Vehicular Ad-Hoc Networks (VANETs) play a significant role in Intelligent Transportation Systems (ITS). The vehicles exchange awareness and safety information with surrounding vehicles using IEEE 802.11p Dedicated Short Range Communications (DSRC) and IEEE 1609 Wireless Access in Vehicular Environment (WAVE)-based short-range communication networks in 5.9 GHz band. These safety messages include periodic beacon messages and event driven emergency messages that are shared in Control Channel (CCH). By collecting these messages and applying clustering we can facilitate new applications like network congestion control, traffic control etc. The partitioning-based clustering methods like k-means and the Partitioning Around Medoids (PAM) are well known in data science. This paper focuses on understanding the message clustering intuitively. The clustering quality is measured using silhouette plot and average silhouette width. The simulation is carried out in OMNeT++ and SUMO-based Veins framework. A simple approach using a package called RInside is explored for fast prototyping of machine learning algorithms in OMNeT++ simulations.

Vehicular Ad-Hoc Networks (VANETs) play a significant role in Intelligent Transportation Systems (ITS). The vehicles exchange awareness and safety information with surrounding vehicles using IEEE 802.11p Dedicated Short Range Communications (DSRC) and IEEE 1609 Wireless Access in Vehicular Environment (WAVE)-based short-range communication networks in 5.9 GHz band. These safety messages include periodic beacon messages and event driven emergency messages that are shared in Control Channel (CCH). By collecting these messages and applying clustering we can facilitate new applications like network congestion control, traffic control etc. The partitioning-based clustering methods like k-means and the Partitioning Around Medoids (PAM) are well known in data science. This paper focuses on understanding the message clustering intuitively. The clustering quality is measured using silhouette plot and average silhouette width. The simulation is carried out in OMNeT++ and SUMO-based Veins framework. A simple approach using a package called RInside is explored for fast prototyping of machine learning algorithms in OMNeT++ simulations.