A novel method for mobile application recognition in encrypted channels

In the field of mobile application traffic analysis, existing methods for accurately identifying encrypted traffic often encounter challenges due to the widespread adoption of encryption channels and the presence of background traffic. Consequently, this study presents a novel mobile application traffic identification model that is in encrypted channels. The proposed model utilizes an adaptive feature extraction technique that combines Convolutional Neural Networks (CNNs) and Gated Recurrent Units (GRUs) to effectively extract abstract features from encrypted mobile application traffic. Additionally, by employing a probability-based comprehensive analysis to filter out low-confidence background traffic interference, the reliability of recognition is further enhanced. Experimental comparisons are conducted to validate the efficacy of the proposed approach. The results demonstrate that the proposed method achieves a remarkable classification accuracy of 95.4% when confronted with background traffic interference, surpassing existing techniques by over 15% in terms of anti-interference performance.