Professional Documents
Culture Documents
TSN - Time Sensitive Networking: White Paper
TSN - Time Sensitive Networking: White Paper
Figure 1: Transformation from the automation pyramid to the automation pillar in future automation networks
Figure 2: Time division multiplexing permits the reservation of time slots within a cycle in order to enable the timely transmission of periodic real-time data
developing into an automation pillar. in-vehicle networks. Even today, it is documents. In the following sections
This is a continuous long-term foreseeable that some of TSN’s target of the white paper, we will provide an
transformation process. In the context markets will differ significantly from overview of the most important TSN
of the automation pillar, the full value of each other concerning their requirements. mechanisms and how these
TSN becomes apparent: It enables the For example, a deterministic and, at mechanisms interact with each other.
demand-oriented scaling of real-time the same time, fault-tolerant data
mechanisms. It also allows for planning transmission may be absolutely Prioritization based on timing with the
of the network topology and the necessary in one application, but Time-Aware Shaper
bandwidth and, at the same time, in another case, the fault tolerance Until now, Class of Service (CoS)
retains investment security in the requirements may only be of secondary mechanisms such as the strict priorities
networks due to the backwards importance. To cater to these different according to IEEE 802.1Q have not been
compatibility to the proven, existing requirements, TSN is designed as a able to guarantee the forwarding of
Ethernet technology. modular system in which the exact time-critical data traffic at a fixed point
characteristics of deterministic data in time: A low-priority Ethernet frame
transmission – and the associated already in the sending queue of a switch
TSN – Mechanisms and hardware and software requirements – can delay the transmission of other
interdependencies can be adapted to the respective data frames, even when they are tagged
requirements. Following this logic, TSN with the highest priority (i.e. priority 7).
TSN extends Ethernet-based data is not specified in one single standards This is possible in each Ethernet switch
communication with novel mechanisms document, but rather through a family along the transmission path. As one
to provide determinism. This is done on of international standards that have of the central components of TSN, the
such a level of performance that TSN been in development since 2012 at the “Enhancements for Scheduled Traffic”
can satisfy even the most demanding IEEE 802 and its TSN Task Group [1]. introduce the possibility of prioritizing
requirements of modern control By now – with only a few exceptions – the data transmission of conventional
networks, e.g. in industrial automation, all mechanisms of the TSN family are Ethernet frames on a temporal level.
automotive manufacturing and available as finished standards For the first time with Ethernet, this
Figure 3: The Time-Aware Shaper (TAS) implements time-based prioritization via the newly-introduced Time-Aware Gates that are located between the CoS
queues and the selection function for the packets to be sent
Figure 4: The guard band in TSN prevents best effort frames from extending into a time slot that is reserved for real-time data, but it decreases the available bandwidth
Figure 5: With the method of Ethernet frame pre-emption, the guard band size can be reduced from the maximum size of an Ethernet frame to the size of a partial packet
best-effort Ethernet frame and can stop critical traffic will not be disturbed, as transmission of the frame can then be
it from violating the subsequent time the best-effort packet will be buffered interrupted and suspended at a multiple
slot, as shown in Figure 4. At the same until the next best-effort time slot of the 64 byte limit before completion of
time, however, a guard band inevitably starts, but it would experience a delay the current time slot. The transmission
leads to unwanted downtime in the in transmission and the bandwidth is then resumed and completed at the
network, as nothing can be transmitted, that can effectively be used would be beginning of the next best-effort phase.
and thus to a waste of bandwidth. restricted. Frame pre-emption allows the guard
band to be limited to the maximum size
As an alternative to the explicit guard Interruption of Ethernet frames of the Ethernet fragments, with 64 byte
band, the Transmission Selection In order to maximize the bandwidth as the minimum. With Fast Ethernet
mechanism (see Figure 3) can utilization for best-effort Ethernet networks, for example, this results in
additionally take the packet length of frames, the IEEE 802 working group a reduction in the “dead time” of up to
an Ethernet frame into account. This also developed a method for Ethernet 0.12 ms per transition from a best-effort
works as long as the size of the packet frame pre-emption (IEEE 802.1Qbu- phase to a phase with time-critical data
is known at the point in time when 2016 [3], IEEE 802.3br-2016 [4]), which traffic. This subsequently results in
the transmission decision is taken. was completed in June 2016. Using significantly less wasted bandwidth.
