A network traffic data generation model based on AOT-DDPM for abnormal traffic detection

With the rapid development of the Internet, network security issues have become increasingly prominent, and accurate detection of abnormal traffic has become a research hotspot in this field. However, in the actual detection, the imbalance of traffic data leads to the low accuracy of model detection. The above problems can be effectively solved by generating traffic data. However, the existing data generation models are difficult to reflect the potential sample distribution of the generated objects and the model training is unstable, which leads to the low quality of the generated samples. To solve the above problems, this paper proposes a network traffic generation model based on AOT-DDPM. Firstly, the model proposed an adaptive sampling strategy in the reverse diffusion process, which dynamically updated the number of samples in each class by learning the distribution of the original data and the imbalance of each class to improve the quality of generated samples. Secondly, using the Transformer network as the core structure of the reverse process to effectively capture the feature distribution between data and improve the model’s data generation performance. Finally, the ODE Solver was introduced to generate data, so as to generate high-quality samples in shorter time steps, and finally obtain balanced data. Through verification on three public datasets, the experimental results show that the AOT-DDPM model is superior to other comparison methods in terms of the effectiveness of data generation, and can solve the problems of data generation in the field of network abnormal traffic detection.