needs the software layer.

What Edge Orbital Is Building

The spatial intelligence platform we are developing at Edge Orbital integrates three layers that have not been combined before in a wearable form factor:

• GPS-TDMA mesh network — precise position, time-synchronized peer-to-peer communications that work without cell or internet
• WiFi CSI sensing — through-wall presence detection leveraging the same 2.4/5GHz radio stack as the mesh
• Edge AI classification — on-device ML classifiers running on a sub-1W processor, no cloud required

The Tripwire Recon platform — now live on the App Store — represents the first layer of this stack deployed at consumer scale. The sensing and mesh layers are in active development.

For defense and commercial partners evaluating WiFi CSI integration or licensing discussions, contact us through the data room.

For the technical architecture behind our GPS-TDMA mesh substrate, see our complete guide to Edge Orbital Sync.

— CJ Wolff, Published Patent Inventor | Founder, Edge Orbital | 33-year wireless veteran