Adaptive Threshold for Unknown Traffic Identification

Network traffic classification has become an important foundation of network security. However, as the types of protocols and applications of the network continue to increase, unknown network traffic is also emerging. In the face of unknown network threats, how to identify unknown network threats in a complex network environment to make corresponding preparations in advance has become extremely important. Aiming at the influence of unknown traffic on classification accuracy in the prediction process, this paper proposes an Adaptive Threshold for Unknown Traffic Identification (AT-UTI) algorithm using particle swarm optimization algorithm to optimize the search of the set threshold, to reduce the impact of unknown traffic on the accuracy of the model. We evaluated our model achieving an accuracy of 93.27%. Our results demonstrate the effectiveness of AT-UTI in unknown traffic identification.