Mr. Beckmann, Mr. Rettig, Beckhoff has been cooperating with Huawei since the end of 2017 - in what area and why?
Guido Beckmann: The cooperation is specifically about two projects in which we are evaluating new communication technologies together with Huawei. One is about the topic of 5G. Here we are working together with Huawei's so-called X-Lab. In the second project, which we have launched with Huawei's Network Technology Lab, we are looking at switched and routed networks. With 'X-Ethernet' and 'Deterministic IP', Huawei has developed its own technologies in this area.

Thomas Rettig: It is simply interesting for us to investigate whether and how technologies from other industries, such as the information and communication technology sector, can be used for industrial communication. Incidentally, in both cases Huawei approached us with the question of whether we would be interested in a corresponding cooperation.

Thomas Rettig, Senior Management Control System and Communication Architecture at Beckhoff Automation.

© Beckhoff

Rettig: In gigabit communication, for example, ten 100 Mbit channels can be defined. As these are always transmitted one after the other on a rolling basis as individual fixed bit blocks, communication is deterministic, jitter-free and easy to configure. And: I don't have to do anything external to synchronize X-Ethernet switches - communication always takes place at edge start to each other. For us, the appeal of the technology lies in the fact that we can use it to integrate standard Ethercat devices directly via a heterogeneous X-Ethernet network in which other standard Ethernet communication takes place in parallel. From the Ethercat master's point of view, this only results in a transmission delay similar to that of a cable. Huawei wants to reduce this to one microsecond per switch cycle. At the recent Hannover Messe - where we already had a corresponding demo running - we were already at 3.5 microseconds with the first prototypes. This enabled us to demonstrate a 50 microsecond cycle from an Ethercat master to a remote Ethercat segment without any further adjustments via a 2-stage X-Ethernet network.

Ultimately, however, X-Ethernet is a proprietary technology from Huawei, isn't it?
Beckmann: Correct. But Huawei's efforts are definitely aimed at standardizing the technology and publishing an international specification. What's more, the whole thing is based on the IEEE 802.3 standard, even if the switching property is special.

For which specific applications would a technology such as X-Ethernet be predestined?
Beckmann: Basically, there are similar applications to Ethernet TSN, for example if you want to do image processing in parallel in the machine - and not just with one camera, but also with several. Another example would be the use of our XTS linear transport system - also a very 'data-hungry' application, where it could definitely be advantageous to use X-Ethernet to distribute the data.

What specific advantages would X-Ethernet have over the currently much-vaunted Ethernet TSN?
Rettig: As X-Ethernet - as already mentioned - involves the transmission of fixed bit blocks instead of telegrams, you no longer have to deal with delay times of different lengths. This simplifies the configuration enormously. I also no longer need to adapt to 1588 synchronization, which has to work with Ethernet TSN between the switches and the end devices. The same applies to MAC address adaptation, which is necessary with TSN in order to define the streams for virtually every end device behind the TSN network. With X-Ethernet, I only have to define the channel once and then I don't care what kind of Ethernet communication runs over it - whether it's standard Ethernet, UDP/IP, Profinet or Ethercat.

And what are the disadvantages of X-Ethernet?
Beckmann: One is that you reserve a fixed bandwidth. This means that if I define a 100 Mbit channel, it is definitely missing from the 1 Gbit quota and cannot be dynamically allocated elsewhere. However, I can flexibly adapt my Ethercat configuration in the master. In other words, if the Ethercat network is extended under the X-Ethernet network, there is no need to worry about the configuration of the X-Ethernet switches, because X-Ethernet is just like a 100 Mbit cable.

The whole world finally wants 'an' Ethernet-based communication standard for the age of Industry 4.0 - the aforementioned Ethernet TSN + OPC UA is highly regarded here. Does it still make sense to deal with a proprietary solution such as X-Ethernet - if you disregard the advantages mentioned?

Beckmann: First of all, let me be clear: Our activities with regard to X-Ethernet are about technology evaluation - nothing more! No products have been discussed here at all. However, if we and our customers discover that new technologies have clear advantages over established solutions, then these are always of interest to us and we take a close look at them accordingly.

Although X-Ethernet is roughly on the same level as TSN, we do not see it as direct competition, at least not at the moment. This does not mean that we will be less involved with TSN. However, it is also a fact that our biggest headache with TSN is still the question of what will happen in terms of configuration or when there will be something sensible in this regard and who will ultimately agree on what.

Basically, we believe that OPC UA TSN technology makes sense for what it was originally intended for - i.e. for communication from controller to controller or for vertical upward communication. We think it is less useful for I/O communication at bus coupler level - partly due to the complex configuration of TSN networks.

However, the Shaper Group in particular is proclaiming Ethernet TSN + OPC UA all the way down to I/O communication, thereby proclaiming the end of previous Ethernet-based fieldbuses!

Rettig: Perhaps. However, you should also bear in mind that the established Ethernet-based fieldbuses set the bar very high in terms of low costs, diagnostics and machine set-up. What happens if something doesn't work? Then I need to be able to find the fault in my machine very quickly. Or what happens if I actually have to expand my network? How will the entire communication be set up again? - You don't want to become worse with a new technology!

I believe that it will be five years at the earliest before we can really assess how things will develop in this respect - in other words, whether Ethernet TSN could reach a comparable level at field level to today's Ethernet-based solutions.

