6.808: Mobile and Sensor Computing aka IoT Systems Lecture 4: Device-Free Localization and Seeing Through Walls
Last Lecture: Device-based Localization
This Lecture: Using radio signals to track humans without any sensors on their bodies
This Lecture: Using radio signals to track humans without any sensors on their bodies Operates through occlusions
Example: WiTrack
Device Device in another room
Applications Smart Homes Energy Saving Gaming & Virtual Reality
Measuring Distances Tx Rx Distance = Reflection time x speed of light
Measuring Reflection Time Option1: Transmit short pulse and listen for echo Tx pulse Rx pulse Time Reflection Time
Measuring Reflection Time Option1: Transmit short pulse and listen for echo Tx pulse Rx pulse Signal Samples Time Reflection Time Capturing the pulse needs sub-nanosecond sampling Why? and why was this not a problem for Cricket?
Capturing the Distance = time x speed pulse needs sub- “smallest “smallest distance nanosecond time” resolution” sampling 10 cm = Δ t × (3 × 10 8 ) Why? Δ t = 0.3 ns 0.3ns period => how many samples per second? Multi-GHz samplers SamplingRate = 1 are expensive, have Δ t high noise, and create 3GSps! >> MSps for WiFi, large I/O problem LTE… Why was this not a because speed of ultrasound problem for Cricket? 10 cm = Δ t × 345 SamplingRate = 1 Δ t ≈ 3 kbps
FMCW: Measure time by measuring frequency Transmitted Frequency Time t How does it look in time domain?
An FMCW example
FMCW: Measure time by measuring frequency Transmitted Frequency Received Δ F Δ F Reflection Time = slope Time t t+ Δ T How do we measure Δ F?
Measuring Δ F • Subtracting frequencies is easy (e.g., removing carrier in WiFi) • Done using a mixer (low-power; cheap) Power Transmitted Mixer FFT Received Δ F Signal whose frequency is Δ F let’s talk about FFTs a bit — freq
Measuring Δ F • Subtracting frequencies is easy (e.g., removing carrier in WiFi) • Done using a mixer (low-power; cheap) Power Transmitted Mixer FFT Received Δ F Signal whose frequency is Δ F Δ F ➔ Reflection Time ➔ Distance
Challenge: Multipath ➔ Many Reflections Tx Rx Reflection Power Multi-paths mask person Distance
Static objects don’t move ➔ Eliminate by subtracting consecutive measurements Multi-path Power @ time t - Distance Multi-path Why 2 peaks when we only have one moving Power @ time t+30ms person? Distance = 2 meters Power Distance
Challenge: Dynamic Multipath The direct reflection arrives before dynamic multipath! Tx Rx Moving Person Dynamic Power Multi-path Distance
Mapping Distance to Location Person can be anywhere on an ellipse whose foci are (Tx,Rx) d Rx Rx’ Tx By adding another antenna and intersecting the ellipses, we can localize the person
Challenge: Dynamic Multipath Fails for multiple people in the environment, and we The direct reflection arrives before dynamic multipath! need a more comprehensive solution Tx Rx Moving Person Dynamic Power Multi-path Distance
How can we deal with multi-path reflections when there are multiple persons in the environment?
Idea: Person is consistent across different vantage points while multi-path is different from different vantage points
Combining across Multiple Vantage Points Experiment: Two users walking Setup Single Vantage Point Mathematically: each round-trip distance can be mapped to an ellipse whose foci are the transmitter and the receiver
Mapping 1D to 2D heatmap
Combining across Multiple Vantage Points Experiment: Two users walking Setup Two Vantage Points
Combining across Multiple Vantage Points Experiment: Two users walking Setup 16 Vantage Points Localize the two users
How can we localize static users?
Dealing with multi-path when there is one moving user Tx Rx We eliminated direct table reflections by subtracting consecutive measurements Needs User to Move
Dealing with multi-path when there is one moving user Tx Rx STATIC We eliminated direct table reflections by subtracting consecutive measurements Needs User to Move
Exploit breathing motion for localize static users • Breathing and walking happen at different time scales – A user that is pacing moves at 1m/s – When you breathe, chest moves by few mm/s • Cannot use the same subtraction window to eliminate multi-path
30ms subtraction window User walking @ 1m/s User Still (Breathing) 8 7 6 Distance (meters) 5 4 3 2 1 0 -4 -3 -2 -1 0 1 2 3 4 Distance (meters) Cannot localize Localize the person
3s subtraction window User walking User Still (Breathing) Person appears in two Localize the locations person
Centimeter-scale localization without requiring the user to carry a wireless device Localize the two users
Want a silhouette 2 People are points .5 1 .5 Localize the two users 0 -0.2 0 0.2 0.4 0.6 0.8
Approach: Combine antenna arrays with FMCW to get 3D image • 2D Antenna array gives 2 angles • FMCW gives depth (1D) 2D array 1D 1D
Challenge: We only obtain blobs in space Output of 3D RF Scan Blobs of reflection power 40
Challenge: We only obtain blob in space At every point in time, we get reflections from only a subset of body parts. At frequencies that traverse walls, human body parts are specular (pure mirror) RF Scanning Cannot Capture Setup Reflection
Solution Idea: Exploit Human Motion and Aggregate over Time RF Scanning Setup
Solution Idea: Exploit Human Motion and Aggregate over Time RF Scanning Setup New Previous Location Location Combine the various snapshots
Human Walks toward Sensor 3m 2.5m 2m Chest (Largest Use it as a pivot: for motion Convex Reflector) compensation and segmentation
Human Walks toward Sensor 3m 2.5m 2m Head Right arm Left arm Lower Torso Feet Chest (Largest Use it as a pivot: for motion Combine the various snapshots Convex Reflector) compensation and segmentation
Human Walks toward Sensor
Sample Captured Figures through Walls Sample Captured Figures through Walls 2 2 2 2 1.5 1.5 1.5 1.5 y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) 1 1 1 1 0.5 0.5 0.5 0.5 0 0 0 0 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters)
Sample Captured Figures through Walls Sample Captured Figures through Walls Through-wall classification accuracy of 90% among 13 users 2 2 2 2 2 2 2 2 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) y-axis (meters) 1 1 1 1 1 1 1 1 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0 0 0 0 0 0 0 0 -0.2 0 0.2 0.4 0.6 0.8 -0.2 -0.2 0 0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 -0.2 -0.2 0 0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 -0.2 0 0.2 0.4 0.6 0.8 x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters) x-axis (meters)
Lecture Summary • Device-free localization via radio reflections • FMCW as a way to estimate distance • Multipath problem • Extending to multiple people and static humans • Beyond Localization
Recommend
More recommend
