AI‐Enhanced AODV Routing for Intelligent Traffic Management in Smart City Internet of Things Networks

ABSTRACT The rapid expansion of smart cities has increased reliance on the Internet of Things (IoT) to manage urban infrastructure efficiently. However, traditional communication protocols often face challenges in dynamic, high‐density environments, particularly regarding routing stability and real‐time traffic control. To address these limitations, this study proposes an AI‐enhanced ad hoc on‐demand distance vector (AODV) routing protocol integrated with intelligent, real‐time traffic management tailored for smart city IoT networks. The system combines distributed AODV‐based communication with a gated recurrent unit (GRU) model that predicts traffic flow patterns using real‐time sensor input and historical data. GRU is selected for its computational efficiency and ability to model temporal dependencies, making it well suited for time‐sensitive smart city applications. The model is trained on the Traffic4Cast dataset—comprising real‐world spatiotemporal traffic data from major European cities—and optimized using the Adam optimizer to ensure fast convergence and high prediction accuracy. The proposed framework follows four phases: (1) traffic data acquisition and preprocessing, (2) GRU‐based traffic forecasting, (3) congestion‐aware route optimization, and (4) dynamic AODV route adjustment based on predicted traffic conditions. Simulation experiments conducted in NS‐3 demonstrate that the framework improves the packet delivery ratio by 13.8%, reduces end‐to‐end latency by 17.6%, and boosts throughput by 20.9% compared with baseline AODV implementations. Continuous real‐time integration of GRU predictions enables the system to adapt efficiently to evolving traffic conditions. This work presents a scalable, AI‐driven communication infrastructure designed to enhance urban mobility, reduce congestion, and advance sustainable smart city development.