Abstract :
[en] VANET safety applications broadcast cooperative
awareness messages (CAM) periodically to provide vehicles with
continuous updates about the surrounding traffic. The periodicity
and the spatiotemporal information contained in these messages
allow a global adversary to track vehicle movements. Many
privacy schemes have been proposed for VANET, but only few
schemes consider their impact on safety applications. Also, each
scheme is evaluated using inconsistent metrics and unrealistic
vehicle traces, which makes comparing the actual performance
of different schemes in the wild more difficult.
In this paper, we aim to fill this gap and compare different
privacy schemes not only in terms of the privacy gained but
also their impact on safety applications. A distortion-based
privacy metric is initially proposed and compared with other
popular privacy metrics showing its effectiveness in measuring
privacy. A practical safety metric which is based on Monte Carlo
analysis is then proposed to measure the QoS of two safety
applications: forward collision warning and lane change warning.
Using realistic vehicle traces, six state-of-the-art VANET privacy
schemes are evaluated and compared in terms of the proposed
privacy and safety metrics. Among the evaluated schemes, it was
found that the coordinated silent period scheme achieves the best
privacy and QoS levels but fully synchronized silence among all
vehicles is a practical challenge. The CAPS and CADS schemes
provide a practical compromise between privacy and safety since
they employ only the necessary silence periods to prevent tracking
and avoid changing pseudonyms in trivial situations.
Scopus citations®
without self-citations
37