WI-FI FINGERPRINT-BASED APPROACH TO SECURING THE CONNECTED VEHICLE AGAINST WIRELESS ATTACK

Natarajan Meghanathan et al. (Eds) : NETCOM, NCS, WiMoNe, GRAPH-HOC, SPM, CSEIT - 2016
pp. 211– 217, 2016. © CS & IT-CSCP 2016 DOI : 10.5121/csit.2016.61518
WI-FI FINGERPRINT-BASED APPROACH
TO SECURING THE CONNECTED
VEHICLE AGAINST WIRELESS ATTACK
Hyeokchan Kwon, Sokjoon Lee and Byung-ho Chung
Electronics and Telecommunications Research Institute,
218 Gajeong-ro, Yoseong-gu, Daejeon, Republic of KOREA
{hckwon, junny, cbh}@etri.re.kr
ABSTRACT
In this paper, we present wifi fingerprint-based approach to securing the connected vehicle
against wireless attack. In current connected vehicles such as Tesla EV, Mitsubishi outlander
PHEV etc., there is a wi-fi access point on the vehicle to connect to the mobile device which has
telematics apps installed. And generally the wi-fi access point is managed by the head unit
system in the vehicle. Currently, the headunit in the vehicle utilizes white-list that contain MAC
addresses of the pre-registered (i.e authorized) device. However, the white-list based
mechanism cannot detect the device that forges its MAC address with authorized one. This
paper presents security mechanism to detect rogue telematics device that has a spoofed (i.e,
forged) MAC by analysing wi-fi fingerprint. We generate wi-fi fingerprint by analysing radio
frequency features such as error vector magnitude (EVM), frequency offset, I/Q offset, sync
correlation and so on. And we also utilizing distance information for improving detection ratio.
The prototype of the proposed mechanism is implemented in this work, and we provide
experimental results.
KEYWORDS
Connected Vehicle Security, Wireless Attack, Wi-Fi Fingerprint, Telematics Device
Authentication
1. INTRODUCTION
Recently, various telematics apps for diagnostic the car, setting the configuration, locating the car,
locking it remotely etc. are exist and they use wireless network such as wi-fi, Bluetooth etc. to
connect to the vehicle. In current connected vehicles such as Tesla EV, Mitsubishi outlander
PHEV etc., there is a wi-fi access point on the vehicle to connect to the mobile device which has
telematics apps installed. And generally the wi-fi access point is managed by the head unit
system in the vehicle.
In recent years, some hacking accidents have occurred with connected vehicles providing wi-fi
access. For example, in this year, Mitsubishi outlander PHEV was hacked [1] by cracking wi-fi
PSK (Pre-shared key) and analysing binary protocol using MITM (Man in the middle attack). In
this case, the hackers were able to disable the theft alarm, unlock the car, turn the light, pop the
window/jimmy, turns on pre-heating, pre-cooling and so on. For another example, the Tesla EV
was also hacked [2] by using malicious wi-fi hotspot which is connected to a car’s web browser.
In this case, the hackers were able to remotely unlock the door, take over control of the dashboard

212 Computer Science & Information Technology (CS & IT)
computer screen, open the door, move the seats and activate the indicators and windscreen wipers,
as well as fold in the wing mirrors while the vehicle was in motion. And they were also able to
take remote control of Tesla’s brakes and door locks from 12 miles away.
In this paper, we present wi-fi fingerprint-based approach to securing the connected vehicle
against wireless attack. Currently, with regard to wi-fi access, the head unit check MAC address
of the device in order to determine whether the device is authorized or not. To do this, head unit
use white-list which consists of MAC addresses of the pre-registered device. However, the white-
list based mechanism cannot detect the device that forges its MAC address with authorized one.
This paper presents security mechanism to detect rogue device that has a spoofed (i.e, forged)
MAC by analysing wi-fi fingerprint. We generate wi-fi fingerprint by analysing radio frequency
features such as error vector magnitude (EVM), frequency offset, I/Q offset, sync correlation and
so on. And we also utilizing distance information for improving detection ratio. The prototype of
the proposed mechanism with considering EVM as a radio frequency feature is implemented in
this work, and we provide experimental results. So far, security research regard to security
vulnerability/threat on connected vehicle has not been conducted.
The rest of the paper is organized as follows. The wi-fi fingerprint-based telematics device
authentication mechanism and experiment result is described in section 2. Finally, conclusion is
given in section 3.
Wi-Fi access point
Mobile Telematics device
IVN
CAN, FlexRay, Most,
ethernet.. …GW
Head unitTelematics apps
OBD2
Figure 1. Wi-fi based connected vehicle
2. WI-FI FINGERPRINT-BASED TELEMATICS DEVICE
AUTHENTICATION MECHANISM FOR CONNECTED VEHICLE
In order to authenticate telematics device, we utilizes the radio frequency features and distance
information (i.e. RSSI). In the wi-fi fingerprint generation phase, the wi-fi access point on the
vehicle collects and analysis wi-fi signals of the device by moving the physical location of the
device and generates wi-fi fingerprint. In the verification phase, it estimates the distance from
device to vehicle by using RSSI value, and it analyses wi-fi fingerprint data with the best nearby
radio frequency features in current relative distance with the vehicle. Figure 2 shows the overall
architecture of this mechanism.

