TSN is the hottest technology in industrial communications right now. In fact, TSN (Time Sensitive Network) Time Sensitive Network is one of the newest but not unfamiliar technology, so what kind of network is TSN? Why is TSN considered to be the unified standard for future industrial communications, and will TSN replace all types of real-time industrial Ethernet? ......

In the face of these questions, CE China's Lao Shi and B&R's Lao Song started a root-and-branch dialog, hoping to give you a comprehensive and correct understanding of TSN.

1. TSN stands for Time Sensitive Network, what does time sensitive mean? Are other networks not time sensitive? Are there metrics?

A: To be honest, initially when TSN was introduced at Bergerac HQ in 2016, it didn't seem like a technology for industrial automation to me because industrial networks like PROFINET, POWERLINK, EtherCAT, etc. can already achieve relatively high real-time metrics and it seems to be extremely rare to encounter any current 100μS cycle time can not handle the case, industrial networks talk about "Determistic", that is, deterministic network, because the control is based on "isochronous synchronization", the industrial field has long been realized "time sensitive The industrial field has already realized "time-sensitive". Therefore, personally, I felt that this term was developed by IT people who did not understand industry, because the original description of TSN was to give "deterministic" and "real-time" transmission capabilities to Ethernet, because standard Ethernet does not have this capability. This is because standard Ethernet does not have this capability, but real-time Ethernet does.

So what is the point? The key purpose of a TSN is to transmit data over the "same" network, i.e., to transmit cyclical control communication needs and non-cyclical data over the same network, which is the core of what a TSN is all about.

Because the IT and OT convergence process will now encounter a very big problem, that is, cyclical data and non-cyclical data often need to be transmitted over two networks, so you will find that the controller is two ports, a real-time Ethernet, a standard Ethernet, the real-time network in general terms of the language of the machine and the control of the system, while the standard network is used for the transmission of management-level information.

Typically, the QoS (Qualityof Service) metrics of a network include cycle time, latency, jitter, and several other metrics. Generally, hard real-time will be refreshed at the level of a hundred microseconds, jitter is controlled at the level of a few tens of nS, and latency is at the level of microseconds. There is no particular official definition of real-time, because the application scenarios are different.

2. One of the characteristics of TSN is high real-time performance. Doesn't TSN have an advantage over these in terms of real-time performance?

A: Although the key claim of TSN is the "same" network transmission, its scheduling mechanism is designed by R&D staff to allow for high real-time responsiveness, and the TSN for Gigabit Ethernet that has been developed by Bergerac has been tested with a jitter of 50nS (jitter usually describes the worst case scenario). describes the worst-case scenario), and the fastest cycles achievable are in the 10μS class,? Therefore, for existing industrial control, the real-time performance of TSN is fully guaranteed.

TSN is clearly much more capable than today's mainstream real-time Ethernet, with 18x higher overall performance according to Gigabit test data from Bergaplast HQ (which averages performance over increasing node counts, data loads, and so on).

3. How does TSN achieve time sensitivity? Did it make any modifications to the standard Ethernet standard?

A. TSN itself is a set of standards that encompasses three key aspects of clock synchronization, data scheduling, and network configuration. TSN refers only to the standards for the data link layer, which must be clarified, and it can be implemented using either IEEE802.3 Ethernet or IEEE802.3cg standard networks for the physical layer, while the data link layer Bridged networks are used, as well as different strategies for scheduling the data flow, also known as Shaper-Shapers such as CBS Credit Based Shaper, Qbv-Time Aware Shaper TAS, CQF-Cyclic Queuing and Forwarding, and ATS-Asynchronous Transport Shaper. To understand this you have to understand how Ethernet itself is transmitted, and then understand that TSN is mainly designed as a policy scheduling mechanism in the place of Transmit Selection.

4. TSN has become more and more popular in the industrial world in the last couple of years. Where does it come from and is it unique to our industry? Is it unique to our industry? Is it also called TSN in other industries? Who is leading this standard globally?

A: TSN is not from industry, TSN was first in the field of audio and video transmission, and then in the automotive field in 2012, the establishment of the IEEE802.1Q work, and to the industrial has been a 2015 thing, the establishment of the IEEE802.1 TSN Working Group.

In the automotive industry they are generally known as AVB-Audio Video Bridge, which is made up of IEEE802.1Qav, IEEE802.1AS, and IEEE802.1Qat (which has been used as the base standard for IEEE802.1Q).

There are also a large number of time-sensitive networking applications in the aerospace sector, which can be referred to as AS6802, and in the industrial sector, the IEC and IEEE have collaborated to form the IEC60802 working group for interoperability standards development for TSN networks.

The current driving organization for TSN is Avnu, including the IIC, and the OPC UA Foundation have also joined in the effort to drive this technology, and they will be working with the major Shaper vendors*** to drive the development of TSN technology.

