Sensors guard privacy
Technology Research News
In a world where sensor networking and
location tracking technology is becoming increasingly sophisticated and
prevalent, preserving privacy is an increasingly difficult challenge.
Researchers from the University of Colorado at Boulder have addressed
the problem with a way to set up networks of tiny sensors that allows
users to gain useful traffic statistics but preserves privacy by cloaking
location information for any given individual.
"We realized that privacy policies tend to get incredibly complex
because they need to define in detail under which circumstances who should
get access to what information," said Marco Gruteser, a researcher at
the University of Colorado.
At the same time, privacy is poised to get even more complicated,
said Gruteser. Researchers are developing wireless networks of tiny, cheap,
powerful sensors. "Sensor network technology promises to enable vast arrays
of sensors monitoring many aspects of our daily lives," he said.
Such high-precision data can give away the identity of the people
being monitored, said Gruteser. For example, an array of infrared sensors,
which key off the heat human bodies emit, may track movements throughout
an office building. "If such movements [were] correlated with knowledge
of individuals' office locations, most monitored subjects could be identified
simply by checking... which place they spend most of their time," Gruteser
The researchers' software uses the computational abilities built
into today's sensors to automatically adjust the precision of location
data and removes obvious identifiers like names in order to make such
correlations more difficult.
The researchers' algorithm defines the strength of privacy as
the number of people any given individual is indistinguishable from. The
software can be set to a minimum level of privacy, for instance, to make
it impossible to distinguish a given person from five of her colleagues.
"The algorithm monitors the overall number of people and adaptively changes
the precision of reported locations -- say, from a room level to a building
floor level -- to maintain the predefined minimum level of privacy," he
The researchers' key idea was realizing that networks of sensors
offer enough computational power to allow for such privacy-enhancing algorithms,
said Gruteser. The main challenge to turning sensor nodes into a trustworthy,
privacy-protecting network was designing an algorithm that distributed
the needed computation among a large number of sensors, he said.
Distributing the information among nodes yields another advantage,
said Gruteser. "If an adversary succeeds in compromising one sensor node,
he does not gain access to all information collected by the network,"
Putting privacy protection in the sensor network means there are
fewer places where information can be hacked. Such protection is usually
set up further downstream, in the applications or databases that use the
information collected by sensors. "Our work applies... anonymity or depersonalization
on-the-fly to a stream of location data before it can be stored in a database
that might be exposed to inside attacks or... inadvertent data disclosures,"
The drawback to the approach is that it is only suitable for applications
that do not require user identification and can deal with less precise
location data, said Gruteser.
Appropriate applications include monitoring the use of facilities,
tracking the availability of meeting rooms and offices, collecting retail
store traffic statistics to improve store design and product shelving,
and tracking vehicles in order to better manage traffic, he said.
Privacy-enabled sensor networks would allow service providers
to avoid privacy software and policies at the database and application
level, he said.
To ensure privacy, an organization that wishes to collect data
would have the sensor networks certified by a third-party agency, said
Gruteser. The process could be similar to the federal information processing
computer security standard for cryptographic modules, he said. "This should
increase user trust in the deployment and decrease organizations' exposure
to privacy liabilities," he said.
The scheme is a collection of techniques that have been applied
separately in other contexts, and a nice piece of design that shows how
to avoid aggregation, said Gene Spafford, a professor of computer science
at Purdue University. "This is the first time I've seen that all combined
into the context of sensor systems," he said.
The scheme addresses a general issue of privacy concerns, said
Spafford. "It is difficult to preserve privacy and also meet necessary
accuracy constraints," he said.
But preserving privacy at the sensor level may turn out to be
too vague a scheme for many applications, said Spafford. "For instance,
if I'm installing a building alarm system for after-hours use, I want
to know exactly how many people are in each room," he said.
The scheme also requires that all sensors can be trusted and that
they can't be tempered with, said Spafford. "That may or may not be realistic
in actual application," he said.
The researchers are ultimately looking to provide a way for users
to remain private if they wish to, said Gruteser. "Research in sensor
and wireless networking will, as a side-effect, dramatically increase
the potential for data collection and surveillance," he said. "Our research
seeks to provide a toolkit of techniques that enable users to protect
their location privacy when desired," he said.
The privacy-aware location sensor networks could be used practically
in 3 to 6 years, according to Gruteser.
Gruteser's research colleagues were Graham Schelle, Ashish Jain,
Rick Han and Dirk Grunwald. They presented the work at Usenix HotOS IX:
9TH Workshop on Hot Topics in Operating Systems, in Lihue, Hawaii, May
18-21, 2003. The research was funded by the National Science Foundation
Timeline: 3-6 years
TRN Categories: Cryptography and Security; and Data Acquisition;
Computers and Society
Story Type: News
Related Elements: Technical paper, "Privacy-Aware Location
Sensor Networks," presented at at Usenix HotOS IX: 9TH Workshop on Hot
Topics in Operating Systems, in Lihue, Hawaii, May 18-21, 2003, and posted
July 16/23, 2003
Sensors guard privacy
link on tap
Logic clicks with
3D display goes vertical
Gel yields nanotube
gains carbon cones
Jolts mix micro fluids
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link