Hackers easily seize control of nearly 100 traffic lights (Wired UK)


A typical intersection configuration

Halderman et al., University of Michigan


Taking over a city’s intersections and making all the
lights green to cause chaos is a pretty bog-standard Evil Techno
Bad Guy tactic on TV and in movies, but according to a research
team at the University of Michigan, doing it in real life is within
the realm of anyone with a laptop and the right kind of radio.
In a paper
published this month
, the researchers describe how they very
simply and very quickly seized control of an entire system of
almost 100 intersections in an unnamed Michigan city from a single
ingress point.

The exercise was conducted on actual stoplights deployed at live
intersections, “with cooperation from a road agency located in
Michigan.” As is typical in large urban areas, the traffic lights
in the subject city are networked in a tree-type topology, allowing
them to pass information to and receive instruction from a central
management point. The network is IP-based, with all the nodes
(intersections and management computers) on a single subnet. In
order to save on installation costs and increase flexibility, the
traffic light system uses wireless radios rather than dedicated
physical networking links for its communication infrastructure –
and that’s the hole the research team exploited.

Wireless security? What’s that?

The systems in question use a combination of 5.8GHz and 900MHz
radios, depending on the conditions at each intersection (two
intersections with a good line-of-sight to each other use 5.8GHz
because of the higher data rate, for example, while two
intersections separated by obstructions would use 900MHz). The
900MHz links use “a proprietary protocol with frequency hopping
spread-spectrum (FHSS),” but the 5.8GHz version of the proprietary
protocol isn’t terribly different from 802.11n.

In fact, using unmodified laptops and smartphones, the team was
able to see each intersection’s broadcast SSID, though they were
unable to join the networks due to the protocol differences. The
paper notes that researchers could have reverse-engineered the
protocol to connect but instead chose to simply use the same type
of custom radio for the project as the intersections use. Lest you
think that’s cheating, the paper explains the decision like
this:

“We chose to circumvent this issue and use the same model radio
that was deployed in the studied network for our attack. While
these radios are not usually sold to the public, previous work has
shown social engineering to be effective in obtaining radio
hardware [38]….Once
the network is accessed at a single point, the attacker can send
commands to any intersection on the network. This means an
adversary need only attack the weakest link in the system.”

The 5.8GHz network has no password and uses no encryption; with
a proper radio in hand, joining is trivial.


Nodes in the traffic light network are connected in a tree-topology IP network, all on the same subnet

Halderman et al., University of Michigan


Smash box

After gaining access, the next step was to be able to
communicate with the controllers that operate each intersection.
This was made easy by the fact that in this system, the
control boxes run VxWorks 5.5, a version which by default gets
built from source with a debug port left accessible for testing.
The research team quickly discovered that the debug port was open
on the live controllers and could directly “read and write
arbitrary memory locations, kill tasks, and even reboot the
device.”

Debug access to the system also let the researchers look at how
the controller communicates to its attached devices — the traffic
lights and intersection cameras. They quickly discovered that the
control system’s communication was totally non-obfuscated and easy
to understand — and easy to subvert:

“By sniffing packets sent between the controller and this
program, we discovered that communication to the controller is not
encrypted, requires no authentication, and is replayable. Using
this information, we were then able to reverse engineer parts of
the communication structure. Various command packets only differ in
the last byte, allowing an attacker to easily determine remaining
commands once one has been discovered. We created a program that
allows a user to activate any button on the controller and then
displays the results to the user. We also created a library of
commands which enable scriptable attacks. We tested this code in
the field and were able to access the controller remotely.”

Once total access to a controller and its commands was
gained, that was it — at that point, the team had full control
over every intersection in the entire network. They could change
lights at will and even control each intersection’s cameras. The
paper lays out several potential activities that an attacker could
engage in, including connecting from a moving vehicle and making
all lights along the attacker’s path green, or purposefully
snarling traffic to aid in the attacker’s escape after a crime.

More worrying is the ability of an attacker to engage in a type
of denial-of-service attack on controlled intersections by
triggering each intersection’s malfunction management unit, which
would put the lights into a failure mode — like all
directions blinking red — until physically reset. This would,
according to the paper, let “an adversary… disable traffic lights
faster than technicians can be sent to repair them.”

Mitigation

The paper closes by pointing out a number of ways in which the
gaping security holes could be easily closed. Chief among the
recommendations is some kind of wireless security; the paper points
out that the 5.8GHz systems support WPA2 encryption, and enabling
it is trivial. The 900MHz systems are more secure by virtue of not
using a frequency band easily accessible by consumer laptops and
smartphones, but they also support the older WEP and WPA wireless
encryption standards.

But a layered defense is best, and as such the paper also
recommends stricter controls over the traffic systems’ IP networks
— firewalling devices and strictly controlling the type of network
traffic allowed.

Further, though many of the components in the network support
some kind of username and password authentication scheme, the
report ominously points out that “all of the devices in the
deployment we studied used the default credentials that came built
into the device.” Doing some basic housekeeping and changing the
credentials on the VxWorks intersection controllers and the
wireless network components would go a long way toward frustrating
attacks.

Should we panic?

It’s hard to not get a little antsy when confronted with
research showing that vital pieces of public infrastructure are
sitting essentially unsecured. The paper’s conclusion is clearly
stated: “While traffic control systems may be built to fail into a
safe state, we have shown that they are not safe from attacks by a
determined adversary.” There is plenty of blame to be cast, from
the local agencies deploying infrastructure hardware in an unsafe
state to the manufacturers helping them set things up.

In fact, the most upsetting passage in the entire paper is the
dismissive response issued by the traffic controller vendor when
the research team presented its findings. According to the paper,
the vendor responsible stated that it “has followed the accepted
industry standard and it is that standard which does not include
security.”

This article originally appeared on Ars Technica.

If the article suppose to have a video or a photo gallery and it does not appear on your screen, please Click Here

21 August 2014 | 10:32 am – Source: wired.co.uk

Leave a Reply

Your email address will not be published.