Systematic literature review of AI/ML techniques applied to VANET routing

Abstract

Vehicular Ad Hoc Networks (VANETs) have arisen as powerful network models for road safety and infotainment applications. These networks are characterized by high node mobility, communication intermittency, and unreliability, which deteriorates the network’s performance. Articles have proposed intelligent algorithms to solve VANET’s performance concerns. This study aims to review the literature regarding intelligent VANET routing protocols, by focusing on techniques used and how their performance was assessed. We used the snowballing procedure to collect studies that propose novel solutions by means of intelligent algorithms. The 86 included studies reported that heuristics, fuzzy logic and reinforcement learning approaches are the most popular and effective methods to improve VANET routing performance. We also demonstrate that the community has yet to find a consensus as to how to evaluate routing protocol performance. We suggested the use of an evaluation and comparison framework for transparent routing protocol design and selection in future vehicle applications.