Novel Vehicular Compatibility-Based Ad Hoc Message Routing Scheme in the Internet of Vehicles Using Machine Learning

The sixth-generation cellular networks bring about proactive communications with predictive decision making by incorporating artificial intelligence (AI) and machine learning (ML) in vehicular networks, toward envision of the Internet of Vehicles (IoV). Currently, vehicular communications suffer from unreliable communication links due to multihop
ad hoc
communications and the high-mobility environment. The available literature falls short in providing a reliable routing scheme that proactively and accurately estimates or predicts connectivity duration between two vehicles. In this study, we highlight the need for communication route compatibility (connectivity duration) as a route selection parameter along with trustworthiness. We propose an ML and analytical compatibility-based
ad hoc
routing protocol that allows a vehicle to estimate or predict the compatibility time of all candidate routes, to choose the best route. We evaluated one analytical and five ML classification techniques on our OpenStreemMap (OSM) and SUMO mobility trace generated data set (Seoul and Berlin). Our exhaustive simulation demonstrated that our proposed scheme (six variations) dismisses all short-lived routes and achieves 2–3 times higher packet delivery ratio in comparison to the existing hop count-based routing (AOMDV and trust cryptographic secure routing). The proposed scheme disregards paths having few intermediate nodes for long-lasting paths with the expenses of a few extra hops. We also present a comprehensive comparative study to evaluate ML techniques based on the well-known metrics, such as accuracy, time, misclassification,
$F1$
-score, etc.