VANET alert endorsement using multi-source filters

We propose a security model for Vehicular Ad-hoc Networks (VANETs) to distinguish spurious messages from legitimate messages. In this paper, we explore the information available in a VANET environment to enable vehicles to filter out malicious messages which are transmitted by a minority of misbehaving vehicles. More specifically, we introduce a message filtering model that leverages multiple complementary sources of information to construct a multi-source detection model such that drivers are only alerted after some fraction of sources agree. Our filtering model is based on two main components: a threshold curve and a Certainty of Event (CoE) curve. A threshold curve implies the importance of an event to a driver according to the relative position, and a CoE curve represents the confidence level of the received messages. An alert is triggered when the event certainty surpasses a threshold. We analyze our model and provide some initial simulation results to demonstrate the benefits.