ACM SIGCOMM Software Radio
Implementation Forum (SRIF) 2014
SDR-based Passive Indoor
Localization System for GSM
Islam Alyafawi, Desislava Dimitrova, Torsten Braun
Universität Bern
[email protected], cds.unibe.ch
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Passive Localization of Wireless Devices
>
System can
— overhear radio (e.g., GSM, WiFi) signals,
— process them to retrieve user identity, and
— locate user based on the signal properties.
>
System components based on software-defined radio
—
—
—
—
>
Radio signal acquisition
Signal property retrieval, e.g., timestamps, power levels
Message parsing, e.g., identifiers
Localization algorithms
Applications
— Analysis of customer behaviour in shopping centres /
amusement parks
— Analysis of number of people and movements in public areas
Chicago, August 18, 2014
2
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Passive Localisation System
AN: anchor node / radio sensor
wireless
device
signal
processing
and
localization
algorithm
storage
base station
Chicago, August 18, 2014
3
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Universal Software Radio Peripheral
> USRP hardware is
controlled by open source
USRP Hardware Driver,
which translates
instructions between
FPGA hardware and
signal processing
software
> GNUradio applications
— Airprobe intercepts
GSM downlink messages.
— OpenBTS implements
base station protocol stack
up to layer 3.
Chicago, August 18, 2014
4
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
System Implementation
Chicago, August 18, 2014
5
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
GSM Message Capturing
a. Sample capturing
b. GNUradio
low pass filter
c. Interpolator
d. Time synchronization
— Training sequence
discovery
— Normal burst detection
— Message reconstruction
e. Message parsing
Alyafawi et al.: Real-Time Passive Capturing of the GSM Radio, IEEE ICC 2014
Chicago, August 18, 2014
6
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Localization Algorithms
>
>
>
Range-based positioning using Time/Angle (Difference) of
Arrival, Received Signal Strength (RSSI) and multi-lateration
Finger-printing
Proximity-based positioning, e.g., Centroid
Trilateration
Chicago, August 18, 2014
Centroid
7
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Linear Weighted Centroid (LWC)
Chicago, August 18, 2014
8
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Differential RSS
𝑑𝑑
𝑑𝑑0
−ψ
>
𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑃𝑃𝑟𝑟 𝑑𝑑 = 𝐴𝐴 − 10 𝛼𝛼 log
>
>
Select 3 ANs with largest RSS values
Calculate DRSS values between each AN pair
>
𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑗𝑗 = 𝑃𝑃𝑟𝑟 𝑑𝑑𝑖𝑖 − 𝑃𝑃𝑟𝑟 𝑑𝑑𝑗𝑗 = 10 𝛼𝛼 log
𝑑𝑑𝑗𝑗
𝑑𝑑𝑖𝑖
− ψ𝑖𝑖𝑖𝑖
— X = RSS1 – RSS2
— Y = RSS1 – RSS3
— Z = RSS2 – RSS3
>
>
>
>
>
Q1 = Y/X
Q2 = Z/X
Q3 = Z/Y
ω1:ω2:ω3 = Q1:Q2:Q3
ω1+ω2+ω3 = 1
Chicago, August 18, 2014
9
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Combined Differential RSS (CDRSS)
1. Form all possible K triangles
2. Calculate weights ωik,DRSS
— i = 1,2,3
— k = 1..K for all K triangles
3. Calculate weights
ωi,CDRSS for the 3 ANs
with highest RSS
Chicago, August 18, 2014
10
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Weighted Circumcenter (WCC)
1.
2.
3.
Form triangle using 3 ANs with largest
RSS values
Calculate circumcenter
Calculate DRSS values
–
–
–
–
4.
5.
6.
X = RSS1 – RSS2
Y = RSS1 – RSS3
Z = RSS2 – RSS3
h1 = X/Y, h2 = X/Z, h3 = Y/Z
Move circumcenter point
to each AN:
Calculate AN weights ωi,WCC using
differential RSS for new triangle
Estimate coordinates
of mobile device
Chicago, August 18, 2014
11
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Localization Performance
Chicago, August 18, 2014
12
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Impact of Open/Closed Doors
Chicago, August 18, 2014
13
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
RSS of (Non-)Line-of-Sight Signals
Chicago, August 18, 2014
14
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Summary and Outlook
>
>
>
SDR systems allow new opportunities for signal processing
Positioning based on proximity-based localization algorithms
(CDRSS and WCC) outperform LWC
Promising results but challenges remain,
main challenge: multi-path mitigation
Chicago, August 18, 2014
15
Torsten Braun: SDR-based Passive Indoor Localization System for GSM
Thanks for your attention !
>
>
[email protected]
cds.unibe.ch
Chicago, August 18, 2014
16