FENet/IP: Uncovering the Fine-Grained Structure in IP Addresses
Fei Du, Xiuguo Bao, Yongzheng Zhang · 2019
Classifying IP addresses based on network traffic behavior is essential, such as network measurement, network QoS, and network security. Many previous studies have focused on coarse-grained classification, and these do not meet the increasingly diverse needs of applications. In this paper, we introduce FENet/IP, a novel approach for uncovering Internet address structure via learning the continuous feature representation of IP device's connection patterns, employ a combination of low-dimensional embedding and machine learning techniques to probabilistically model IP addresses. Our approach is useful in exposing the fine-grained structural characteristics of the existing Internet address space. Experimental results show the effectiveness of FENet/IP over existing state-of-the-art techniques. In several real-world networks from active IP addresses, we have achieved very high classification accuracy and stability.