@Article{cmc.2020.09695,
AUTHOR = {Pengshou Xie, Guoqiang Ma, Tao Feng, Yan Yan, Xueming Han},
TITLE = {Behavioral Feature and Correlative Detection of Multiple Types of Node in the Internet of Vehicles},
JOURNAL = {Computers, Materials \& Continua},
VOLUME = {64},
YEAR = {2020},
NUMBER = {2},
PAGES = {1127--1137},
URL = {http://www.techscience.com/cmc/v64n2/39350},
ISSN = {1546-2226},
ABSTRACT = {Undoubtedly, uncooperative or malicious nodes threaten the safety of Internet 
of Vehicles (IoV) by destroying routing or data. To this end, some researchers have 
designed some node detection mechanisms and trust calculating algorithms based on 
some different feature parameters of IoV such as communication, data, energy, etc., to 
detect and evaluate vehicle nodes. However, it is difficult to effectively assess the trust 
level of a vehicle node only by message forwarding, data consistency, and energy 
sufficiency. In order to resolve these problems, a novel mechanism and a new trust 
calculating model is proposed in this paper. First, the four tuple method is adopted, to 
qualitatively describing various types of nodes of IoV; Second, analyzing the behavioral 
features and correlation of various nodes based on route forwarding rate, data forwarding 
rate and physical location; third, designing double layer detection feature parameters with 
the ability to detect uncooperative nodes and malicious nodes; fourth, establishing a node 
correlative detection model with a double layer structure by combining the network layer 
and the perception layer. Accordingly, we conducted simulation experiments to verify the 
accuracy and time of this detection method under different speed-rate topological 
conditions of IoV. The results show that comparing with methods which only considers 
energy or communication parameters, the method proposed in this paper has obvious 
advantages in the detection of uncooperative and malicious nodes of IoV; especially, with 
the double detection feature parameters and node correlative detection model combined, 
detection accuracy is effectively improved, and the calculation time of node detection is 
largely reduced.},
DOI = {10.32604/cmc.2020.09695}
}