STRONG CONFRONTATION TRAFFIC CLASSIFICATION METHOD BASED ON GRAPH NEURAL NETWORK

In strong confrontation communication environment, data packets are easy to be damaged or lost, which will greatly reduce the the reliability of network traffic classification. The traditional traffic classification method solves the problem of traffic classification to a certain extent, but it is difficult to accurately describe the classification characteristics of traffic under the condition of strong countermeasure communication. Aimed at the above problems, a strong adversarial network traffic classification method based on graph neural network is proposed. The data in the complete traffic was used to randomly extract the data to form a stream to simulate the transmission of data packet damage or loss. The graph structure was used to represent the flow, and three different methods were used as the vertex characteristics of the graph, and the constituent graph was passed into the graph neural network for classification. Experiments were carried out on the real network traffic dataset. The experimental results show that the proposed method is better than the existing methods in terms of classification accuracy and has stronger stability.