In addition to X-Ethernet, you also showed 'Deterministic IP' as part of the trade fair demo in Hanover - what is this about?

Beckmann: Huawei's deterministic IP approach is about creating real-time properties on a routed network - i.e. on layer 3. To understand: In a factory, there is usually a need for halls, cells or subnets to communicate with each other. If this is to take place in real time via a cable, a technology such as TSN no longer helps, as this is a layer 2 switch technology. Instead, I then have to go one step further - i.e. to a routed level.

Rettig: With this in mind, Huawei has developed Deterministic IP, another hardware-supported technology that can be used to request or reserve cyclical bandwidth in the router. Routing from one port to another port takes less than 50 microseconds. This means fast routing at IP address level with deterministic properties. This can be used, for example, to define a data stream through the factory in order to calculate a digital twin in a remote server room, which in turn affects the process in real time. Another application example would be the topic of PLC cloudification - in other words, how we could improve control from a local cloud in the future.

Let's move on to your second project with Huawei - wireless communication via 5G.
Beckmann: Here, too, it's initially about technology evaluation and also about clearly demonstrating our requirements for industrial-grade wireless communication to an important player in the ICT market that manufactures both chips and end devices. In turn, we want to understand what the ICT sector is capable of achieving in this respect.

Why technology evaluation - aren't the key points regarding 5G already clearly defined?
Rettig: As far as Release 15 of the standard is concerned, which is now being frozen and which the industry will be using from next year, this may be the case. What we are talking about with Huawei is Release 16 of the standard, which will only be defined at the end of 2019 and then implemented!

The new and exciting thing about Release 16 for us is that there will be three different operating modes for 5G in future, which can run in parallel: In addition to the 'Enhanced Mobile Broadband' mode 1) - i.e. 'normal' data communication with an extended bandwidth of 20 Gbit/s as we already have in Release 15 - the new modes 'Ultra-Reliable Low-Latency Communication' 2) and 'Massive Machine-Type Communication' 3) will be added. These two in particular will make 5G really interesting for the industry!

What exactly is behind these two modes?
Rettig : uRLLC is about transmission that is as deterministic as possible. The plan is to achieve a latency time of one millisecond from hop to hop - i.e. from the base station to the end device - as well as a very high availability of 99.999% from a radio technology perspective. Admittedly, the latter is very low from our point of view if you take wired communication as a basis. However, this is an extremely good value for wireless communication.

The mMTC mode, on the other hand, should make it possible to integrate up to one million 'connected devices' - in our case, for example, wireless sensors on a machine - per square kilometer into a 5G network.

Although the data rate is then reduced, the energy consumption is correspondingly low, which in turn allows for enormously long battery life for the sensors. For a simple sensor, for example, which only needs to transmit a temperature once or twice a day, it is therefore sufficient to only support mMTC. This allows the sensor to be built smaller and therefore more cost-effectively. In a nutshell: mMTC is what makes the Industrial Internet of Things possible in the first place.

Beckmann: What you have to understand in this context: You don't get everything at once with 5G! If I have a 20 Gbit data rate, I can no longer transmit it in 1 ms. This is precisely why the different modes have now been defined, called 'network slicing'. It is important that the infrastructure to be set up - i.e. my base station and my core network - supports everything I want to use. In future, there will be devices or chipsets that support all three modes. For a PLC, for example, it may make sense for it to support a high bandwidth in addition to short-cycle communication in accordance with uRLLC if a vision stream is also to be transmitted.

In any case, we are pleased to see that the 5G community organized in the so-called 3GPP committee is addressing precisely these issues. In other words, it no longer only has smartphones in mind, but is now also approaching new industries.

What other topics still need to be addressed in order to further optimize 5G for use?
Beckmann: One example is the topic of 'private networks', which we are now also addressing via the ZVEI or the 5G ACIA group set up in April in the direction of the Federal Network Agency. What is this about? Many of our customers don't want to rely on the antenna mast in front of their house to implement their machine-to-machine communication. Instead, they want to be able to install their own infrastructure - a 'private' network, so to speak.

In this context, questions such as: Which frequencies or which spectrum can be used for this? Can I use these not only in Germany but also throughout Europe? And what if I sell my machine to the USA - which frequencies do I need then? Although these questions are more of a political nature, they are no less important.

Rettig: Last but not least, the reliability of the connection and susceptibility to interference will remain critical issues. After all, 5G is a shared medium, which means that if it works today, I don't know whether it will still work tomorrow.

As far as susceptibility to interference is concerned, this can be reduced by increasing the frequencies. If I don't transmit at 3 or 5 GHz but at 25 GHz or more, I can no longer get out of my building. As a result, I no longer disturb my neighbors and can no longer be disturbed from outside. However, this also limits the range, which the telecom providers don't like because they then have to install their masts at shorter distances.

However, the latter would be particularly interesting for industry, because: Ultimately, I can also use it to protect my factory - keyword security! The disadvantage, however, is that the radio illumination of my plant is more difficult. But I believe that a lot can be achieved in this respect too.

Beckmann: The decisive factor in everything we have discussed so far is that we as the automation industry must now succeed in clearly communicating our requirements for wireless communication to the ICT sector so that these are taken into account in Release 16 of 5G. If we don't succeed, we can move on to 6G again!