Beckhoff EtherCAT Terminals
Industrial Ethernet Book Issue 96 / 5
Request Further Info   Print this Page   Send to a Friend  

TSN: a definitive kit for demanding automation networks

Based on the key value propositions of Time-Sensitive Networking, and the fact that it can be a major factor in facilitating the transition to future automation networks, TSN is more than a new set of Ethernet networking protocols. It is the future of automation networks and of Industrial Ethernet itself.

UNTIL RECENTLY, TIME-SENSITIVE NETWORKING (TSN), the emerging family of IEEE 802 standards for real-time Ethernet, was merely in the back of the mind of the broad industrial automation community. Curiously, this was despite the fact that TSN was and still is specifically being designed for the automotive and industrial control markets.

The publication of the initial "core" TSN standards by the IEEE 802.1 and IEEE 802.3 committees along with first demonstrators highlighting the practical feasibility of this new technology, however, significantly changed the perception of TSN in the automation industry. This change was further fueled by the OPC Foundation starting to work on specifications that allow combining the device and communication models of OPC UA with the real-time capabilities of TSN.


TSN can be utilized by all industrial protocols.

TSN technology on the move

With this move, TSN suddenly was in the spotlight of major ventures such as Industrie 4.0. For the first time, a vendor-neutral real-time Industrial Ethernet solution was able to offer universal compatibility with conventional Ethernet, without the constraints of being limited to a vendor-specific ecosystem.

By offering the possibility to merge traffic of different priorities and requirements on a single network, while also providing strict latency guarantees for time-sensitive control traffic, TSN is the technology that enables the converged industrial networks of the future. These are networks where time-sensitive control traffic on the local automation network shares the same infrastructure as non-time-critical statistical data, logs and other background information that is on its way to a local server or a wide-area uplink to a cloud infrastructure.

But topics that emerge in this particular fashion and technologies that promise so much tend to be overhyped, up to the point where the hype collapses in its entirety when it becomes clear that a lot has been promised, but the actual delivery is lackluster.

For this reason, let us take one step back and have a good look at TSN. What is the actual value proposition that TSN brings to the table? Setting aside the current hype, what will be the real benefit for operators of mission-critical network infrastructure? And finally, is TSN just a fleeting topic that will not deliver on promises, or is it really going to change the way we look at automation networks?

The value proposition of TSN

The key value proposition of TSN is twofold. First of all, TSN is not a single standard or technology. Instead, it is a set of Ethernet features that is defined as a family of standards. As such, each standard tackles a separate problem space and can be combined as required by a given application scenario.

In addition, TSN is developed and will subsequently be maintained by the same IEEE 802.1 and IEEE 802.3 standardization groups that previously defined the basic Ethernet and Bridging/Switching technologies that are in use today.

This is great news because it means that TSN can be seamlessly integrated into today′s Ethernet-based automation networks. Moreover, this also ensures the evolution of TSN with future Ethernet standards, whichever new technologies they will yield.

The value propositions in detail

Up to this point, choosing real-time capable Ethernet technology inherently means a commitment to one of the many real-time Industrial Ethernet technologies, and mostly closed systems. With some exceptions, these technologies share a basic compatibility with Ethernet, but the layer 2 real-time communication mechanisms are usually incompatible between the various vendor-specific solutions. As a result, selecting a specific network infrastructure hardware such as an Ethernet switch, implies a commitment to a particular real-time technology.

This, in turn, has the side effect that network operators typically find themselves in a lock-in situation concerning the corresponding vendor-specific ecosystem. The alternative is to accept a patchwork installation of many incompatible systems that introduces a lot of unnecessary complexity in network design.

As a consequence, network operators commonly have to cope with high maintenance efforts and a significant total cost of ownership. These circumstances are a major factor that is currently crippling the real-time automation market - a market that otherwise, and especially with the emergence of the Industrial Internet of Things and the Smart Factory, has an excellent growth potential.

This growth potential stems from the fact that many IIoT networks will need to carry a lot of different types of traffic with varying requirements, all at the same time. This starts with the real-time traffic for closed loop control and ends with background data transmissions, such as low-bandwidth log data or low-priority file transfers - with many possible shades of grey in between. Many IIoT applications, which already have a clear value proposition, are directly related to those requirements.

One example is predictive maintenance, where large amounts of usage and statistical data from manufacturing machinery and other devices have to be transmitted to a central data compactor such as a server or gateway for statistical analysis. This data often comes from areas in the network that have some or even strict real-time requirements.

Merging these data transmissions with the time-critical control traffic on the same network, without any negative impact, is a key requirement for this kind of IIoT application. But it will only be realized if the enabling technology, the control network, can provide this capability. Without TSN, network operators would be left between a rock and a hard place: either boxed in on a technology island with a vendor-specific real-time protocol or not able to realize the full IIoT potential with conventional Ethernet technology.

TSN breaks this crippling deadlock by introducing IEEE 802-standardized real-time technology and improves the universally compatible baseline layer 2 communication. Suddenly, standard Ethernet is able to provide a level of determinism that was previously only offered by vendor-specific solutions. By offering the same or better determinism and real-time characteristics, TSN essentially eliminates the necessity to utilize incompatible communication infrastructure hardware and allows merging time-critical and background traffic in a single network.

