An Enhanced Intrusion Detection System Using Edge Centric Approach

Abstract

In the ever-evolving landscape of Cybersecurity, the detection and mitigation of network intrusions and anomalous activities remain formidable challenges. Conventional methods for identifying threats often encounter difficulties in scaling up and adapting swiftly, as they heavily rely on labeled network data. Furthermore, a narrow focus on individual data points may inadvertently overlook critical details at the packet level, thus exposing vulnerabilities that malicious actors can exploit. To confront these ongoing challenges head-on, Graph Neural Networks (GNNs) emerge as a promising solution. Their innate ability to comprehend complex network structures equips them with the capability to provide deeper insights into the dynamics of network traffic. By harnessing the power of GNN, it autonomously detects and comprehends intrusions and anomalies, surpassing the limitations of conventional techniques. Through experimentation and evaluation on real-world datasets, the proposed system demonstrates promising results in accurately identifying and classifying network intrusions.