Lightweight Traffic Classification Model Based on Deep Learning

The development of mobile computing and the Internet Things (IoT) has led to a surge in traffic volume, which creates heavy burden for efficient network management. management requires high computational overheads make classification, is even worse when edge networks; existing approaches sacrifice efficiency obtain high-precision classification results, are no longer suitable limited resources scenario. Given problem, generally huge parameters especially complexity. We propose lightweight model based on Mobilenetv3 improve it an ingenious balance between performance lightweight. Firstly, we adjust scale, width, resolution substantially reduce number computations. Secondly, embed precise spatial information attention mechanism enhance flow-level feature extraction capability. Thirdly, use multiscale fusion features traffic. Experiments show that our excellent accuracy operational efficiency. designed work reached more than 99.82%, parameter computation amount significantly reduced 0.26 M 5.26 M. In addition, simulation experiments Raspberry Pi prove proposed can realize real-time capability network.