5. TSN is going to be the standard Ethernet protocol in the future and is more advanced than the previous standard Ethernet protocols, so does that mean that all Ethernet protocols for residential and commercial industrial use will be TSN in the future?

A. TSN is a VLAN technology, that is, Vitural Local Area Network, which clearly defines it as a LAN, and a virtual LAN, it does not have to necessarily become a commercial and civilian all-purpose standard, the difference between TSN domains and non-TSN domains lies in the VLANID, i.e., into the TSN network will be given by the switch to the VLAN labeling, and then with the help of the The VLAN tag is removed when the switch enters the TSN network and is then transmitted over the network with the help of the TSN mechanism, but when the switch leaves the TSN network, the VLAN tag is removed and it can also be transmitted as a standard Ethernet frame. Therefore, TSN switches will work with normal switches without problems.

Therefore, it is not mandatory for a commercial or residential network to become a TSN network, it depends entirely on the needs of the application itself, especially the economics, so it is not necessary if there are no specific real-time needs, and currently TSN networks seem to be focused more on industrial IoT applications.

6. Why has TSN received so much attention from the industry in the past two years? Can it solve any problems that can't be solved today?

A. The reason for the TSN fire is that the demand for a large number of IoT applications has arisen because, as you have to think about the problem of transmitting periodic and non-periodic data over the same network, the bandwidth requirements are greater than ever before.

You can see in this diagram, which describes several scenarios:

(1) Audio-video synchronization: In fact, if you look at CCTV, you see that the announcer's mouth and voice will be out of sync-which also belongs to the quality problem because the audio and the video are not synchronized or like the speakers in the big theaters, think about it if they can't be synchronized. If they don't synchronize, there will be repetitive sounds, which is also a quality issue with data transfer.

(2). ADAS, as opposed to traditional automotive, will require multiple LIDARs (like three front and three rear), including security systems, and these more sensors will require more bandwidth than ever before.

(3). AR/VR, machine vision: the increased use of these technologies in industrial scenarios will make the bandwidth demand greater.

Therefore, there is a real need for TSN, and as connectivity becomes more widespread, there is a need for greater network capacity to deliver it.

7. With all the advantages of time-sensitive Ethernet, why wasn't time-sensitivity considered when the earlier standard, Ethernet, was introduced? Why do we need to add time sensitivity now? Was it not originally thought of, or was it not possible with the original technology, or was the cost too high for other reasons?

A: When Volkswagen came to China in the 90's to invest in car factories, China's highways were just starting to open up, and there weren't enough parking spaces in many of the older neighborhoods, so who would have thought that cars would be so popular today?

By the same token, you don't really have an incentive to develop a technology when you don't need it, because if no one is using it, the R&D investment is meaningless, so it takes time for any technology to catch on, and as of right now, the TSN hasn't even hit its stride yet because the need for connectivity has only just begun to emerge in the last couple of years, and there are a lot of factories that are still in a state of standalone mode. A large number of factories are still at the stage of stand-alone production. As a result, TSN is now in a rainy day phase, as big data adoption is only beginning locally, not in an explosive phase.

Of course, you are right, to realize the TSN such a network does require a very large investment in technology, because its complexity is more than the existing network, like clock synchronization than the IEEE1588 has the need for reliability, the scheduling mechanism is also more varied and flexible, which are required to chip processing, including Gigabit Ethernet processing chip, transmission cables, switching chips, etc., which are cost, and only in the face of a huge number of applications. These are all costs, and only in the face of huge market opportunities, chip makers will be motivated to invest in research and development. A large number of chip vendors are now investing in this because they see a bright future - one that will make TSNs easier to use and less costly - and the potential scale of TSNs makes them more competitive than previous networks.

8. What are the latest developments in TSN-related product development, test beds, and vendor support to date?

A. Currently, there are several important testbed promoters for TSN, one is the German LNI-which is the testbed of the Industry 4.0 organization, one is the testbed in the IIC, and the other is the testbed organized by Huawei ECC, and Bergerac is actively participating in the construction of the Testbed in all three organizations.

At present, the mainstream automation vendors have released their respective TSN products or test products, like B&R released a TSN product at SPS 2017, while SIEMENS released a Profinet over TSN product at the Hanover Show 2018, and Mitsubishi released a CC-Link IE TSN product in 2019. Huawei, TTTech, CISCO, MOXA, and Hirschmann also released TSN switch products.

9. Although TSN has been hot for the past two years, and almost everyone is optimistic about and supportive of it, why is it that there are not many actual products from the actual products, or they have not been mass-produced and sold, and there are almost no formal industrial applications, and where is the main card? When is it expected that TSN will start to be used in real applications on the ground?

