8/31/2009 8:00am-10:00am ECOT 831
|
Quantifying and Improving Wireless Privacy
Computer Science PhD Candidate
Wireless devices have become ubiquitous in the daily lives of most people,
and thus magnify their potential to harm users' privacy. The main goal of this
dissertation is to identify and quantify privacy vulnerabilities in common
mobile wireless devices and present practical methods to mitigate them.
In this dissertation a holistic approach to achieve the goal of improving
wireless privacy is taken, not focusing on a single layer, but focusing on the
two key substrate layers, the link-layer and the physical-layer, and their
interactions. In order to make this problem more concrete this investigation
focus on 802.11, which is the dominant short range wireless protocol.
However, the solutions are general enough to be applied to most wireless
networks.
To understand the nature and different types of privacy threats, a detailed
measurement of link-layer and physical-layer behavior of wireless devices is
conducted. Using these measurements, it is demonstrated that wireless device
drivers can be fingerprinted quickly and accurately using link-layer
information, and that messages can be correlated back to their transmitting
device using physical-layer information, even if there are no identifiers
present at the link-layer. In response to these threats, first the design and
build a prototype implementation of a wireless protocol that encrypts all bits
transmitted at the link-layer is presented. Second, this work presents
techniques to mitigating physical layer-privacy threats and quantify their
effectiveness. Finally, it presents and evaluates methods to facilitate the
establishment of trust in wireless protocols without requiring an out-of-band
exchange of information. The key contribution of this dissertation is to
present a framework of practical and generalized solutions that improve
wireless privacy as a whole, and that can be applied to current and future
wireless protocols.
|