Human Presence Detection via Wi-Fi Range-Filtered Doppler Spectrum on Commodity Laptops

This paper introduces a novel, low-complexity Human Presence Detection system for commodity laptops that utilizes the built-in Wi-Fi hardware and a new Range-Filtered Doppler Spectrum technique to achieve privacy-preserving, calibration-free occupancy sensing without requiring external sensors or infrastructure.

The Gist

Imagine your laptop is currently a bit "dumb." It only knows you're there if you touch the keyboard or move the mouse. If you walk away to grab a coffee, it keeps running, draining your battery, and leaving your screen open for anyone to see. If you walk back, it stays asleep, waiting for you to tap it.

This paper introduces a way to make your laptop "smart" enough to know you're there, approaching, or leaving—without adding any new cameras, sensors, or extra gadgets. It does this using the Wi-Fi card that is already inside your computer.

Here is how it works, explained through simple analogies:

1. The Problem: The "Blind" Laptop

Currently, laptops use one of three ways to know if you are around:

• Special Sensors: Like adding a tiny radar or infrared eye. It works well, but it costs money and uses extra battery.
• Cameras: It looks at you. But people hate this because it feels like a spy is watching them, and it drains the battery fast.
• Old Wi-Fi Tricks: It listens to Wi-Fi signals bouncing around, but it's like trying to hear a whisper in a noisy room. It can't tell if you are at your desk or if your cat is walking by in the other room.

2. The Solution: The "Echo-Location" Laptop

The authors (from Intel) figured out how to turn the laptop's existing Wi-Fi chip into a radar.

Think of your laptop as a bat in a cave. It sends out Wi-Fi signals (like sound waves) that bounce off everything in the room—your desk, the walls, and you.

• Monostatic Sensing: Usually, radar needs one device to shout and a different one to listen. This laptop does both! It uses one antenna to shout and another to listen, all on the same device.
• The Magic: When you move, the Wi-Fi signal bounces back slightly changed. By measuring these tiny changes, the laptop can tell exactly how far away you are and how fast you are moving.

3. The Secret Sauce: "Range-Filtered Doppler Spectrum" (RF-DS)

This is the fancy name for their new trick. Let's break it down with an analogy.

The "Noisy Room" Problem:
Imagine you are trying to hear a friend whispering across a crowded room. The problem is that the walls are echoing loudly, and the air conditioning is humming. These are "static" noises (like the walls or the desk) that drown out your friend (you).

The Old Way (2D Map):
Old methods try to map every single echo in the room at once. It's like trying to listen to every conversation in a stadium simultaneously. It takes a lot of brainpower (computing power) and battery, and it's hard to pick out the specific person you care about.

The New Way (RF-DS):
The authors invented a "Smart Filter."

1. Range Filtering: They tell the laptop, "Ignore everything behind the wall and everything far away. Only listen to the space between 0 and 2 meters (your desk area)." It's like putting on noise-canceling headphones that only let in the voice of the person sitting right in front of you.
2. Doppler Spectrum: They look specifically at the movement of the signal. If you are breathing, your chest moves slightly. This creates a tiny "wiggle" in the Wi-Fi signal. The new method is sensitive enough to hear that "wiggle" even if you are sitting perfectly still.

4. The "Sleeping Guard" Strategy (Adaptive Power)

One of the biggest worries with always-on sensing is battery drain. If the laptop is constantly shouting "Hello? Anyone there?" at full volume, your battery will die by lunch.

The authors added a Smart Sleep Mode:

• Idle Mode: When the laptop thinks no one is around, it whispers very slowly (checking 10 times a second). It's like a guard taking a nap but keeping one ear open.
• Detection Mode: The moment it hears a tiny movement (like you walking toward the desk), it instantly wakes up and shouts loudly (checking 100 times a second) to track exactly where you are.
• Result: It saves massive amounts of battery because it only goes into "high-alert" mode when it actually detects motion.

5. Why This Matters

• Privacy: No cameras. No one is watching you. It just listens to invisible radio waves.
• Convenience: Your laptop will wake up before you even touch it when you walk in. It will lock itself and dim the screen the moment you walk away.
• Battery Life: Because it's so efficient, it can run 24/7 without killing your battery.
• Works Everywhere: They tested it on different laptops (HP and Lenovo) and in different offices. It didn't need to be "retrained" or calibrated for each room. It just works.

In a Nutshell

This paper shows how to turn a standard laptop into a smart, privacy-friendly guardian using only the Wi-Fi chip it already has. It uses a clever "filter" to ignore background noise and focus only on you, and it sleeps lightly to save battery, waking up instantly when you approach. It's like giving your laptop a pair of invisible, super-sensitive ears that only listen for you.

Technical Summary

Here is a detailed technical summary of the paper "Human Presence Detection via Wi-Fi Range-Filtered Doppler Spectrum on Commodity Laptops."

1. Problem Statement

Human Presence Detection (HPD) is critical for intelligent power management (e.g., sleep/wake cycles) and security (e.g., auto-lock) in personal computing devices. However, existing solutions for laptops face significant limitations:

• Dedicated Hardware: Sensors like Time-of-Flight (ToF) or infrared increase cost, power consumption, and design complexity, often limiting them to premium devices.
• Camera-Based Solutions: These raise privacy concerns and struggle in low-light conditions or when the camera is obstructed.
• Traditional Wi-Fi Sensing:
  • Simple Statistical Methods: Rely on amplitude/phase variance but lack spatial (range) information, leading to high false positives from ambient motion.
  • Machine Learning Approaches: Require heavy computational resources and extensive environment-specific training, making them unsuitable for real-time, always-on laptop operation.
  • Bistatic Architectures: Most current Wi-Fi sensing relies on external Access Points (APs), which introduces variability and requires infrastructure.
• Monostatic Challenges: While commercial Wi-Fi Network Interface Controllers (NICs) can perform monostatic sensing (Tx and Rx on the same device), standard processing (2D Range-Doppler Maps) is computationally expensive and power-intensive, making it unsuitable for battery-powered laptops.

