Deterministic Ethernet Communication for Time-sensitive Applications

Author: Jonas Gesch

Job Title: Field Application Engineer
 

Feature Introduction – TSN Part 1

Introduction

The key to success for modern industrial automation has always depended on timely and reliable data transfer. As factories migrate toward Industry 4.0 architectures – integrating motion control, computer vision, and edge-AI analytical systems onto shared networks – those demands continue to increase. Critical control loops require strict execution boundaries which is why delayed or dropped control frames can halt entire production lines. Therefore, modern real-time systems introduce stricter network requirements to ensure they stay operational and – most importantly – within safe parameters.
Time-Sensitive Networking (TSN) addresses this challenge by introducing deterministic real-time communication capabilities to Layer 2 Ethernet networks, providing:
• Precise time synchronization
• Bounded low latency
• Minimal transmission jitter
• Zero congestion packet loss

This is Part 1 of a two-part article and introduces the concept of TSN and provides an overview on TSN standards and how they can benefit industrial Ethernet networks with hard real-time requirements. It also discusses IEEE 802.1AS gPTP in more detail with other TSN features following in Part 2.

Standard Non-Deterministic Ethernet vs. Hard Real-Time Requirements
After it was introduced about 4 decades ago Ethernet has been adopted rapidly, becoming the most predominant Layer 2 protocol in local networks – which it remains to this day. It is designed to deliver data traffic according to the best-effort principle, meaning that data packets are sent out as soon as possible, but without any guarantee when (or even if) they arrive. Industrial Ethernet switches operate using a store-and-forward mechanism with first-in-first-out buffers as displayed in Figure 1. First, all incoming packets are typically queued into an ingress buffer, after which they enter the switch matrix that directs the packets to one or multiple egress buffers according to the packet’s destination. How much packet delay is introduced between the ingress port and the egress port of the Layer 2 switch mainly depends on the buffer state. With multiple ports forwarding ingress packets to a single port, the egress buffer can fill up quickly, potentially leading to packet drops if data cannot be sent out quickly enough.
 

 

Even when using QoS features to prioritize mission-critical data traffic, timely frame delivery cannot be guaranteed. A high-priority frame can still be delayed by more than 700 µs on a 100 Mbps link if a 9 KB Jumbo frame has already started transmitting on the physical layer. As a result, the time needed to send a data packet from device A to device B can vary significantly, depending on the current network utilization.
This makes deterministic communication over standard Ethernet impossible, which poses a major obstacle to industrial networks with hard real-time requirements. Some devices rely on exact timing for control message input or stop functioning completely if critical frames are delayed or dropped. Even though several Ethernet-based industrial communication protocols like EtherNet/IP, PROFINET or EtherCAT addressed this by improving the coordination, reliability and forwarding delay between communication partners, those protocols either still suffer from the drawbacks inherent to store-and-forward switch architectures or use specialized ASIC hardware, requiring dedicated network segments that are isolated from standard Ethernet traffic, which adds architectural complexity when integrating IT/OT infrastructure.

Guaranteed Network Performance With High-Priority TSN Streams
Time-Sensitive Networking (TSN) has been specifically designed to meet those hard real-time requirements and deliver deterministic communication that is timely, stable, and predictable, while also sharing the same network with standard Ethernet traffic. TSN achieves deterministic communication by synchronizing timing, scheduling critical traffic, and reserving bandwidth to ensure key control data behaves more like a punctual train service rather than regular road traffic. It evolved from the IEEE Audio Video Bridging (AVB) Task Group out of the necessity to send Audio/Video streaming over Ethernet with strict real-time performance. Since then TSN has been adopted across a wide range of applications including automotive, semiconductors, and factory automation among others and has been adopted by large organizations like the CC-Link Partner Association (CLPA) who incorporated it into their CC-Link IE TSN factory automation concept.¹
To allow time-sensitive data traffic to share the same physical network infrastructure with standard Ethernet traffic, TSN introduces the concept of high-priority streams as shown in Figure 2. A stream is a unidirectional flow of data between a talker and a listener, which carries time-critical data and uses TSN features to guarantee predictable data delivery. TSN streams can also be grouped in stream collections to share configurations for per-stream TSN features like PSFP or FRER.
 

 

Standard Ethernet traffic will simply be treated as best-effort and does not utilize TSN features. This ensures that standard Ethernet devices can share the same network infrastructure with time-sensitive devices without interfering with the high-priority traffic, provided that every Ethernet switch on the path between talker and listener is TSN-capable, such as the 9561-8GT4XS-TSN industrial Ethernet switch offered by Volktek.

The Key TSN Standards That Enable Deterministic Ethernet
TSN is not defined as a single standard but instead consists of multiple standards from the IEEE 802.1 group. As a result, not all network devices that claim to be TSN-capable necessarily support the same standards, which is why special attention is warranted when comparing TSN standard compliance between devices, especially from different manufacturers. Figure 3 shows an overview of the key TSN standards as well as their relation to each other. 

