5G URLLC: A Case Study on Low-Latency Intrusion Prevention
Sebastian Gallenmüller, Johannes Naab, Iris Adam, Georg Carle
Abstract
5G introduces different communication types to support novel applications, e.g., industrial control, depending on ultra-reliable, low-latency communication (URLLC) to provide their service. To realize this type of communication, network operators offer virtual end-to-end networks, called network slices, to their customers, and service level agreements of the operators promise a certain quality of service for the provided network slices. The main challenge is the creation of network slices that adhere to these strict requirements despite virtualized resources shared across different network slices. In this article, we analyze the latency performance of typical virtualized network functions. Based on these results, we derive guidelines to lower latency and propose a system architecture for hosting low-latency network functions. We measure the latency performance of a security network function, an intrusion prevention system based on Snort 3, and demonstrate that URLLC-compliant latency performance is achievable. Our entire architecture relies on off-the-shelf hardware and widely adopted software components, making our findings highly applicable to situations where low latency is crucial. All artifacts used in this article, the investigated software, the pcap traces, and the experiments scripts are publicly available at https://gallenmu. github.io/low-latency/.