Impact of 5G Latency and Jitter on TAS Scheduling in a 5G-TSN Network: An Empirical Study

TL;DR

This study empirically evaluates how 5G latency and jitter affect TAS scheduling in TSN networks, highlighting the need for precise TAS parameter tuning to maintain deterministic industrial communications over wireless 5G links.

Contribution

It provides an experimental analysis of 5G's impact on TSN scheduling, offering guidelines for setting TAS parameters to ensure end-to-end determinism in industrial networks.

Findings

Bounded latency and jitter require careful TAS parameter tuning.

Excessive offset can cause delay or loss of determinism.

High-order percentile measurements help in setting appropriate TAS parameters.

Abstract

Deterministic communications are essential to meet the stringent delay and jitter requirements of Industrial Internet of Things (IIoT) services. IIoT increasingly demands wide-area wireless mobility to support Autonomous Mobile Robots (AMR) and dynamic workflows. Integrating Time-Sensitive Networking (TSN) with 5G private networks is emerging as a promising approach to fulfill these requirements. In this architecture, 5G provides wireless access for industrial devices, which connect to a TSN backbone that interfaces with the enterprise edge/cloud, where IIoT control and computing systems reside. TSN achieves bounded latency and low jitter using IEEE 802.1Qbv Time-Aware Shaper (TAS), which schedules the network traffic in precise time slots. However, the stochastic delay and jitter inherent in 5G disrupt TSN scheduling, requiring careful tuning of TAS parameters to maintain end-to-end determinism. This paper presents an empirical study evaluating the impact of 5G downlink delay and jitter on TAS scheduling using a testbed with TSN switches and a commercial 5G network. Results show that guaranteeing bounded latency and jitter requires careful setting of TAS transmission window offset between TSN switches based on the measured 5G delay bounded by a high order p-th percentile. Otherwise, excessive offset may cause additional delay or even a complete loss of determinism.