Classification of IPSec VPN Encrypted Traffic Based on a Hybrid Method

Internet traffic classification has gained significance with the rapid growth of Internet networks and online applications. In this paper, we propose the IPSO-AlexNet-LSTM model, which effectively distinguishes encrypted network traffic. By combining the payload features and time-flow features of packets into feature vectors, AlexNet is used to extract the static features of IPSec VPN packets, and then the effectiveness of Long Short-Term Memory (LSTM) networks is utilized to capture the temporal dependencies of the packets. To further optimize model performance, an improved Particle Swarm Optimization (PSO) heuristic algorithm is employed to fine-tune the hyperparameters of LSTM, ensuring enhanced robustness and generalization when handling the random and dynamic behaviors of malicious attacks. Finally, the experimental results demonstrate that the proposed approach successfully identifies IPSec VPN application traffic types and achieves a balanced trade-off between classification accuracy and speed, enabling efficient recognition of IPSec VPN encrypted traffic.