Proposed data verification method capable of detecting more than 93 percent of all misbehavior attempts by using independent V2V location verification.

The method, which uses both V2V and V2I communications to validate a vehicle's indicated location, was found in a simulation to be most effective in urban areas with high traffic densities.

Date Posted
12/31/2018
Identifier
2018-B01337
TwitterLinkedInFacebook

A Method of Misbehavior Detection with Mutual Vehicle Position Monitoring

Summary Information

One potential threat to connected-vehicle networks is cloud-based attacks. Such attacks would hijack vehicle-to-cloud (V2C) communications and send falsified information, such as incurring congestion by falsely reporting traffic accidents or fabricating driving and position information.



The authors of the paper offer a method to detect misbehavior from aggregated data on a cloud server by utilizing information from surrounding vehicles. For instance, it is possible to verify the position of a vehicle that reports falsified location data through the independent verification of nearby vehicles that are within the range of vehicle-to-vehicle (V2V) communication. The method proposed within the paper combines this validation with a further cross-reference against local base stations to verify that location information from a given vehicle is accurate. If enough discrepancies are observed, the vehicle is assumed to be transmitting falsified data.



METHOD

To evaluate the effectiveness of the detection procedure, the authors performed a trial using the network simulator Scenargie. A variety of misbehavior methods were examined, to account for possible attempted attacks. The impact of changing the "threshold value," or the number of times the location-validation was performed, was examined.

FINDINGS
The method described in the paper was able to detect 93 percent of the falsified data, even when only one validation check was performed. This rate increased to 100 percent with higher threshold values.

However, the false positive rate was also found to increase with higher threshold values. This effect was most significant for lower traffic densities; when the simulation was re-run with a density value typical for urban areas, the false positive rate was sharply reduced. As misbehavior attacks would be more disruptive in densely trafficked areas, this indicates that the proposed method would be most effective in precisely the areas it would be used.

Goal Areas
Results Type
Deployment Locations