Normalized multi-dimensional parameter based affinity propagation clustering for cellular V2X

Abstract

This paper introduces a novel Vehicular Ad Hoc Networks (VANET) clustering scheme, titled “Normalized MultiDimensional Parameter based Affinity Propagation Clustering (NMDP-APC).” The similarity function of NMPD-APC consists of normalized multi-dimensional parameters in order to represent VANET's motion dynamics. This similarity function is the key in the Affinity Propagation Clustering (APC) scheme to adequately representing the homogeneity of traffic dynamics of VANET. In contrast to the previous research on APC scheme for VANET, our proposal does not associate any un-stable future prediction term. We also applied recently introduced Cellular V2X radio capability in this study, which significantly contributed to the reduction of communication latency. In conducting simulations, Gazis-Herman-Rothery (GHR) car following model was applied on the real sampled traffic data. The simulation resulted remarkable effects on the normalized multidimensional parameters along with successful VANET clustering. The simulation also demonstrated required minimum number of iterations for the clustering and controllability of clustering granularity in the proposed NMDP-APC scheme on the real traffic data.