Modeling and Analysis of New Hybrid Clustering Technique for Vehicular Ad Hoc Network

Hazem Noori Abdulrazzak, Goh Chin Hock, Nurul Asyikin Mohamed Radzi, Nadia M.L. Tan, Chiew Foong Kwong

Research output: Journal PublicationArticlepeer-review

2 Citations (Scopus)


Many researchers have proposed algorithms to improve the network performance of vehicular ad hoc network (VANET) clustering techniques for different applications. The effectiveness of the clustering model is the most important challenge. The K-Means clustering algorithm is an effective algorithm for multi-clusters that can be used in VANETs. The problems with the K-Means algorithm concern the selection of a suitable number of clusters, the creation of a highly reliable cluster, and achieving high similarity within a cluster. To address these problems, a novel method combining a covering rough set and a K-Means clustering algorithm (RK-Means) was proposed in this paper. Firstly, RK-Means creates multi-groups of vehicles using a covering rough set based on effective parameters. Secondly, the K-value-calculating algorithm computes the optimal number of clusters. Finally, the classical K-Means algorithm is applied to create the vehicle clusters for each covering rough set group. The datasets used in this work were imported from Simulation of Urban Mobility (SUMO), representing two highway scenarios, high-density and low-density. Four evaluation indexes, namely, the root mean square error (RMSE), silhouette coefficient (SC), Davies–Bouldin (DB) index, and Dunn index (DI), were used directly to test and evaluate the results of the clustering. The evaluation process was implemented on RK-Means, K-Means++, and OK-Means models. The result of the compression showed that RK-Means had high cluster similarity, greater reliability, and error reductions of 32.5% and 24.2% compared with OK-Means and K-Means++, respectively.

Original languageEnglish
Article number4720
Issue number24
Publication statusPublished - Dec 2022


  • K-Means clustering
  • cluster evaluation
  • clustering
  • energy
  • rough set
  • unsupervised machine learning

ASJC Scopus subject areas

  • Computer Science (miscellaneous)
  • Mathematics (all)
  • Engineering (miscellaneous)


Dive into the research topics of 'Modeling and Analysis of New Hybrid Clustering Technique for Vehicular Ad Hoc Network'. Together they form a unique fingerprint.

Cite this