A Lightweight Message Authentication Framework in the Intelligent Vehicles System

The International Journal of Engineering and Science (IJES)
|| Volume || 6 || Issue || 6 || Pages || PP 33-39 || 2017 ||
ISSN (e): 2319 – 1813 ISSN (p): 2319 – 1805
DOI: 10.9790/1813-0606013339 www.theijes.com Page 33
A Lightweight Message Authentication Framework in the
Intelligent Vehicles System
1
Mostafa El-Said, 2
Alexander Arendsen, 3
Samah Mansour
1
School of Computing and Information Systems Grand Valley State University Allendale, MI 49401-9403
2
University of Central Florida Computer Science Dept 100 Weldon Boulevard, Sanford, FL 32773
3
School of Computing and Information Systems Grand Valley State University Allendale, MI 49401-9403
--------------------------------------------------------ABSTRACT-----------------------------------------------------------
Intelligent Vehicles System (IVS) supports a wide variety of Advanced Driver Assistance System (ADAS)
services such as vehicle visibility detection. In implementing this service, the message authentication is a vital
design parameter that protects victim vehicles from being tricked into accepting false messages as legitimate
ones and make a false decision based on the incoming message. However, implementing message authentication
service is too expensive especially if vehicles, initially, don’t trust each others or there is no certificate of
authority in place.
In this research, we investigate the use of the Basic Safety Message (BSM) behavior over time as a metric to
allow a receiving vehicle to anticipate at what distance it will continue to receive BSMs from within-range
vehicles. Therefore, the victim vehicle would reject the BSM messages that fall outside its acceptance window.
Simulation experiments are setup to study the realistic behavior of the BSM messages in different environment
characteristics including changing the vehicle size, number of road lanes and vehicle speed. Research findings
suggested that the lightweight message authentication can assist vehicles in estimating the duration for a trusted
relationship among those that are located within range of each others.
Keywords - ITS, DSRC, message authentication, Visibility.
-------------------------------------------------------------------------------------------------------------------------
Date of Submission: 14 January 2017 Date of Accepted: 09 June 2017
-------------------------------------------------------------------------------------------------------------------------------------------
I. INTRODUCTION
In a connected vehicle system, Basic Safety Message (BSM) is used to convey road safety messages on US high
ways [1]. Connected vehicles are communicating using either Vehicle-to-Vehicle (V2V) or Vehicle-to-
Infrastructure communication architecture. V2V communications aims to provide each vehicle with 360-degree
situational awareness of nearby vehicles, and by complementing onboard sensors in detecting possible crash
scenarios [3]. Moreover, in V2V or V2I architecture, vehicles use Dedicated Short Range Communications
(DSRC) transmissions to share Advanced Driver Assistance System (ADAS) safety information services such as
road conditions or vehicle’s location, weight, size or its heading direction, probability of vehicle collision with
neighboring vehicles. Authors in [1], indicated that a recent report issued by the National Highway Traffic
Safety Administration United States (NHTSA) shows that implementing just two V2V safety applications
(Intersection Movement Assist (IMA) and Left Turn Assist (LTA)) will potentially prevent 25,000 to 592,000
crashes, save 49 to 1,083 lives, avoid 11,000 to 270,000 maximum abbreviated injuries, and minimize 31,000 to
728,000 property damage. Research efforts to better use the BSM message in the dissemination process of
safety messages presented in the work presented by authors in [1]
Authors in [1], developed a Tractor-Trailer Basic Safety Message (TT-BSM) extensions to correctly represent
the location of the articulated vehicles in V2X communications. The authors’ approach aims to reduce the
potential for false safety warnings messages in the DSRC-based systems. Authors indicated that the additional
data sent in the BSM will have a minor effect on the over-the-air data traffic or the channel utilization efficiency
because the number tractor-trailer vehicles is fairly a small fraction of the overall vehicles on the road.
However, attacking and altering the safety messages sent in the V2V environment becomes possible due to the
fact that vehicles have various penetration inputs. Consequently, this will lead to defeat the purpose of
implementing a V2V communication to promote awareness and safety on our roads.
Vehicle attack surface represents all the possible ways to attack a victim vehicle. The target could be the
vehicle’s On Board Unit (OBU) or the entire vehicle system. The vehicle has multiple entry points to
synchronize the attack that is represented by the system’s possible weaknesses. In a V2V or V2I network
architecture, an attacker will consider all the possible ways to inject malicious or fake data into the vehicle
system via any of its sensors such as RADAR, LIDAR, ultrasound or the DSRC sensor such as shown in Fig. 1.