The transmission decision therefore this method, conventional Ethernet
depends on whether the next packet frames can be divided into small packet As a component of the TSN toolkit,
in the queue still “fits” and can be fragments of at least 64 bytes and Ethernet frame pre-emption can, of
completely transmitted – before the each fragment can be transmitted course, also be used separately from
transition to the next time slot. Still, the separately. As shown in Figure 5, this the TAS mechanism. In this case, for
situation may arise where a packet can enables the transmission of an Ethernet example, the CoS priority assigned to
no longer be transmitted if it is too large frame to be started within a best-effort the time-critical control traffic can be
or if the remaining time in the time slot phase, despite insufficient remaining configured as “express” – effectively
is too short. In this case, the time- time to transmit the whole frame. The as preferred traffic that is not to
Figure 6: Precise time synchronization is a prerequisite for the TSN Time-Aware Shaper
Figure 7: With the Credit-Based Shaper, data streams with reserved bandwidths are handled with higher priority than best effort traffic, as long as positive
transmission credit is available
threshold are exceeded. Accordingly, is then prioritized by the respective working group for the predecessor
it is not always possible to predict the traffic shaper in a way that ensures that technology of TSN, Audio/Video
exact time of communication in these certain latency bounds for the time- Bridging (AVB). As the name suggests,
scenarios. During the data transmission critical data can be achieved. However, this technology primarily serves audio/
however, clearly defined latency limits a compromise for the flexibility in video and comparable applications.
should typically be adhered to in order prioritization gained in this way is a The aim of the Credit-Based Shaper
to ensure a control decision based on conceptually lower accuracy in the is to secure the maximum bandwidth
current information. But the TAS achievable latency and jitter guarantees required for an audio/video transmission
mechanism can only provide low in comparison to the TAS mechanism. in a defined time interval without
latency and jitter guarantees if the end noticeably interrupting the best-effort
device meets predictable transmission As part of the standardization activities data traffic that shares the same network.
times. Therefore, this mechanism is at the IEEE, three different traffic shapers To achieve this, the Credit-Based Shaper
only suitable to a limited extent for such are currently being considered for use allocates transmission credit to data
traffic patterns to ensure prioritization with TSN: streams with reserved bandwidth.
of the process data. Initially this credit is 0.
• Credit-Based Shaper
(CBS; IEEE 802.1Qav-2009 [10])
In addition to the TAS mechanism, TSN As long as the transmission credit is
therefore offers further prioritization • Cyclic Queuing and Forwarding in the positive range (≥0), data packets
mechanisms, so-called traffic shapers. (CQF; IEEE 802.1Qch-2017 [11]) with reserved bandwidth are transmitted
These afford the reservation of worst- • Asynchronous Traffic Shaping preferentially (see, for example, the
case bandwidth requirements for time- (ATS; IEEE P802.1Qcr [12]) transmission of the first AVB frame
critical control data based on defined marked in blue in Figure 7 on the left).
observation intervals (e.g. 250 μs). The Credit-Based Shaper was already Through any such preferential
The forwarding of such reserved traffic developed in 2009 by the IEEE 802.1 transmission, the transmission credit
Figure 8: Using Cyclic Queuing and Forwarding, data streams with reserved bandwidth are transmitted intermittently by one hop in the direction of the receiver
with each cycle
time window. In other words: while malicious intent, the failure of a network as shown in Figure 9, at the first
TAS acts on the output side for time- element or a communication line can network element (e.g. a switch) on the
controlled sending, Ingress Policing also lead to interference in deterministic transmission path. At the destination,
offers similar functionality on the input data transmission. In order to avoid the first replicated data packet is
side upon reception, i.e. before the the packet loss associated with such a forwarded to the application layer.
packet is processed by the switch or disruption, a fault-tolerance procedure Subsequently received packet
end device. Due to implementation was developed at the IEEE with duplicates, on the other hand, are
complexities associated with time- IEEE 802.1CB-2017 [16], which works detected via a new redundancy field
based policing, it is not yet completely similarly to the already established, in the Ethernet header, the so-called
clear if and in which scenarios this full seamless redundancy mechanisms R-TAG, and rejected. This ensures that
version of the Ingress Filtering and High Availability Seamless Redundancy redundant data transmission according
Policing mechanism will be used. (HSR) and the Parallel Redundancy to IEEE 802.1CB is transparent for
Bandwidth monitoring, on the other Protocol (PRP) according to IEC 62439-3, higher-level layers in the network stack
hand, is commonly used today in and was kept compatible with them. and does not have to be considered
switches for automation networks. IEEE 802.1CB is therefore a static separately there.
