Centre based evolving clustering framework with extended mobility features for vehicular ad-hoc networks

• Main Author: Talib, Mohammed Saad
• Format: Thesis
• Language: English; English
• Published: 2021
• Subjects: T Technology (General); TE Highway engineering. Roads and pavements
• Online Access: http://eprints.utem.edu.my/id/eprint/26073/1/Centre%20based%20evolving%20clustering%20framework%20with%20extended%20mobility%20features%20for%20vehicular%20ad-hoc%20networks.pdf; http://eprints.utem.edu.my/id/eprint/26073/2/Centre%20based%20evolving%20clustering%20framework%20with%20extended%20mobility%20features%20for%20vehicular%20ad-hoc%20networks.pdf; http://eprints.utem.edu.my/id/eprint/26073/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=121253

Vehicular Ad hoc Network (VANET) clustering is an active research area where a group of connected vehicles forms an ad hoc network. A stable cluster is essential for routing and data dissemination in VANET to avoid various issues such as packet loss, broadcast storm, and increased overhead resulting in an unstable clustering. Therefore, clustering is regarded as an essential part of the hierarchy of intelligent transportation systems. The literature contains numerous approaches for VANETs clustering. The majority of the approaches follow heuristic-based protocol combined with various connected phases and processes, such as cluster formation, cluster head selection, and cluster maintenance. Due to the high mobility of vehicles in VANET, it is more attractive to adapt the evolving data clustering to an evolving VANET clustering framework. The inclusion of extended mobility features has not been observed in most of the clustering approaches. The required extended mobility features are essential to overcome the challenges of vehicle movement. Moreover, relying on the non-valid assumptions such as the nature of the spherical cluster and the pre-knowledge about the number of clusters may not be feasible in many cases. In addition, most of VANETs clustering approaches use simple evaluation methodology where most of the approaches disregard a significant issue in the evaluation methodology. This thesis presents VANETs clustering framework called Centre-based Evolving Clustering with Grid Partitioning (CEC-GP). This framework uses an evolving data clustering algorithm by adopting the concept of grid granularity to capture the features of a cluster more efficiently. CEC-GP includes extended mobility features and provides the capability to avoid spherical assumptions for clusters, which is employed in most of the distance-based clustering. Besides, this framework offers high performance even with the challenging and high mobility scenarios related to the variability of mobility behaviour. The developed CEC-GP also includes an integrated approach that combined all clustering tasks such as cluster formation, cluster head selection, and cluster maintenance. Finally, CEC-GP shows a better stability performance compared with "Centre-based Stable Clustering (CBSC)" and "Evolving Data Clustering Algorithm (EDCA)" based on different performance metrics such as the clustering efficiency, the cluster head, and cluster member duration, the cluster head change rate, and the number of created clusters. The performance evaluation results show CEC-GP is better compared with the other two benchmarks in term of stability and consistency.