MECGuard: GRU enhanced attack detection in Mobile Edge Computing environment

Abstract

With the commercialization of 5G technology, various Mobile Edge Computing (MEC) services are being deployed widely. Generally, MEC services rely on MEC devices and servers deployed at the edge of the network. Whether it is a MEC device or an edge server, most of them lack computing resources, and it is difficult to implement powerful security capabilities. Moreover, there are a large number of MEC service providers, different standards, and different protocols, which extend the attack interface of MEC services. In response to this situation, this paper proposes MECGuard, an attack detection solution designed for the MEC environment based on deep learning technology. Based on its distributed architecture designed for the MEC environment, MECGuard implements a lightweight TCP-level protocol extractor based on Decision Tree, and an attack detection network based on Gated Recurrent Unit (GRU). Experiments prove that MECGuard could have a good performance of malicious traffic detection in the EMC environment.