Computer Science & Information Technology (CS & IT) 213
Collect
wireless signal
(for each zone)
Analyze
Radio frequency
features
(training)
(MAC,
wi-fi fingerprint )
at zone #
Collect
wireless signals
& RSSIs
Comparing
Wi-fi
fingerprint
Analyze
Radio frequency
feature
MAC forged device?
Authorized device?
Wi-fi fingerprint generation phase
Relative
distance
estimation
Zone #Wi-fi
fingerprint
Wi-fi
fingerprint
Zone 1
Zone 2
Zone 3
Wi-fi fingerprint verification phase
Figure 2. Overall process of wi-fi fingerprint–based MAC verification for telematics device
2.1. Wi-fi fingerprint generation mechanism
In the Wi-fi fingerprint generation phase, the head unit registers MAC address of the authorized
device. And it determines a wireless signal collection zone, and moves wireless device to the
measurement point and generates wireless wi-fi signals. The head unit collects wi-fi signals from
the authorized device, and then analyse them by machine learning algorithm such as K-NN, SVM
and so on, and creates wi-fi fingerprint of the authorized device. In this paper, we use K-NNDD
(K-Nearest Neighbour Data Description) [5] for training radio frequency of authorized devices.
The relative distance and RF fingerprint information is stored in wi-fi fingerprint database.
In this paper, we applied error vector magnitude (EVM) as a RF feature. EVM is a vector
magnitude difference between an ideal reference signal and measured signal. Figure 3 shows the
concept of error vector magnitude (EVM) and mathematical formula for deriving EVM value.
, where
, (formula 1)
Q
I
Phase
Error
Error
Vector E.V.M.(Error Vector Magnitude)
= |Error Vector|
IQmeas
IQref(ideal)
ErrQ
ErrI
Figure 3. The concept of the Error Vector Magnitude
EVM is calculated by comparing the difference between measured signals with an ideal reference
signals for determining the error vector. The EVM value is the root mean square value of the
error vector over time at the instants of the symbol clock transitions. There are various reasons of
mismatching measured signal with reference ideal signal such as hardware impairment, channel
characteristics, noise at the receiver and modulation error. By using modulation error, we can

identify particular wireless devices with different manufacturer or different wifi-chipset or even
the same manufacturer/wifi-chipset.
2.2. Wi-fi fingerprint verification mechanism
In the Wi-fi fingerprint verification phase, the head unit collect and analyse RF signals of the
device and extracts MAC address, RSSI and radio frequency features. And then it estimates the
relative distance from the device to the vehicle by analysing the RSSI value. To calculate distance
from RSSI values we used the following formula:
The correction factors are derived through iterative experiments by minimizing the difference
from the value by distance estimation algorithm with real distance. The notation d in this formula
is a distance. The head unit then selects the radio frequency features having the highest P(radio
frequency features | d) of the MAC of the device. P(radio frequency features | d) means a
probability of radio frequency features in a given zone. Then it creates radio frequency signature
from the radio frequency features by using machine learning algorithm. Then it determines
whether the device having cloned MAC by comparing the wi-fi signature of the device with the
device having same MAC in the database. In this paper, we use K-NN(K-Nearest Neighbor)
algorithm for comparing measured radio frequency signature with reference radio frequency
signature.
Relative zone estimation
- Fitering RSSI
- Relative distance estimation
Zone 1
Zone 2
Zone 3
Forged device
Wi-fi Connection trial
Collect wireless signal
Generate wi-fi fingerprint
verify wi-fi fingerprint
{RSSI, MAC, radio frequency feature}
Zone 2
Wi-Fi signature
wi-fi
fingerprint
database
MAC spoofed (forged) device
Not accept connection
Figure 4. Detection process of rogue device
2.3. Experiments result
We developed hardware platform to collect wi-fi radio frequency signal and extract wireless
features. Figure 5 shows the developed HW platform which can be installed to the head unit in
the connected vehicle. This hardware platform includes Atheros 9380 WLAN chipset for
monitoring wi-fi signals. We developed test platform with related user interface and dashboard,
and we developed also wireless attack tool for evaluating developed system. Figure 6 shows the
screenshots of our attacking tool to create cloned MAC. Figure 7 shows the UI of test platform
for MAC spoofing device through wi-fi signature analysis and verification. Table 1 shows the
experiment result.