 

 

The most important standard in the TSN family is IEEE 802.1AS, which defines the Generalized Precision Time Protocol (gPTP). It provides the nanosecond-level clock synchronization that is required for time-scheduled TSN algorithms. IEEE 802.1Qbv introduces the Time-Aware Shaper (TAS), which schedules the high-priority TSN streams and makes sure the real-time requirements are met. The IEEE 802.1Qbu standard defines mechanisms for Frame Preemption to increase bandwidth efficiency that might be compromised when using TAS. It is not technically dependent on IEEE 802.1Qbv but achieves the greatest effect when used in conjunction with it. The same applies to IEEE 802.1Qci, which can enforce the egress scheduling performed with TAS at the downstream ingress port using Per Stream Filtering and Policing (PSFP). Furthermore, it ensures that TSN streams comply with specified limits for bandwidth and frame sizes to protect the high-priority streams from interference through rogue or malfunctioning nodes. Another TSN standard which operates independently from the scheduling algorithms is IEEE 802.1CB. Frame Replication and Elimination for Redundancy (FRER) replicates frames from TSN streams and transmits them over several redundant paths to achieve zero failover-time redundancy. 

The TSN Data Path
In its core architecture the data path in TSN switches still resembles the store-and-forward principle, which is the reason why they can share the same network with standard Ethernet traffic. To guarantee real-time performance for high-priority streams, TSN eliminates the non-deterministic queuing jitter inherent to standard store-and-forward architectures by inserting TSN mechanisms into the data path that manipulate the frame forwarding.
When looking at the TSN data path in Figure 4 from ingress port to egress port, PSFP will be applied first, even before frames enter the ingress queue. Should a frame violate any filters set with PSFP, it will immediately be dropped before affecting any downstream traffic. Then FRER eliminates any duplicate frames at the ingress stage. After the frames pass through the ingress queue and into the switch matrix, FRER replicates frames to send them onto redundant egress paths. After entering the egress queue, TAS gates control the exact timing when each of the queues are allowed to transmit frames to schedule dedicated interference-free transmission windows for high-priority frames. Finally, Frame Preemption operates at the egress port MAC layer where it can interrupt normal traffic to give priority to time-critical traffic. 

 

 

Part 2 of this article will go more in depth about the capabilities of these features and what scenarios they might be applied in, while the following section will discuss how gPTP forms the foundation of the real-time capabilities of TSN and its time-scheduled features.

 

IEEE 802.1AS gPTP – Nanosecond-level Network Time Synchronization
The Generalized Precision Time Protocol (gPTP) provides the nanosecond-level synchronization required for time-scheduled TSN features to function across a distributed network. As visualized in Figure 4, it operates by electing a Grandmaster Clock that distributes a common time reference to all TSN nodes via specialized synchronization packets and path delay measurements. The Time-Aware Switches require the precise clock synchronization to coordinate time-scheduled TSN features it can also be utilized in End Stations for time-critical command execution or highly accurate timestamping necessary for sensor fusion. 
 

 

The time reference is propagated through the TSN network by the Time-Aware Switches which act as gPTP Bridges that assume the master role for clock synchronization with downstream devices similar to Boundary Clocks in the original IEEE 1588 standard (PTP). Figure 6 shows the synchronization handshake performed between the Master and Slave clocks by measuring the Peer-to-Peer delay (Pdelay). It is worth noting that unlike PTP, gPTP does not allow the use of Transparent Clocks that simply relay the time messages sent from the Grandmaster Clock after adding the delay measurement. This improves synchronization accuracy² and scalability since the Grandmaster Clock only needs to perform the Pdelay measurements with its direct neighbors instead of End-to-End time synchronization with every single End Station³, 

 

 

Additionally, the Neighbor Rate Ratio keeps track of the clock drift between peers caused by the relative difference in oscillator frequencies. With this value, TSN devices can automatically correct local clock drift to keep their network time accurate between Grandmaster synchronization messages.⁵ The updated 802.1AS-2020 revision further added support for multiple time domains, allowing multiple gPTP instances to share the same physical interfaces. Since this can significantly increase the amount of Pdelay measurements that need to be performed, it also introduced the Common Mean Link Delay Service (CMLDS), which allows sharing Pdelay measurement results for one port across all gPTP instances.⁶ Avoiding redundant Pdelay measurements reduces both CPU load and bandwidth utilization.

Why gPTP (802.1AS-2020) Matters:
• Nanosecond-level time synchronization: Establishes a unified, highly precise clock reference across the entire network.
• Foundation for time-aware features: Provides the strict timing required for scheduled packet forwarding in Time-Aware Switches.
• Deterministic End Station control: Enables precise timestamping for time-critical command execution and multi-sensor data fusion.
• Multi-domain isolation: Supports multiple isolated clock domains over shared physical hardware with minimum overhead.