A Lightweight Message Authentication Framework in the Intelligent Vehicles System
Furthermore, the attacker constructs a dynamic threat model to study and track the victim vehicle to learn about
how these sensors communicate and at what time they are active. Then, choosing the location, entry point and
the right time to inject the malicious data becomes a critical factor to execute the hacking activity [2].
Figure 1. V2V Attack Scenario
Efforts made research and industry communities to protect the BSM messages over the air. In [3], authors
introduced an On-board Unit (OBU) VBSS solution, which generates a short-term certificate and signs Basic
Safety Messages (BSM) to preserve privacy and enhance security. In this work, authors developed their
solution based on the Public Key Infrastructure (PKI) where vehicles create groups. Each group will have its
public and private keys, and a vehicle will sign a message using its group’s private key. Moreover, authors
encouraged research community to continue conducting further research to validate the proposed vehicle-based
approach in V2V credential management.
In this research, we are investigating another Vehicle-Based Security System (VBSS) for a V2V system to
provide security and privacy via developing a lightweight message authentication framework solution. The
proposed framework allows a victim vehicle to discriminate against spoofed BSM messages. When building this
framework, several design goals were considered and special efforts were made to:
- Provide on a vehicle infrastructure-less solution.
- Rely on a Certificate Authority-less solution. Therefore, no need to the use of certificate authority body
hosted on a fixed Radio Service Unit (RSU).
- Predict the authentication session duration for how long the two vehicles will build their trust relationship via
message authentication.
The proposed solution’s framework consists of two phases such as described below:
Phase-I:
 In this phase, the authentication session lifetime is estimated by having two vehicles predicting a Visibility
Time Window (VTW) factor based on the timing and the sequence of the BSM messages and the vehicle’s
speed. So, essentially the VTW window defines the duration for the authentication session. Simulation
experiments will be conducted to determine the VTW in different simulation environments. Messages arrived
outside the VTW may be ignored because they may redundant or malicious.
Phase-II: Monitoring the consistency of a
 In this phase, the sending vehicle will calculate a Message Authentication Code (MAC) and the receiving
vehicle will verify this code over the duration of the VTW window such as follows:
a. At the sending vehicle, the MAC value is calculated based on hashing the vehicle VIN number, location and
size. Then the sending vehicle sends out its BSM message along with the calculated MAC value.
b. At the receiving vehicle, the OBU verifies the BSM message integrity using the incoming MAC value as well
as calculating a hash value such as follows:
i. Use the camera sensor to capture a picture for the sending vehicle,
ii. Use the picture estimation algorithm in [4] to estimate the vehicle’s size,
iii. Hash the sending vehicle’s estimated size from step (ii), VIN number, location,
iv. Compare the value calculated in above in (iii) with the incoming hash value and
v. If the two hash values match, this means that the message is coming from a trusted source; otherwise, it will be
disregarded and dropped.
Since calculating the VTW window is cornerstone for this solution framework, we will focus on implementing
the framework’s first phase in this paper. The remainder of the paper is organized as follows. Section 2
provides an overview on the BSM message format, Section 3 describes how the simulation experiments are
designed and executed. Sections 4 and 5 conclude the paper and outline the future work.
Attacker’s
Malicious
Warning
Message Attack
Vehicle
Victim
Vehicle

II. BASIC SAFETY MESSAGE (BSM)
By default, BSMs are periodically broadcasted to nearby vehicles carrying the safety messages. However, user
custom information can be added to the BSM messages using custom fields. An example for custom info is a
pull-over BSM message sent to a police vehicle to a violating vehicle. Messages that carry such sensitive info
should be verified to ensure its legitimacy while protecting the identity of the driver or vehicle [3].
According to the SAE J2735 DSRC definition, BSM message consists of data elements (DEs) and data frames
(DFs). The data frame consists of one or more data elements or other data frames. Furthermore, the BSM
message consists of two sections such as shown in table 1. The first section is a required section of any BSM
message and is known as Basic Vehicle State with a size of 39 Bytes. The second section of the BSM is optional
and contains the Vehicle Safety Extensions and the Vehicle Status data frames. In general, vehicles periodically
broadcast the first section of the BSM message only. However, in some events such as emergency braking, the
BSM message can be further described by setting the corresponding event flag in the second part of the BSM,
[1].
Table 1: Basic Safety Message (BSM)
BSM Data Item Sequence BSM
Part
Type Bytes
Message ID I Data Element 1
Message Count I Data Element 1
Temporary ID I Data Element 4
Time I Data Element 2
Latitude
PositionLocal3D
I Data Element 4
Longitude I Data Element 4
Elevation I Data Element 2
Positioning Accuracy I Data Frame 4
Transmission & Speed
Motion
I Data Frame 2
Heading I Data Element 2
Steering Wheel Angle I Data Element 1
Accelerations I Data Frame 7
Brake System Status Control I Data Frame 2
Vehicle Size Vehicle Basics I Data Frame 3
To keep track of the number of BSM messages sent out over time, the BSM message format contains a field
called MsgCount. It is simply a number that serves the same purpose as sequence numbers in traditional
networking protocols. Fig. 2 is developed by a 3D Simulink model. It displays the progression of the MsgCount
field of two transmitting vehicles over time (one vehicle is pink, the other is yellow). Fig. 2 shows that two cars
are traveling together from the right of the screen to the left, while a third one is traveling on the opposite
direction is the receiving vehicle.
Figure 2. MsgCount Field Progression over Time for Two BSM Messages
III. EXPERIMENTAL ANALYSIS
We carried out multiple simulation experiments to calculate the visibility time between two testing vehicles. The
visibility time represents the amount of time during which the transmitting vehicle was within the receiving
vehicle's range. In the next sections, we’ll introduce our simulation environment implemented by PreScan and
the details of the conducted experiments to calculate the visibility time in different settings.