Hence, one of the major factors that held back real-time Ethernet up to now is gone - TSN will enable network operators to utilize real-time communication in one universally compatible layer 2 network, with all the above described advantages. This is a key value proposition of TSN and the main reason, why its importance to the future of automation networks cannot be overstated.

At the same time, this does not mean, however, that the established Industrial Ethernet ecosystems, such as Profinet or EtherNet/IP, will be superseded by TSN - nothing could be further from the truth. The real value proposition of ecosystems such as Profinet or EtherNet/IP presents itself above the layer 2 network technology in data, device and communication models as well as network and device management functionalities.

With TSN, all Industrial Ethernet ecosystems can utilize the same excellent level of determinism with standardized hardware that is offered by many silicon and device vendors, while maintaining their distinctive characteristics and unique selling points on the application level. Staying within the proven IEEE 802.1 and IEEE 802.3 ecosystem also allows the quick utilization of technical advancements made by these groups, such as bandwidth increases or new physical media. Especially the utilization of higher network speeds in proven and tested silicon chips has been a challenge for many Industrial Ethernet vendors in the past, and the innovation cycles of standard Ethernet could not be met by many vendor-specific solutions.

Building on strengths

Now, Industrial Automation ecosystems can build on their respective strengths, their device and service models, while utilizing the standardized network infrastructure that is offered by TSN. This creates a win-win situation between the Industrial Ethernet vendors and the network operators. The vendors can focus on their core competencies, avoid very costly development efforts for vendor-specific silicon and, at the same time, the real-time market is no longer crippled, which will result in healthier growth rates.

Likewise, the network operators can utilize a commodity layer 2 network infrastructure, which will drive down device costs and the total cost of ownership for the whole network, while no longer being locked into one vendor-specific technology island. This, in turn, will vastly improve the acceptance of real-time Ethernet technology and allows the further development of industrial networks to meet the requirements of the future.

One example of such a win-win situation is the work that is currently underway at the OPC Foundation: In conjunction with the specification of a publish/subscribe-based communication model, the OPC Foundation is currently working on the integration of TSN into their OPC UA ecosystem. Most importantly, this includes interfaces that allow the configuration of the different mechanisms provided by TSN via an OPC UA configuration model. This way, OPC UA can utilize the full range of TSN features and combine it with its own configuration and management solutions. Thus, this work constitutes a blueprint of how TSN can be utilized by higher-layer automation protocols in the future.

All industrial protocols can benefit directly from the TSN mechanisms that are defined at layer 2 of the OSI reference model. More precisely, the automation protocol can utilize Ethernet frames and encapsulate its payload directly or it can utilize IP and TCP/UDP encapsulation, depending on the given protocol architecture. In addition, the managed objects that are defined by the IEEE 802.1 for TSN can be directly mapped to managed objects of the particular automation protocol in order to be able to configure TSN directly with their well-known administration and configuration tools.

Another important aspect of the value proposition of TSN is its inherent modularity. Hence, it is possible to choose the required TSN mechanisms depending on the particular application scenario. This allows decreasing the overall complexity and configuration of the network as well as to reduce the hardware and software requirements of the affected devices. When a network is designed with Ethernet switches that have a common TSN feature baseline, the actual device configuration on the switches will define which features of that common baseline are used and which are not used. Through configuration, different zones in the network with different service levels and varying network complexity can be defined. Not all TSN features that are available have to be configured and used - only those that are necessary to fulfill the application requirements need to be considered.

This also implies that not all devices always need to support all TSN functions, as long as there is a common feature denominator on all devices inside a network. But what exactly is the actual minimum feature set that all products need to support in order to offer basic TSN services?


TSN devices can have a modular design to support different components of TSN.

TSN - the building blocks

The technical basis for TSN with strict latency guarantees in the sub-millisecond range is time synchronization. When network devices are configured to utilize a common clock, all participating devices need to agree on a common understanding of time. Only then can a TSN network can be set up and operated. Thus, when viewing TSN as building blocks, time synchronization is the minimum feature set and a mandatory requirement that always needs to be satisfied on all devices that are participating in TSN networks.

The methods how to achieve synchronization can be chosen freely, depending on the specific application needs. The TSN standardization group has specified IEEE 802.1AS, a profile of the well-known IEEE 1588 Precision Time Protocol (PTP), as a primary time synchronization candidate for TSN. But TSN is not limited to the use of this particular profile. Any IEEE 1588 profile or configuration can be used, as long as all devices in the TSN network are configured to use the same or a compatible synchronization method. This is especially important when TSN is used in environments, where IEEE 1588 with a specific profile is mandatory, for example in the Power Automation market.

