Citation

Al-Selwi, Hatem Fahd and Aziz, Azlan Abdul and Bin Abas, Fazly Salleh and Alias, Mohamad Yusoff and Bin Jamian, Saifulnizan and Aziz, Nor Azlina Ab. (2025) Weather-Aware Spatial-Temporal Graph Convolutional Networks for Traffic Forecasting. In: 2025 Multimedia University Engineering Conference (MECON), 21-23 July 2025, Cyberjaya, Malaysia.

Text 489.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

Vehicular traffic flow prediction is a crucial component of Intelligent Transportation Systems (ITS), essential for mitigating congestion and optimizing traffic management. Additionally, accurate traffic forecasting plays a vital role in resource allocation for vehicular communication networks, where efficient network resource distribution depends on precise predictions. While deep learning-based methods have significantly improved traffic prediction, existing models still struggle to fully capture the complex spatiotemporal dynamics of traffic data. Recent advancements have framed traffic prediction as a graph network problem, leveraging Graph Neural Networks (GNNs) to model spatial and temporal dependencies more effectively. This study proposes an enhanced Spatial-Temporal Graph Convolutional Network (ASTGCN) that integrates weather data to improve traffic flow prediction accuracy. The proposed model is evaluated on multiple datasets to assess the impact of weather conditions on predictive performance. Experimental results show that incorporating weather data leads to an average improvement of 7% over original model.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Traffic Prediction, Deep Learning, Graph Neural network, Machine learning, Connected Vehicles
Subjects:	T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL1-484 Motor vehicles. Cycles
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	17 Mar 2026 03:02
Last Modified:	19 Mar 2026 01:37
URII:	http://shdl.mmu.edu.my/id/eprint/15469

Downloads

Edit (login required)