Deep Learning Model for Intelligent Location Prediction in VANETs : LSTM-CNN A Hybrid Approach

Abstract

This research focuses on vehicle location prediction in the context of connected and intelligent transportation systems (C-ITS), particularly wi thin vehicular ad hoc networks (VANETs). Accurate and reliable location is essential for the safety, routing, and coordination of autonomous and connec ted vehicles. However, GPS signals are often degraded or unavailable in real world scenarios such as urban canyons and tunnels. To address this challenge, we propose a hybrid architecture based on deep learning that combines convolutional neural networks (CNNs) and long short term Memory (LSTM) models. The system is trained and evaluated on syn thetic trajectory datasets simulating realistic driving patterns (circular, he lical, and L-shaped). Various error metrics such as MSE, RMSE, MAE, R² Score, RPE, and MADE are used to evaluate performance. The results demonstrate that the LSTM-CNN architecture outperforms other models in terms of localization accuracy, making it a promising solution for real-time applications in dynamic vehicular environments.