Based on the common understanding of time in all TSN devices, i.e. switches and end points, traffic scheduling and shaping mechanisms can be implemented. Traffic scheduling in TSN is done through the Time-Aware Scheduler that is defined in IEEE 802.1Qbv. Based on the common time on all devices, individual time-slices for different traffic types (TDMA - Time Division Multiple Access) can be configured. This ensures that, for example, background traffic does not negatively interfere with time-sensitive control traffic. Thus, the Time-Aware Scheduler allows an ability to specify latency guarantees with hard bounds.

In addition to the scheduling of hard real-time traffic, TSN also offers methods to shape traffic in order to achieve soft real-time guarantees or to specifically target certain traffic patterns. TSN offers a variety of traffic shaping mechanisms, each aimed at different use cases. All of these mechanisms can be combined in a device to serve different traffic patterns at the same time. The IEEE P802.1Qch cyclic traffic shaper, for example, can be used, where repeating traffic patterns are used, but there is some flexibility in the exact delivery precision of each individual frame. Likewise, the credit-based shaper from IEEE 802.1Qav can be employed where soft real-time traffic such as a CCTV video stream is competing for bandwidth with background traffic, such as file transfers. In this case, the credit-based shaper will provide a fair distribution of bandwidth to all involved communication streams.

The modular design of TSN also allows the Time-Aware Scheduler to be combined with one or several of the traffic shaping methods, in order to enable the coexistence of different traffic types (Hard Real-Time, Soft Real-Time, and Background Traffic) on the same network depending on which methods are supported by the network devices.

Especially mission-critical networks may also require fault-tolerance mechanisms in combination with TSN. This is true, for example, in case of power automation networks, where the recovery of a lost frame at a higher-layer protocol would violate timing requirements and where the loss of a sampled value cannot be tolerated. For the particular use case of sampled values that are transmitted cyclically, the combination of IEC 62439-3 HSR or IEEE P802.1CB seamless redundancy and the IEEE P802.1Qch traffic shaper presents itself as the logical conclusion. For a hard real-time motion control use case, for example, the P802.1Qch shaper may be replaced by the IEEE 802.1Qbv Time-Aware Scheduler. The redundancy mechanisms then still provide the same level of fault-tolerance, but the 802.1Qbv Scheduler offers tighter timing for the control traffic, as required by the application.

The overall idea of TSN is a scalable solution that will fit many use cases in automation networks and, in doing so, it will not always be necessary to utilize the full feature set of TSN. Therefore, the market will see a lot of devices that are aimed at very specific use cases and will only support a specific set of TSN functions, as well as more complex devices that will provide support the full range of all TSN standards functions in order to cater to a broader range of application scenarios.

A device vendor such as a manufacturer of Industrial Ethernet switches may, for example, choose to only implement a select TSN feature-set, depending on market that the device is targeting, thus allowing the device vendor to offer a sweet-spot concerning the technical complexity of the device and the target price/performance range. In the world of conventional Ethernet switching, such feature differentiation in different devices at different price points is nothing novel - the ability to mix and match devices due to the universal compatibility has been a key success factor for Ethernet in the last decades.

With TSN, this flexibility is now expanded to the area of real-time networking, which, up to this day, has been an "all-or-nothing" approach. In such a flexible real-time Ethernet ecosystem, protocol conformance statements and certification authorities, such as AVnu, then will ensure that TSN capabilities provided in a device can clearly be identified by a customer and that interoperability is ensured.

The device vendors will clearly announce the different TSN functions in the device specifications - this will, in essence, become part of a device datasheet, for example, for an Ethernet switch. Through the protocol conformance statement and the AVnu certifications, a customer can always clearly see the TSN capabilities of each device.

The future powered by TSN

Despite the continuing work on the TSN standards, core specifications have already been completed and are readily available for their integration into upcoming products. Most notably, these specifications include the IEEE 802.1Qbv Time-Aware Scheduler. As such, they already can accommodate very demanding application requirements and, therefore, can provide substantial benefits in today′s industrial automation networks. At the same time, the IEEE 802.1 standardization process ensures that all future TSN specifications are backward compatible and will work nicely with the current TSN feature-set. Hence, even when implementing and utilizing TSN features in automation networks today, these feature will continue to operate and be useful as the standardization of TSN progresses.

Based on the key value propositions of TSN and the fact that it can be a major factor in facilitating the transition to future automation networks, TSN is more than a new set of Ethernet networking protocols. It is the future of automation networks and of Industrial Ethernet itself. Its modularity, flexibility, and scalability will ensure that device vendors will be able to provide products that fit market-specific price-performance sweet spots. Network operators do not need to be afraid to bet on "the wrong horse", as TSN is vendor-neutral and the IEEE specification ensures investment security by ensuring compatibility with future devices.

Dr. Oliver Kleineberg, Axel Schneider and Dr. René Hummen, Hirschmann Automation and Control.


Source: Industrial Ethernet Book Issue 96 / 5
Request Further Info    Print this Page    Send to a Friend  

Back

Sponsors:
sps connect: Automation goes digital
Accelerate your HART data at the speed of Ethernet
Industry of Things World

Get Social with us:



© 2010-2020 Published by IEB Media GbR · Last Update: 20.10.2020 · 14 User online · Privacy Policy · Contact Us