3.1 Prescan Simulation Engine
PreScan is a simulation development environment that is used to simulate realistic driving conditions and allow
for testing users’ algorithms and ADAS services in realistic simulation environment. It supports
communications using various sensor technologies such as built-in vehicle camera, LIDAR, RADAR, and GPS
in V2V and V2I architecture. PreScan supports three design paradigms: model-based controller design (MIL),
real-time tests with software-in-the-loop (SIL) and hardware-in-the-loop (HIL) systems [4 and 5]. PreScan
provides a GUI that enables us to build a simulation scenario and model sensors, while the Simulink and
MATLAB interface allows us to add a control system to define the simulation building blocks to test the ADAS
application or the user algorithm. There are four steps to build and run a simulation scenario such as described
below and in Fig. 3.
a. Building the simulation scenario
A dedicated GUI is used to build and modify traffic scenarios using an existing database of road sections, trees,
buildings, traffic signs, different vehicles such as cars, trucks and various weather conditions such as rain, snow
and fog.
b. Modeling sensors
Various types of sensors such as radar, laser, camera, ultrasound, infrared, GPS and antennas for vehicle-to-X
(V2X) communication can be added with various parameters to adjust to fit the simulation scenario.
c. Adding control system
A Matlab/Simulink interface enables us to control the vehicle movement algorithm as well as sensor fusion and
adding any user custom module to perform a mathematical operation such as calculating the inter-vehicle
distance in a simulation environment and exporting collected data to output files.
d. Running the simulation experiment
A 3D visualization viewer allows users to monitor the progress of the simulation experiment as well as analyze
the obtained results.
Figure 3. PreScan Simulation Engine Life Cycle [5]
3.2 Building Simulation Experiments
PreScan has been chosen to implement the simulation environment because it supports DSRC communications.
We setup and ran PreScan and Simulink simulation experiments to determine the visibility time between two
testing vehicles using different scenarios such as described in table 2, 3, 4 and 5:
Table 2: Control Experiment Parameters
Experiment1: Control Experiment Parameters
Vehicle Size Vehicle Speed Vehicle Trajectory
Sending Vehicle Receiving Vehicle Sending
Vehicle (m/s)
Receiving
Vehicle (m/s)
Road
Length(m)
Number of
Lanes
Sedan - Citreon C3 -
Travels along westernmost
northbound lane, turns left
at intersection
Sedan - BMW Z3 - Travels
along southernmost
eastbound lane, turns left at
intersection
15 15 120 2 opposite
lanes

Table 3: Truck Experiment Parameters
Experiment 2: Truck Experiment Parameters
Sending Vehicle Receiving Vehicle Sending Vehicle
(m/s)
Receiving
Vehicle (m/s)
Road
Length(m)
Number of
Lanes
Truck/Nissan Cabstar -
Travels along westernmost
northbound lane, turns left at
intersection
Sedan - BMW Z3 Travels
along southernmost
intersection
lanes
Table 4: Five Lanes Experiment Parameters
Experiment 3: Five Lanes Experiment Parameters
Sending Vehicle Receiving Vehicle Sending
Vehicle (m/s)
Receiving
Vehicle (m/s)
Road
Length(m)
Number of
Lanes
Sedan - Citreon C3 - Travels
along westernmost northbound
lane, turns left at intersection
along southernmost
intersection
lanes
Table 5: 25 Meter per Sec Speed Experiment Parameters
Experiment 4: Twenty-Five MPS Experiment Parameters
Sending Vehicle Receiving Vehicle Sending Vehicle
(m/s)
Receiving
Vehicle (m/s)
Road
Length(m)
Number
of Lanes
Sedan - Citreon C3 - Travels
along westernmost northbound
lane, turns left at intersection
along southernmost
intersection
lanes
A snapshot from the real time simulation environment for each experiment is depicted in Fig 4.
(a) Real Time View of the Simulation Environment For the
Control Experiment
(b) Real Time View of the Simulation Environment For
the Truck Experiment
(c) Real Time View of the Simulation Environment For the
5 Lanes Experiment
(d) Real Time View of the Simulation Environment For
the 25 m/s Speed Experiment
Figure 4. PreScan Simulation Environment