2. Methodology: Range-Filtered Doppler Spectrum (RF-DS)

The authors propose a novel, low-complexity solution using monostatic Wi-Fi sensing on the built-in NIC of a laptop. The core innovation is the Range-Filtered Doppler Spectrum (RF-DS) technique, which replaces the traditional 2D Range-Doppler Map (RDM) with a more efficient, task-specific approach.

Key Technical Components:

• Monostatic Sensing: Utilizes a single laptop's Tx and Rx antennas to capture Channel State Information (CSI) from the Long Training Field (LTF) of Wi-Fi frames. The system performs hardware synchronization (delay and phase alignment) to mitigate clock offsets.
• Range-Selective Processing: Instead of computing a full 2D FFT over all subcarriers and time frames, the system focuses only on specific "range gates" (spatial zones) relevant to the user (e.g., 0–2m for "at desk," 2–5m for "approaching").
• Enhanced Self-Interference Cancellation (FIR Filtering):
  • Standard DC removal (subtracting the mean) often suppresses slow motions like breathing.
  • The authors employ a Moving Target Indication (MTI) filter using a high-bandpass Finite Impulse Response (FIR) filter. This effectively removes static clutter (walls, furniture) and Tx/Rx leakage while preserving micro-Doppler signatures (e.g., breathing at ~5–10 mm/s).
• Matched Filtering for Doppler Spectrum:
  • For a specific range gate $R_i$, the system performs phase-compensated summation across subcarriers. This aligns signals from the target range coherently while attenuating others.
  • A 1-D FFT is then applied to the time-series of these range-specific signals to generate a Doppler spectrum.
• Temporal Windowing:
  • Instead of single-frame analysis, the system accumulates Doppler spectra over a sliding window ($M_w$) to create a time-Doppler map. This improves estimator stability and reduces false negatives.
• Adaptive Multi-Rate Framework:
  • Idle Mode: Operates at a low frame rate (10 Hz) to conserve power.
  • Detection Mode: Upon detecting motion, the rate switches to 100 Hz for high-resolution tracking.
  • State Logic: Uses SNR thresholds and range interpolation to classify states: Present (<2m), *Approaching/Leaving* (2–5m), and *Absent* (>5m). A multi-frame majority voting scheme ensures robust state transitions.

3. Key Contributions

1. First Standalone Monostatic HPD: Demonstrates a device-free HPD solution using only the internal Wi-Fi NIC of commercial off-the-shelf laptops (Wi-Fi 6E and Wi-Fi 7), requiring no external infrastructure or sensors.
2. Novel RF-DS Algorithm: Introduces a low-complexity method that applies range-area filtering before Doppler analysis. This reduces computational load by avoiding continuous 2D RDM computation while maintaining high sensitivity to micro-motions.
3. Adaptive Power Framework: A dynamic system that adjusts CSI sampling rates based on user activity, enabling "always-on" capability with minimal battery impact.
4. Generalizability: The solution operates without environment-specific calibration or retraining, working across different office layouts and hardware generations.

4. Experimental Results

The system was validated in a furnished office environment using an HP laptop (Wi-Fi 7) and a Lenovo laptop (Wi-Fi 6E).

• Micro-Motion Detection: The RF-DS method with MTI filtering successfully detected human breathing at 3 meters, a task where traditional DC cancellation failed due to signal masking.
• Approach/Leave Tracking: The system tracked a user moving from 0.5m to >6m and back. RF-DS provided smoother and more robust range/velocity estimates compared to full 2D FFT maps, particularly in low-speed and long-range scenarios.
• Cross-Platform Accuracy:
  • Lenovo (Wi-Fi 6E): 94.0% accuracy.
  • HP (Wi-Fi 7): 96.5% accuracy.
  • The system correctly classified "Present," "Absent," "Approaching," and "Leaving" states without calibration.
• Latency: The average state estimation latency was 0.48 seconds (max 0.89s), which is sufficient to trigger wake-up events before the user physically reaches the device.

5. Significance and Impact

• Privacy-Preserving: Offers a robust alternative to camera-based monitoring, addressing growing user privacy concerns.
• Cost and Power Efficiency: Eliminates the need for dedicated sensors (ToF/IR), reducing hardware costs and power consumption. The adaptive processing framework ensures the system is viable for battery-operated devices.
• Scalability: Since it relies on standard Wi-Fi protocols and built-in NICs, this solution can be deployed across millions of existing and future laptops without requiring new hardware or complex environmental training.
• Practical Deployment: The ability to distinguish between user presence, approach, and departure with sub-second latency enables seamless "Wake-on-Approach" and "Lock-on-Leave" features, enhancing both user experience and security.

In conclusion, this paper establishes a new paradigm for Human Presence Detection by transforming commodity Wi-Fi hardware into a low-power, high-accuracy radar system through intelligent signal processing (RF-DS) rather than hardware augmentation.