Outlook
While Part 1 covered the motivation behind TSN, its core standard architecture, and the time-synchronization foundation laid by gPTP (IEEE 802.1AS), Part 2 dives into the operational scheduling, ingress filtering, and redundancy features: TAS, Frame Preemption, PSFP, and FRER.
To see how these standards are implemented in industrial hardware, explore Volktek’s industrial TSN switch portfolio, including the 9561-8GT4XS-TSN and the btPoE-enabled 9561-8GP4XS-TSN.

FAQ
Q1: What differentiates TSN from other Ethernet-based real-time protocols like EtherCAT or PROFINET IRT?
Answer: EtherCAT and PROFINET IRT implement specialized ASIC hardware to meet real-time network performance. Transmitting standard Ethernet frames over this hardware can disrupt the real-time capabilities and therefore those protocols are typically deployed in isolated network segments. In contrast, TSN is designed to support standard Ethernet traffic alongside traffic with hard real-time requirements, allowing more open network architectures.

Q2: Can standard legacy Ethernet devices be used in a TSN network?
Answer: Yes. Non-TSN Ethernet traffic is treated as best-effort and processed using standard store-and-forward behavior alongside high-priority TSN streams. However, to guarantee real-time bounded latency end-to-end, every switch along that specific TSN stream’s path must be TSN-capable. Placing a non-TSN legacy switch in the middle of a stream path introduces unmanaged queuing jitter and breaks gPTP time synchronization.

Q3: What is the primary difference between time synchronization using standard IEEE 1588 and 802.1AS gPTP?
Answer: While IEEE 802.1AS-2020 is derived from IEEE 1588, gPTP enforces specific mandates for network simplicity and performance: 
1.    Requires Peer-to-Peer delay measurement between adjacent links, decreasing load on the Grandmaster. 
2.    Prohibits Transparent Clocks, requiring switches to operate as gPTP bridges (acting similar to Boundary Clocks). 
3.    Incorporates Neighbor Rate Ratio tracking to correct local oscillator drift between synchronization cycles. 
4.    Supports multiple time domains that can share Pdelay measurement results.

Q4: What happens when the gPTP Grandmaster Clock becomes unreachable?
Answer: When the Grandmaster Clock stops transmitting synchronization frames, the gPTP Best Master Clock Algorithm will automatically start negotiating a new Grandmaster according to the configured priorities. IEEE 802.1AS-2020 supports multiple simultaneous time domains that enable immediate failover to a redundant Grandmaster clock.

Q5: What is the clock accuracy that can typically be achieved with IEEE 802.1AS gPTP?
Answer: The achievable clock accuracy depends on several factors, such as the capabilities of the Grandmaster Clock and Time-Aware Switches, the number of hops, oscillator stability and the hardware timestamp implementations. That being said, the synchronization accuracy between two direct neighbors often falls within 10 to 100 nanoseconds.

About Volktek
Volktek Corporation is a Taiwan-based manufacturer specializing in industrial Ethernet networking solutions with more than three decades of experience in supporting industrial automation, building management system, maritime, transportation, energy, and critical infrastructure applications. Volktek focuses on delivering reliable, secure, and easy-to-deploy networking products that operate consistently in harsh industrial environments.
With in-house design, manufacturing, and quality control, Volktek supports customers worldwide through the full product lifecycle—from system design and deployment to long-term operation and technical support. Our portfolio emphasizes robustness, long product availability, and practical security features to help customers build resilient and future-ready OT networks.

 

---

¹. CC-Link Partner Association (CLPA), "Network Technology: CC-Link IE TSN"
https://www.cc-link.org/en/networktechnology/features/cclinkietsn/index.html
². Shihui Duan, Renhe Yan, Jie Chen, Meihui Liu and Qichen Xu. 2023. "Comparison and Testing of Time Synchronization Accuracy between IEEE 1588v2 and IEEE 802.1AS.", 2023 13th International Conference on Communication and Network Security, December 2023, Fuzhou, China. ACM, New York, NY, USA, 9 Pages.
https://doi.org/10.1145/3638782.3638828
³. R. Lavoie, "Transparent versus Boundary Clocks (PTP) in Broadcast Environments." Riedel Communications. April 2020. Application Note & Technical Guide. 
https://www.riedel.net/fileadmin/user_upload/800-downloads/07-Guides/Transparent_versus_Boundary_Clocks.pdf
⁴. D. Arnold, "End-to-End Versus Peer-to-Peer", Meinberg. Sep 2013. Technical Blog.
https://blog.meinbergglobal.com/2013/09/19/end-end-versus-peer-peer/
⁵. K. Stanton, "IEEE 802.1AS Tutorial." Intel Corporation. IEEE 802.1 AVB TG November 2008 Plenary. Technical Presentation. 
https://www.ieee802.org/1/files/public/docs2008/as-kbstanton-8021AS-overview-for-dot11aa-1108.pdf
⁶. Meinberg, "What is gPTP? Understanding IEEE 802.1AS." CN Rood. Technical Blog.
https://cnrood.com/news-articles/what-is-gptp/