Lightweight Application Identification Framework based on 1D-CNN Approach

With the rapid evolution of network scheduling, the application-driven approach is becoming increasingly important. Accurate application identification is vital in supporting this approach. However, application identification faces challenges in feature perception for applications and performance limitations in network devices. In this paper, we propose a novel lightweight CNN model called 1D-CNN, which introduces attention mechanisms to enhance accuracy. Then, we design an application identification framework based on 1D-CNN. We train the proposed 1D-CNN based on the framework. Simulation results demonstrate that the proposed method achieves a precision of 93.3% in traffic application identification tasks and reduces resource overhead by over 67%.