Traffic Generation

Traffic generation research focuses on creating realistic synthetic traffic data for applications like network security testing, intelligent transportation systems, and autonomous vehicle development. Current efforts leverage deep learning models, particularly diffusion models and transformers, often incorporating graph convolutional networks to capture spatial relationships in road networks and language models to incorporate textual descriptions of traffic scenarios. This work aims to improve the realism and controllability of generated traffic, addressing limitations of previous methods in handling long sequences, unusual events, and diverse traffic patterns, ultimately leading to more robust and reliable systems in various domains.