Time-Sensitive networking

IHS Markit forecasts that the IoT market will grow from 15.4 billion devices in 2015 to more than 30.7 billion devices in 2020, and 75.4 billion by 2025. Bain & Company projects revenues from the IoT alone could reach $470 billion for vendors selling hardware, software and comprehensive solutions by 2020.

With the IoT becoming more ubiquitous, there’s been a big push in the industrial space over the past few years to develop vendor-neutral, universally applicable standards and definitions for the transmission of time-sensitive data over Ethernet networks.

As the industrial Internet of Things (IIoT) – the IoT for the industrial space – continues to expand, its need for reliable and efficient control networks, free of bandwidth and latency issues, becomes even more important to support the large volumes of data – with and without real-time requirements – on which the phenomenon depends.

Working groups fuel success

The Institute of Electrical and Electronics Engineers (IEEE), within the IEEE 802.1 working group that is responsible for LAN/MAN standards, instituted a task group that is responsible for time sensitive networking (TSN). The goal of this task group is to provide specifications that will allow time-synchronized, low-latency streaming services through vendor-neutral Ethernet networks.

Industry partnerships, such as the Industrial Internet Consortium’s effort for TSN technology or the Avnu Alliance, have also been established to foster awareness and compliance with this new framework.

Ethernet networks today

TSN provides low latency guarantees, especially for use cases with synchronized axles and drives, power generation transmission and distribution networks and applications in the transportation industry. The cycle times for time sensitive data in these spaces are often well below a millisecond.

To achieve both low latency and jitter, vendor-specific solutions like EtherCAT, PROFINET IRT and SERCOS III are often used – technologies that rely on conventional Ethernet but utilise additional proprietary mechanisms to enhance standard Ethernet. These proprietary mechanisms are incompatible with each other.

The real-time Ethernet solution market is therefore very fragmented, with a lack of compatibility and little room for market growth. TSN is intended to open up the real-time Ethernet market and establish a standardised, universal physical and data-link layer, which will lead to major cost savings and better investment security for industrial engineers.

Communication with TSN

TSN is comprised of a variety of mechanisms and interdependencies that work together to ensure smooth transmission of real-time data. These features add a level of determinism to Ethernet-based data communication that is able to meet even the highest demands of modern control networks:

• Prioritisation based on timing with the IEEE 802.1Qbv Time-Aware Scheduler (TAS): To avoid bottlenecks in data transmission and the queueing effect – where Ethernet frames with low priority already in transmission could delay frames of the highest priority at every switch along the transmission path – the TAS prioritizes data transmission, sorts them into appropriate time slots and guarantees their forwarding and delivery at a specific point in time. The TAS divides time into segments of equal length, or cycles. This allows for dedicated cycles for the transmission of data packets with real-time requirements and mitigates the possibility of network traffic and latency issues.
• The necessity of guard bands and the interruption of Ethernet frames: As predictability of best effort traffic patterns is low – it’s not always possible to foresee when a specific data packet will be processed – guard bands are used to avoid situations of delayed real-time data processing. Guard bands interfere with transmission of data packets to prevent transmission of best effort Ethernet frames that would intrude into subsequent low-latency time slots. This supports the simultaneous transmission of high-priority and low-priority data on the same network and avoids violation of end-to-end latencies, but can also result in dead-time where the network cannot be utilized. The IEEE 802.1 working group has developed a method to temporarily interrupt the transmission of a long low-priority frame, then transmit the high-priority frame and resume the low-priority frame at a later time. This, combined with the TAS, results in maximum available bandwidth for low-priority background transmissions while still providing timing guarantees for high-priority traffic.
• Synchronous clocks as a prerequisite: Close coordination between all network devices is required to ensure that frames match the correct time slots in each switch as they travel from source to destination. This requires synchronized clocks in all switches and end devices. Time synchronisation in TSN can be done through any means necessary, but the higher the precision, the sharper the TAS can be configured. The Precision Time Protocol (IEEE 1588) is a well-known tool in automation networks to provide time information and can be used for TSN networks, as well.

Shaping & sharing network traffic

Traffic shapers complement the TAS to enable a more granular traffic control. As with the TAS, their use is always connected to the exclusive assignment of one of eight class of service priorities. There are three general types of traffic shapers, which can be combined with the TAS in a network to provide different latency and jitter guarantees, as well as cater to different traffic patterns. The shapers include:

• The Credit-Based Shaper: ensures provision of the maximum required bandwidth for an audio/video transmission, without a noticeable interruption of the best effort data traffic that is simultaneously transmitted.
• The Cyclic Queuing and Forwarding Shaper: Provides timing guarantees based on the number of network “hops” (e.g. switches) along a transmission path, but allows for some jitter, as this shaper collects the data frames with reserved bandwidth received within a cycle and transmit as “prioritised” at the start of the next cycle.
• The Asynchronous Traffic Shaper: Does not require a time synchronisation mechanism and is best suited for the prioritised transmission of event-driven data packets or frames that are needed for the time synchronisation itself.

The future

TSN will play an increasingly important role in industrial applications as it is applicable wherever real-time information is required.

Flexible, intelligent and dynamic production and manufacturing facilities are becoming the norm and as these connected environments continue to grow. Calculable, guaranteed end-to-end latencies, highly limited latency fluctuations (or jitter), and extremely low packet loss will be instrumental to their success.

While the standardisation process is not yet complete, TSN mechanisms are here to stay and engineers across all sectors will need to familiarize themselves with the new framework and make sure they are implementing it appropriately to maximise its benefits.