Improving VANET Data Dissemination Efficiency with Deep Neural Networks

Abstract

Vehicular Ad-hoc Networks (VANETs) play a crucial role in Intelligent Transportation Systems (ITS), but their dynamic nature makes efficient data dissemination challenging. This paper proposes a novel deep learning-based method to optimize data dissemination within VANETs. A realistic dataset is generated through simulations using a modified Breadth-First Search algorithm combined with the Jaccard similarity coefficient to maximize message coverage. A deep neural network (DNN) is trained on this dataset to predict optimal forwarding paths in varying VANET conditions. Integration of this DNN-based protocol into OMNeT++ simulations demonstrates significant improvements in packet delivery ratios, reduced network overhead, and minimized transmission delays compared to existing dissemination protocols.