Open Access

ARTICLE

Graph Neural Network-Assisted Lion Swarm Optimization for Traffic Congestion Prediction in Intelligent Urban Mobility Systems

Meshari D. Alanazi¹, Gehan Elsayed^2,*, Turki M. Alanazi³, Anis Sahbani⁴, Amr Yousef^5,6

1 Department of Electrical Engineering, College of Engineering, Jouf University, Sakakah, 72388, Saudi Arabia
2 Department of Interior Design, College of Engineering, Jouf University, Sakakah, 72388, Saudi Arabia
3 Department of Electrical Engineering, College of Engineering, University of Hafr Al Batin, Hafr Al Batin, 39524, Saudi Arabia
4 Institute for Intelligent Systems and Robotics (ISIR), CNRS, Sorbonne University, Paris, 75006, France
5 Electrical Engineering Department, University of Business and Technology, Ar Rawdah, Jeddah, 23435, Saudi Arabia
6 Engineering Mathematics Department, Alexandria University, Lotfy El-Sied st. off Gamal Abd El-Naser, Alexandria, 11432, Egypt

* Corresponding Author: Gehan Elsayed. Email:

(This article belongs to the Special Issue: Machine Learning and Deep Learning-Based Pattern Recognition)

Computer Modeling in Engineering & Sciences 2025, 145(2), 2277-2309. https://doi.org/10.32604/cmes.2025.070726

Received 22 July 2025; Accepted 11 October 2025; Issue published 26 November 2025

Abstract

Traffic congestion plays a significant role in intelligent transportation systems (ITS) due to rapid urbanization and increased vehicle concentration. The congestion is dependent on multiple factors, such as limited road occupancy and vehicle density. Therefore, the transportation system requires an effective prediction model to reduce congestion issues in a dynamic environment. Conventional prediction systems face difficulties in identifying highly congested areas, which leads to reduced prediction accuracy. The problem is addressed by integrating Graph Neural Networks (GNN) with the Lion Swarm Optimization (LSO) framework to tackle the congestion prediction problem. Initially, the traffic information is collected and processed through a normalization process to scale the data and mitigate issues of overfitting and high dimensionality. Then, the traffic flow and temporal characteristic features are extracted to identify the connectivity of the road segment. From the connectivity and node relationship graph, modeling improves the overall prediction accuracy. During the analysis, the lion swarm optimization process utilizes the concepts of exploration and exploitation to understand the complex traffic dependencies, which helps predict high congestion on roads with minimal deviation errors. There are three core optimization phases: roaming, hunting, and migration, which enable the framework to make dynamic adjustments to enhance the predictions. The framework’s efficacy is evaluated using benchmark datasets, where the proposed work achieves 99.2% accuracy and minimizes the prediction deviation value by up to 2.5% compared to other methods. With the new framework, there was a more accurate prediction of real-time congestion, lower computational cost, and improved regulation of traffic flow. This system is easily implemented in intelligent transportation systems, smart cities, and self-driving cars, providing a robust and scalable solution for future traffic management.

---

Keywords

---