AI-Enabled VANET: Enhancing The QoS Performance Over Traditional Vehicular Network

Vehicular Ad-hoc Networks (VANETs) integrated with Intelligent Transportation Systems (ITS) support real-time communication to improve road safety, traffic control, and driving performance. However, traditional routing protocols often struggle in dynamic VANET environments due to high mobility and topology changes, leading to degraded Quality of Service (QoS). This study proposes a Reinforcement Learning (RL)-based enhancement to the Optimized Link State Routing (OLSR) protocol to address these challenges. Using a realistic intersection scenario at UiTM Section 7, Shah Alam, and tools such as JOSM, SUMO, and NS-2, simulations were conducted across various network densities (50 to 250 nodes). The proposed AI-enhanced OLSR protocol, which features adaptive Multipoint Relay (MPR) selection through reinforcement learning, was evaluated in comparison with the traditional Ad-hoc On-Demand Distance Vector (AODV) protocol. Results demonstrate that the RLintegrated OLSR significantly improves throughput by approximately 4.76 times, increases packet delivery ratio by $0.239 \%$, and reduces end-to-end delay by over 5.5 times. These results underscore the potential of AI-driven routing to transform VANET performance under real-world conditions. The integration of AI enables real-time adaptation to changing network conditions, leading to more reliable and efficient communication. This research confirms the effectiveness of incorporating artificial intelligence into VANET routing, offering a robust solution for dynamic vehicular networks and contributing to the advancement of intelligent transportation systems.