Improvement of False Positives in Misbehavior Detection
Résumé
By faking vehicle information on cloud servers, an adversary may deliberately cause traffic congestion and/or accidents. Misbehavior means sending masqueraded data to cloud servers in this paper. In our previous research, we proposed ”A Method of Detecting Camouflage Data with Mutual Position Monitoring”. Cloud servers can detect masqueraded position data from malicious vehicles by increasing the threshold value of our detecting method. However, there are some problems. In this paper, we clarify what kind of malicious behavior is targeted, and we propose two new measures to address the false positives problem. First, we weight for public vehicles such as police cars, and cloud servers can trust vehicles even if they below the threshold value. Second, we dynamically determine the threshold value with consideration of vehicle density. Next, we evaluate the two methods. We find that the method of weighting for each vehicle was very effective, and the method of dynamic determination also showed good results. There is not much difference between our previous method and weighting for each vehicle at low threshold value, but this new one helps considerably suppress false positives at high threshold. The advantage of the dynamic determination model is that false positives do not depend on each base station, because the threshold is dynamically determined. This works more effectively in lower vehicle densities. Our results indicated that these two countermeasures was practical against false positives.
Origine | Fichiers éditeurs autorisés sur une archive ouverte |
---|
Loading...