A: It seems that you are more anxious than the user, but the general life cycle of industrial products are longer, like CAN bus, Modbus is still in use, in fact, it does not want to think that you imagine so slow, as opposed to the past bus from the concept, local applications, large areas of the mature application of time, the development of TSN is considered to be relatively fast because of the You have to realize that the TSN industry-oriented working group was only formed at the end of 2015, and the kickoff meeting for Shaper was only held in September 2016, so it's kind of pretty fast that you're able to see so many companies launching products in 2018 already.

TSN will have volume product launches in 2019 including the likes of Bergerac and Mitsubishi who are now known to be launching products, industrial products don't get updated as fast as cell phones, Huawei's P20 Mate Pro I haven't even bought it yet, the P30 came out, and industrial products generally have a lifecycle of 15+ years as a machine tool can last for 20+ years.

So actually, TSN has evolved very quickly.

10. What equipment is needed to make up or support a communications network to implement TSN? Will TSN be more expensive to implement compared to other networks?

A: TSN is a VLAN, so it is a lot of implementation is at the software level, need to have to deal with this kind of clock synchronization and scheduling mechanism of the switch, of course, for the controller, the need for TSN chip support, the specific cost of the chip I have not considered, but the development of the law of the matter is the nature of the ****, the cost must be an The cost must be a decreasing process, because a large number of adoptions can reduce the cost.

11. Now I see that foreign vendors are releasing TSN products one after another, but domestic vendors are not, so how can I develop a TSN master or slave or I/O? Is the cost high?

A: This is not exactly the case, Huawei is already ahead of the curve in this area, and their TSN switch has been developed, but it also seems that it has not yet entered the batch stage, because Huawei will consider more scenarios of the problem, and therefore, there will be some of their own design in terms of the shaper, and I recently wrote about the shaper for TSN ( Shaper), but also asked Huawei's two experts, in addition, MOXA also TSN switch products launched, there are some universities, research institutes, enterprises also launched TSN-related technology test products, has not yet been officially released.

At present, the TSN technology development boards include XILINX, as well as TTTech and Intel*** with the launch of the TSN solution, chip vendors NXP, AD, etc. have also launched the TSN development and test chips and test boards.

12. Nowadays, real-time Ethernet and fieldbus are widely used in the control field, especially in the field of motion control, will TSN replace these protocols, and what is the relationship between TSN and these real-time networks?

A: and need to emphasize again, is that TSN is in the ISO/OSI architecture of the second layer, which means that the current TSN can be in various forms, in fact, the Ethernet itself is also the development process, such as Profinet is Profibus over Ethernet, POWERLINK is the CANopen over Ethernet, then, there will be Profinet over TSN, CC-Link IE over TSN these scenarios, this design is often to maintain the continuity of the application layer software to ensure that the availability of existing equipment investment, and Bekaert will choose OPC UA over TSN, the original POWERLINK CANopen application layer will gradually shift to OPC UA, must maintain a software application continuity, which is also in a longer period of time, the application layer to maintain, while the data link layer is gradually shifted to TSN process. In terms of the performance demonstrated by TSN so far, there is not the slightest problem to cope with the task of motion control, and some vendors declared that they will soon launch motion control products with TSN, with Bergerac launching IOs and controllers in 2019 and planning to launch motion control products based on TSN communication in 2020.

This point must be emphasized, for any industrial application, to maintain the stability of the technology, inheritance is a necessary consideration, therefore, the upgrade of the technology must be as smooth as possible transition, the same is true for the user, which is a guarantee of investment security, so, of course, the TSN will not replace the existing network in the short term, but in the longer term, the TSN will become the basis of the network architecture!

ISO/IEC 17025 and ISO/IEC 17025.

The ISO/OSI model is interesting in that it allows each layer to be designed independently of the other, including the physical layer, where the current IEEE802.3 can continue to be used, and of course, physical layers such as PoE (Power onEthernet), SPE (Single Pair Ethernet-IEEE802.3cg), and so on, can be used. .

13. Now that another communication technology, 5G, has arrived, and 5G is also characterized by low latency and high bandwidth, and 5G is going to have the most applications probably in the industry, will TSN and 5G be a competitive relationship?

A: 5G also has for ULL (UltraLower Latency) scenarios, but, 5G belongs to the wireless network, in the industrial field of application, especially similar to the motion control these should not be used 5G, but, based on the IIoT application of the network deterministic assessment of the field can afford 5G can be used! In addition, TSN will have standards for wireless scenarios.

Therefore, evaluating the future of a technology's adoption depends largely on the scenario - they are certainly not in competition, but rather complementary to each other.

14. There are some people who think that there is a lot of talk about TSN, so what is the latest progress on TSN standards? What are the issues that still need to be resolved to accelerate the progress? Now that the technology is evolving so quickly, will TSN be replaced by a new communication technology midway through the process?