![](bg2.png)
regular or even periodic updates issued from a cen-
tral authority to the vehicle fleet. While it may be
possible to achieve two of these goals simultaneously,
achieving all three appears fundamentally difficult.
The implication of this is twofold. First, we be-
lieve this should motivate designers to reconsider the
deployment of cryptographic authentication infrastruc-
ture, which can be very costly.
1
Secondly, we believe
that manufacturers should begin to develop alternative
techniques that filter out potentially malicious V2V
communications by examining the content of the mes-
sages, rather than by relying on authentication data.
1.1 Cryptographic Authentication for
V2V
A secure V2V communications system consists of sev-
eral components. The most important is the On Board
Equipment (OBE), a specialized computer that resides
within each vehicle. This computer is connected to a
radio transceiver and a Global Positioning System unit.
In current V2V proposals, the OBE transmits “basic
safety messages” that embed the vehicle’s exact posi-
tion and trajectory. These messages may be received
by nearby vehicles, and used to display messages to
the driver and/or (in future autonomous vehicles) assist
in making driving decisions.
Because these messages may have safety impli-
cations, it is important to provide a means by which
vehicles can distinguish authentic messages from mali-
cious transmissions sent by unauthorized transmitters.
The canonical method for cryptographic authentication
uses public key digital signatures. In this scheme, users
possess a public key and a secret key, where the former
should be shared and the latter is hidden. Users may
employ the secret key to sign arbitrary messages. The
resulting signature can be verified using the public key.
The fundamental challenge with authentication is dis-
tributing the public keys in a way that users are certain
who a public key belongs to. A common strategy is to
use certificates. A certificate is a digital credential that
contains the public key of a user, and is in turn signed
by a trusted authority known as a Certificate Authority.
Certificates have a set time period during which they
are considered valid. In some systems, there is a revo-
cation process in place to invalidate certificates so the
system can control who can send authentic messages.
The security of this approach therefore makes a
key assumption: the key material needed to generate
digital signatures will be available only to approved
OBE devices, and will not be easy for a malicious party
1
One proposal, called SCMS (Hehn et al., 2014), is ex-
pected to cost approximately $4 billion USD to deploy in
the United States.
to extract from an OBE and duplicate. Because this
last assumption is difficult to guarantee across millions
of vehicles, this motivates a final requirement: if the
cryptographic keys are extracted from an authorized
OBE device by a malicious party, there exists a means
to identify the invalid messages, and disable the stolen
keys.
To make this effort more challenging, modern V2V
security proposals add two additional requirements.
First, individual messages sent by vehicles should not
uniquely identify the vehicle. Moreover, it should be
challenging to link two messages sent by the same
vehicle at different locations. This requirement is in-
tended to prevent the use of V2V communications as
a means to track the location of vehicles. This pri-
vacy goal has been identified a critical requirement of
deployed V2V security systems, and accounts for a
substantial degree of the complexity of deployed pro-
posals such as the U.S. government’s proposed SCMS
system (Hehn et al., 2014).
As a final requirement, today’s V2V systems as-
sume that network connectivity (from vehicles to the
Internet, or to centralized authorities) is fundamentally
unreliable. That is, many vehicles will not be able
to connect routinely to a central authority in order to
obtain additional key material. A V2V security system
must function even without access to a reliable cellular
network. We refer to this final property as robustness.
2 TRADEOFFS
As discussed above, an intelligent vehicular system
must satisfy several distinct requirements that may not
be easy be achieve simultaneously. We first enumerate
these requirements, which we refer to as authenticity,
privacy and robustness.
Authenticity.
This is the property that communica-
tions between vehicles are trustworthy. Messages
received from other vehicles are from exactly who
they say they are from, and all the contents of the
messages are accurate. Vehicles that are found to
be misbehaving are detectable or removed from
the network.
Privacy.
The privacy requirement implies that vehi-
cles are able to communicate without revealing
information that could be used for tracking vehi-
cles. In a private system, transmissions should
not contain uniquely vehicle-identifying informa-
tion, and multiple messages from the same vehicle
should not be linkable. A common benchmark is
that the system should not allow an adversary per-
form a stronger attack on privacy than it would be
Don’t Talk to Strangers
523