Publication:

Network traffic analysis through node behaviour classification: a graph-based approach with temporal dissection and data-level preprocessing

Network traffic analysis is an important cybersecurity task, which helps to classify anomalous, potentially dangerous connections. In many cases, it is critical not only to detect individual malicious connections, but to detect which node in a network has generated malicious traffic so that appropriate actions can be taken to reduce the threat and increase the system's cybersecurity. Instead of analysing connections only, node behavioural analysis can be performed by exploiting the graph information encoded in a connection network. Network traffic, however, is temporal data and extracting graph information without a fixed time scope may only unveil macro-dynamics that are less related to cybersecurity threats. To address these issues, a threefold approach is proposed here: firstly, temporal dissection for extracting graph-based information is applied. As the resulting graphs are typically affected by class imbalance (i.e. malicious nodes are under-represented), two novel graph data-level preprocessing techniques - R-hybrid and SM-hybrid - are introduced, which focus on exploiting the most relevant graph substructures. Finally, a Neural Network (NN) and two Graph Convolutional Network (GCN) approaches are compared when performing node behaviour classification. Furthermore, we compare the node classification performance of these supervised models with traditional unsupervised anomaly detection techniques. Results show that temporal dissection parameters affected classification performance, while the data-level preprocessing strategies reduced class imbalance and led to improved supervised node behaviour classification, outperforming anomaly detection models. In particular, Neural Network (NN) outperformed Graph Convolutional Network (GCN) approaches for two attack families and was less affected by class imbalance, yet one GCN performed best overall. The presented study successfully applies a temporal graph-based approach for malicious actor detection in network traffic data.