Malicious Node Detection Scheme Based on Correlation of Data and Network Topology in Fog Computing-Based VANETs

Abstract

In vehicle ad hoc networks (VANETs), if a legal vehicle node becomes malicious, then it is more likely to tamper with transferred data or provide false data easily. Because the malicious node is a valid internal user in VANETs, its behavior is difficult to be detected only through some cryptographic methods. Then the behavior may cause many serious traffic accidents. Based on the available (unencrypted) data only, how to detect out the internal malicious vehicle nodes by some lightweight methods needs to be researched in VANETs. Additionally, fog computing seamlessly integrates heterogeneous computing resources widely distributed in edge networks and then provides stronger computing services for users. Therefore, in this article, we propose a malicious node detection scheme in fog computing-based VANETs, where the fog server uses the reputation calculation to score each suspicious node based on the correlation of acquired data and network topology. In our proposed scheme, we build a reputation mechanism to score each suspicious node according to the correlation between outlier detection of acquired data and influence of nodes. Based on our proposed experiments, our proposed scheme can efficiently and effectively detect out malicious vehicle nodes so that fog server can acquire more true data.