Enhancing the Security of Local DangerWarnings in VANETs - A Simulative 
Analysis of Voting Schemes 

Benedikt Ostermaier Inst. for Pervasive Computing ETH Zurich 8092 Zurich, 
Switzerland ostermaier@inf.ethz.ch 

Florian Dötzer ASKON ConsultingGroup 60313 Frankfurt, Germany 
florian.doetzer@askon.de 

Markus Strassberger BMW Group Forschung und Technik 80992 Munich, Germany 
markus.strassberger@bmw.de 

The upcoming deployment of vehicular ad-hoc networks does not only facilitate 
novel telematics applications, but also poses strong requirements on security. 
Especially the adoption of active safety applications may raise new threats to 
road safety if security issues are not properly handled, thus thwarting their 
initial purpose. In this paper, a special active safety application is 
considered that enables cooperative foresighted driving through the exchange 
of local danger warnings, which are based on individual observations and refer 
to the current road condition. From a security point of view, the decision 
whether or not such an application should rely on a reported hazard, is a 
crucial issue, which cannot be completely protected by conventional security 
measures. We propose an additional security mechanism based on an information 
centric evaluation of the plausibility of received hazard messages. We 
developed four decision methods, which are based on voting schemes, and 
evaluated them by simulation using two attacks trying to manipulate the 
decision process by distributing false information. Our results indicate that 
the proposed information centric evaluation of remote observations is a 
reasonable means to increase the stability and security of a cooperative local 
danger warning service.