We present results from a system which uses received signal strength (RSS) measurements to estimate the speed at which a person is walking when they cross the link line. While many RSS-based device-free local- ization systems can detect a line crossing, this system estimates additionally the speed of crossing, which can provide significant additional information to a tracking system. Further, unlike device-free RF sensors which occupy tens of MHz of bandwidth, this system uses a channel of about 10 kHz. Experiments with a per- son walking from 0.3 to 1.8 m/s show the system can measure walking speed within 0.05 m/s RMS error.
Link line crossing speed estimation with narrowband signal strength
1. Link Line Crossing Speed Estimation with
Narrowband Signal Strength
Alemayehu Solomon Abrar1
, Anh Luong1
, Peter Hillyard1
, Neal Patwari1,2
1
University of Utah, 2
Xandem Technology
Abstract
We present results from a system which uses received
signal strength (RSS) measurements to estimate the
speed at which a person is walking when they cross
the link line. While many RSS-based device-free local-
ization systems can detect a line crossing, this system
estimates additionally the speed of crossing, which can
provide significant additional information to a tracking
system. Further, unlike device-free RF sensors which
occupy tens of MHz of bandwidth, this system uses
a channel of about 10 kHz. Experiments with a per-
son walking from 0.3 to 1.8 m/s show the system can
measure walking speed within 0.05 m/s RMS error.
Why Walking Speed Estimation?
• Health monitoring for elderly
• Device-free human tracking for security
• Activity recognition in smart home
Existing challenges
• Quantization of RSS
• Noisy channel measurements
• Scarce RF bandwidth resource
• Need for new wireless infrastructure
• Device cost
Proposed Method
RSS measurements from sub-GHz transceivers
• Sub-dB RSS computed from IQ samples taken from a
CC1200 radio[1]
P =
1
N
N
n=1
|sn|2
(1)
• RSS frequency content analyzed
Sub-dB RSS Measurement System
Figure 1. System Overview
• Beaglebone Green (BBG) + TI CC1200 radio.
• Operates at 169 MHz, 434 MHz, or 900 MHz bands.
• Provides a sampling rate of 449 Hz.
Figure 2. Sub-dB RSS measurement system
Human-Induced Shadowing
Diffraction Model for Link Line Crossing
Figure 3. Human modeled as a cylinder
• For a person moving in the −y direction with an
average speed of v, and positioned at (x, y0 − vt) at
any given time t, the average RSS r(t) is derived from
[2]
r(t) = c + 20 log10
R
πay
cos (2πfy(vt − y0)) (2)
Figure 4. Simulated RSS as a function of position.
• Peak RSS frequency decreases as a person approaches
a link line.
0 2 4 6 8 10 12 14
Time (s)
0
5
10
15
20
25
30
35
40
45
Frequency(Hz)
Figure 5. Simulated RSS spectrogram (Line crossed at
t= 7.3 s)
Real Human Link Line Crossing
• Average RSS frequency, not peak frequency, decreases
as a person approaches a link line.
0 10 20 30 40 50 60 70
−46
−44
−42
−40
−38
−36
−34
−32
RSS(dBm)
0 10 20 30 40 50 60 70
Time (s)
0
2
4
6
8
10
12
Avg.frequency(Hz)
unsmoothed
smoothed
Figure 6. Measured RSS (top) and time-averaged RSS
frequency (bottom) (Line crossed at t= 23 & 56 s)
Speed Estimation
• Time of crossing corresponds to the minimum average
frequency.
• The speed is estimated by scaling the average
frequency by α,
ˆv = α min
t
fav(t), (3)
where the minimum is taken over a set time interval.
Experiment
• Experiments are performed in indoor hallways.
Figure 7. Hallways for the experiment
Figure 8. Experiment setup
Results
Estimation Accuracy
• Median error of 0.05 m/s for different speeds.
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Ground Truth (m/s)
0.0
0.5
1.0
1.5
2.0
2.5
EstimatedSpeed(m/s)
Reference
Median
Figure 9. Estimated walking speed vs ground truth when
a person crosses a link.
• Median error of 0.1 m/s for different crossing
positions.
TX 1 2 3 4 5 6 7 8 RX
Position at crossing
0.4
0.6
0.8
1.0
1.2
1.4
EstimatedSpeed(m/s)
Reference
Median
Figure 10. Estimated walking speed vs ground truth
when a person crosses a link as a function of position.
Current and Future Work
• Application of speed estimates to improve device-free
localization and tracking
• Application in activity recognition and other context
aware computing systems
Acknowledgements
This material is based upon work supported by the U.S.
National Science Foundation under Grants #1329755,
1407949, and 1622741.
References
[1] Luong, A., Abrar, A.S., Schmid, T. and Patwari, N., “RSS step size: 1 dB is
not enough!” In Proceedings of the 3rd Workshop on Hot Topics in Wireless,
2016.
[2] Rampa, V., Savazzi, S., Nicoli, M., D’Amico, M., “Physical modeling and
performance bounds for device-free localization systems” IEEE Signal
processing letters, 2015.