The data collected from the simulation experiments are synchronous across all experiments, which means that
the time samples for all experiments match up to that of the control experiment without error. Simulation results
obtained are summarized in table 6.
Table 6: Simulation Results Summary
Experiment Type First BSM Message Time
(Sec)
Last BSM Message Time
(Sec)
Visibility Time (Sec)
Control Experiment (1) 2.2 5.90 3.75
Truck Experiment (2) 2.2 5.95 3.80
Five Lanes Experiment (3) 2.2 6.20 3.95
25 MPS Speed Experiment (4) 1.80 4.25 2.45
We have observed that, as long as the vehicles speed is the same, the visibility time is almost the same such as
shown below:
 Control experiment (2.2 5.90: 3.75Sec),
 Truck experiment (2.25.95: 3.80Sec),
 Five Lanes experiment (2.26.20: 3.95Sec),
 25m/s experiment (1.804.25: 2.45Sec).
This means that the receiving vehicle is able to detect its peer vehicle at an average visibility time of 3.83 Sec.
In other words, the two vehicles will be able to hold an active authenticated communications session for 3.83 Sec
starting from receiving the first BSM message. The only exception here is when the two vehicle are travelling
with a high speed (25m/s), the visibility time = 2.45 Sec. Therefore, the vehicle that is traveling at a higher
speed (25 m/s) was detectable for less time and they are able to hold a shorter V2V communications session. So,
when vehicles travel at the same speed for instance at (15m/s), vehicles can predict for how far and for how long
they will be engaged in an authenticated communication session. Assume that before a target vehicle is being
within range of an attack vehicle, the attack vehicle will know neither its distance from the victim vehicle nor its
victim's receiving range. Consequently, it will be very difficult for an attacker to successfully synchronize an
attack to get a spoofed fake message through without any outside help because the target vehicle is sensitive not
only to the contents of the BSM, but also to the behavior of the BSM's contents over time.
ACKNOWLEDGEMENTS
The authors would like to thank the TASS International Company in MI, USA for allowing us an opportunity to
test and practice with their PreScan Simulation environment.
IV. CONCLUSION
In this paper, authors focused on predicting the authentication session duration based on estimating the visibility
time between 2 vehicles in a V2V vehicle communication scenario. Authors introduced the visibility time
window as a measure for the authentication session duration. Authors conducted simulation experiments using
PreScan to study measure the visibility time between two vehicles in a V2V environment. The effect of several
driving environment conditions have been studied and analyzed such as vehicle size, speed, and number of road
lanes.
We found that it is fairly easy that vehicles can estimate how long their communication session can last for from
receiving the first BSM message. Any messages arrived outside the visibility time window may be ignored
Future Work
Authors would like to investigate the implementation and evaluation of the second phase of the proposed
framework.
REFERENCES
[1]. A. Svenson, G. Peredo, and L. Delgrossi (2015). Development of a Basic Safety Message for Tractor-
Trailers for Vehicle-to-Vehicle Communications. 24th International Technical Conference on the
Enhanced Safety of Vehicles (ESV), Gothenburg, Sweden
[2]. G. Corser, A. Arenas, and H. Fu (2016). Effect on vehicle safety of nonexistent or silenced basic safety
messages. 2016 International Conference on Computing, Networking and Communications (ICNC),
Kauai, Hawaii, USA.
[3]. J. Carter, and N. Paul (2015). Analysis of Vehicle-Based Security Operations. 24th International
Technical Conference on the Enhanced Safety of Vehicles (ESV), Gothenburg, Sweden

[4]. M. Huang, Visibility and confidence estimation of an onboard-camera image for an intelligent vehicle.
(2015), Accessed 2016
[5]. PreScan automotive simulation, https://www.tassinternational.com/prescan, Accesses 2016
[6]. C. Smith (2014). Car Hacker's Manual Paperback, Theia Labs

A Lightweight Message Authentication Framework in the Intelligent Vehicles System

More Related Content

Similar to A Lightweight Message Authentication Framework in the Intelligent Vehicles System (20)

Recently uploaded (20)

A Lightweight Message Authentication Framework in the Intelligent Vehicles System