Characterization of latency and jitter in TSN emulation

TL;DR

This paper evaluates timestamping methods in TSN emulation to accurately characterize latency and jitter, providing insights for optimizing network performance measurement in software environments.

Contribution

It introduces a methodology for software timestamping in TSN emulation, compares techniques, and offers experimental insights for improving TSN emulation accuracy.

Findings

Four timestamping techniques compared in TSN emulation

Bridge latency and jitter characterized in Mininet environment

TSN scheduling solutions validated with measured parameters

Abstract

This research focuses on timestamping methods for profiling network traffic in software-based environments. Accurate timestamping is crucial for evaluating network performance, particularly in Time-Sensitive Networking (TSN). We explore and compare four timestamping techniques within a TSN emulation context, though its findings extend to other network scenarios. The study leverages the Mininet emulator to model TSN networks, defining hosts, bridges, links, and traffic streams. It characterizes bridge latencies and jitter, solves the TSN scheduling problem based on measured parameters, and evaluates the correctness of a deployed schedule for a use case. Key contributions include a methodology for software-based timestamping, solutions for TSN emulation challenges in Linux and Mininet, and experimental insights for optimizing TSN emulation platforms on various system configurations, with and without Intel TCC, either on a high-end workstation or on an industrial PC.