Figure 5. Prototype hardware platform which supporting wireless radio frequency feature extraction
Figure 6. Snapshot of the attack tool, it shows the creation of MAC forging device
Figure 7. Snapshot of the test platform, it shows the MAC verification process in GUI

Table 1. Experiment result (FAR: False Accept Rate, FRR: False Reject Rate, EER: Equal Error Rate)
Test device threshold FAR FRR EER
Mobile device with different chipset
(smart phone w/ Broadcom chipset, laptop computer
w/ intel chipset, laptop computer w/ atheros chipset)
0.91 2.7% 0% 0.8%
Mobile device with same wi-fi chipset
(iphone4s smart phone w/ broadcom’s BCM 4330,
iphone4 smart phone w/ broadcom’s BCM 4329)
0.86 10.04% 0% 5.34%
3. CONCLUSIONS
In this paper, we present wifi fingerprint-based approach to securing the connected vehicle
against wireless attack.
This paper provide the security mechanism to detect rogue device that has a spoofed (i.e, forged)
MAC by analysing wi-fi fingerprint. We generate wi-fi fingerprint by analysing radio frequency
features such as error vector magnitude (EVM). And we also utilizing distance information for
improving detection ratio. The distance information is derived by RSSI value which in included
in wi-fi signal. The prototype of the proposed mechanism with considering EVM as a radio
frequency feature is implemented and we provide experimental results. The proposed mechanism
analyse a characteristics of the wi-fi radio frequency signal of the device for detecting MAC
spoofing device. We also developed wireless attacking tool and test platform with GUI.
In our experiments, the FAR is 2.7% in case that test mobile device has different chipsets and
10.4% in case that test mobile device has same chipsets. The detection rate should be improved
when rogue device with a same manufacturers and wi-fi chipset with authorized one. Currently,
we are designing the algorithm consider additional wi-fi radio frequency features such as IQ
offset, sync correlation and so on.
ACKNOWLEDGEMENTS
This work was supported by Institute for Information & communications Technology Promotion
(IITP) grant funded by the Korea government (MSIP) (No.R-20160226-002842, Development of
V2X Service Integrated Security Technology for Autonomous Driving Vehicle)
REFERENCES
[1] Hacking the Mitsubishi Outlander PHEV hybrid, https://www.pentestpartners.com/blog/hacking-the-
mitsubishi-outlander-phev-hybrid-suv/, Pen test partners, 2016
[2] Hackers take Remote Control of Tesla's Brakes and Door locks from 12 Miles Away,
http://thehackernews.com/2016/09/hack-tesla-autopilot.html, The Hacker News, 2016
[3] Hao, Peng, "Wireless Device Authentication Techniques Using Physical-Layer Device Fingerprint"
(2015). Electronic Thesis and Dissertation Repository. Paper 3440. Western university,
http://ir.lib.uwo.ca/etd/3440
[4] H. Kwon, G.An, S.H.Kim and B.H.Chung, "Detecting cloned devices in wireless network using RSSI
and RF Features", ICONI, Dec., 2014
[5] J. Son and S. Kim, "kNNDD-based One-Class Classification by Nonparametric Density Estimation,"
Journal of the Korean Institute of Industrial Engineers, Vol. 38, No. 3, pp. 191-197, Sep. 2012.

[6] JP Hubaux, S Capkun, J Luo, The security and privacy of smart vehicles, IEEE Security & Privacy
Magazine, 2004
[7] AirTight Patent, Method and system for monitoring a selected region of an airspace associated with
local area networks of computing devices, Patent# US 7,002,943 Feb, 2006
[8] Y. Shi and Michael A. Jensen, Improved Radiometric Identification of Wireless Devices Using
MIMO Transmission, IEEE Transactions on Information Forensics and Security, Dec. 2011
[9] R. Beyah and A. Venkataramen, Rogue-Access-Point Detection - Challenges, Solutions, and Future
Directions, IEEE Security & Privacy, vol.9, issue 5, pp.56-61 (2011)
[10] Agilent 8 Hints for Making and Interpreting EVM Measurements, Agilent Technologies, 2005
AUTHORS
Hyeokchan Kwon
Received PhD degree in computer science from Chungnam National University in 2001.
Since 2001, he is currently a principal researcher in electronics and telecommunications
research institute (ETRI) in Korea. His research interests include automotive security,
wireless intrusion prevention system and IoT security, etc.
Sokjoon Lee
Received MS degree in computer engineering from Seoul National University in 2000.
Since 2000, he is currently a principal researcher in electronics and telecommunications
research institute (ETRI) in Korea. His research interests include cryptographic protocol
and ICT-Physical convergence security such as medical security, automotive security, etc.
Byung-ho Chung
Received PhD degree in computer science from Chungnam National University in 2004.
He joined Agency for Defense Development (ADD) in 1988 where he was a senior
researcher for 12 years. Since 2000, he is currently a principal researcher in electronics
and telecommunications research institute (ETRI) in Korea. His research interests include
automotive security, medical security, wireless security, multimedia security, etc.

WI-FI FINGERPRINT-BASED APPROACH TO SECURING THE CONNECTED VEHICLE AGAINST WIRELESS ATTACK

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to WI-FI FINGERPRINT-BASED APPROACH TO SECURING THE CONNECTED VEHICLE AGAINST WIRELESS ATTACK (20)

Recently uploaded (20)

WI-FI FINGERPRINT-BASED APPROACH TO SECURING THE CONNECTED VEHICLE AGAINST WIRELESS ATTACK