AI-Driven TrafNet-DCMCRP for Congestion-Aware Routing and Traffic Management in VANETs

Abstract

Efficient traffic management and congestion-aware routing are one of the main challenges in Vehicular Ad-hoc Networks (VANETs). In smart city ecosystems, there is a greater need for proper traffic management so that transportation is efficient, the rate of congestion is minimized, and road safety is improved. VANET enables vehicle to-vehicle and vehicle-to-infrastructure communication, and it is expected to be used as a potential platform for dynamic traffic management in the near future. Traditional routing methods, including static protocols or simplistic predictive models, have several limitations. These methods are usually unable to adapt to the changing traffic conditions that occur very rapidly, thus causing inefficient routing, increased congestion, and reduced communication reliability. To overcome these shortcomings, the study proposes a hybrid AI-driven framework, TrafNet: Temporal Driven Adaptive Traffic Intelligence (TCN-DQN), which integrates Temporal Convolutional Networks (TCNs) and Deep Q-Networks (DQNs). The TCN will identify congestion trends by predicting the outcome using sequential data relevant to traffic, hence long-term traffic forecasting. At the same time, the DQN employs real-time traffic conditions to learn adaptive routing policies in real time. This is further achieved with the load balancing and multi-channel communication for more reliable communication with the aid of a Dynamic Congestion-Aware Multi-Channel Routing Protocol (DCMCRP). The proposed system achieves a MAE of 0.01 and a RMSE of 0.02 and significantly outperforms existing approaches. This integration ensures proactive and smart routing of the vehicle, taking care of immediate responses and also of long-term predictions.