Security Landscape of Anomaly based Defence Mechanisms in Edge Environments

Edge computing is a computational paradigm that brings computation closer to the end user. Even though real-time applications benefit from the extremely low response time of such a computing paradigm, the heterogeneity of end devices and the distributed nature of data in an edge computing environment falls prey to cyber attackers to exploit their vulnerabilities. Thus, security in edge computing is a great cause of concern that needs to be addressed. With the use of various Machine Learning(ML) and Deep Learning(DL) techniques, anomaly detection can assist in the real-time detection of security threats and operational concerns, enabling quick response. In this paper, we explore the current state of security in edge computing environments and some of the recent work done to safeguard the network using network intrusion detection systems(NIDS) with an extended focus on the use of Graph Neural Networks(GNNs). Additionally we also discuss the robustness of the NIDS by invading them with the adversarial attacks which are designed so as to fail the existing installed NIDS on the network and investigate the defence mechanisms available to safeguard them.