Last but not least, TSN can also be fault-tolerance procedure in which the
used with existing Layer 2 security redundant transmission paths are The redundancy mechanisms
mechanisms such as MACsec permanently active. In the event of an developed as part of IEEE 802.1CB offer
(IEEE 802.1AE [15]). In this way, the error, the switchover times from one a significant advantage over HSR and
authenticity of the sender can be path to the other path can be avoided. PRP: they can be used on any topology.
checked and only correctly verified IEEE 802.1CB is not limited to the
Ethernet frames can be forwarded. In In order to achieve seamless otherwise mandatory ring topologies
this way, it is possible to handle a large redundancy according to IEEE 802.1CB, or topologies with completely
number of attacks and scenarios with the Ethernet frames are replicated at the independent parallel networks. In
incorrectly configured network beginning of a redundant transmission addition, IEEE 802.1CB supports
participants. path and then transmitted through the redundant transmissions over more
network via the different redundant than two path in order to further
Better safe than sorry: fault tolerance paths. Usually, the frame replication reduce the probability of packet losses
in communication paths takes place either directly at the sending and, thus, eliminates this limitation
In addition to incorrectly configured end device or, if the end device does not of existing redundancy mechanisms.
network subscribers or those with have a redundant network connection However, it must be ensured that all
redundant paths can meet the required
latency guarantees. The management
of requirements and comfortable
configuration of TSN networks is
therefore an important part of a
functioning TSN ecosystem that
consists of both network devices and
network management.
Figure 10: In the centralized TSN configuration approach, the end devices communicate directly with a central configuration instance
Figure 11: The decentralized and hybrid approaches offer a configuration interface to the end devices that is agnostic to the configuration model
Figure 12: Hirschmann Industrial HiVision enables convenient manual engineering and monitoring of TSN networks
This information includes characteristics specific protocols such as OPC UA can Allocation Protocol (RAP, IEEE P802.1Qdd
such as the cycle time and bandwidth be used for the information exchange [19]) to specify a new distributed
requirements. End devices that are between Talker or Listener and CUC. protocol at the IEEE that considers the
interested in a data stream that is The switches are configured using new TSN mechanisms.
announced, so-called listeners, can then existing management protocols such
use this information to register for a as SNMP (Simple Network Management Finally, the hybrid approach combines
specific data stream. After registration, Protocol) or through YANG models the centralized and decentralized
they will receive the data packets with protocols such as NETCONF. approaches. Like with the distributed
belonging to the stream. To interface CNC to the CUC, IEEE approach, the end devices announce
802.1Qcc already defines some basic their requirements via the distributed
The three planned configuration models structures in the form of a YANG model. protocol. However, as shown in Figure
differ in how the device and stream Within the scope of a new standardization 11, the actual TSN configuration takes
requirements are communicated and project (IEEE P802.1Qdj), these basic place centrally. One advantage of this
processed in the network. With the structures will now be extended to a approach is that end devices only
centralized approach, talkers and complete interface definition. This will, have to support a single configuration
listeners communicate via a direct for example, enable a uniform interface protocol, but the network can be either
end-to-end connection with a (logically) based on RESTCONF in the future. managed centrally or configured in a
centralized end device configuration decentralized fashion, depending on the
instance, the Centralized User In contrast to the centralized approach, network operator's preference. SRP in
Configuration (CUC), as shown in the distributed approach propagates IEEE 802.1Qcc has already been extended
Figure 10. the end device requirements in the for this purpose.
network (see Figure 11) and determines
The CUC creates the combined a common configuration of the TSN Even though all three configuration
requirements for the data streams from mechanisms to be used, based on the mechanisms described here are still
the Talker and Listener information locally available information in each under development, it is already
and passes them on to the Centralized device. The Stream Reservation possible today to configure the presented
Network Configuration (CNC), which is Protocol (SRP) that was originally TSN mechanisms. Configuration
also the logical network configuration developed for the TSN predecessor interfaces are available via standardized
instance. The CNC calculates the time technology AVB operates based on protocols such as SNMP. This enables,
slot for a new data stream, for example, this principle. However, it does consider for example, manual engineering of the
based on the information available the TSN configuration requirements, cycle times and time slots of the TAS
regarding the network topology and the as TSN was developed later. Therefore, mechanism using network management
resource reservations already allocated, IEEE 802.1 is currently developing the tools such as Hirschmann Industrial
and configures the network participants Link-local Registration Protocol (LRP, HiVision (see Figure 12).
(e.g. switches) accordingly. Application- IEEE P802.1CS [18]) and the Resource
References
1. https://1.ieee802.org/tsn/
2. https://ieeexplore.ieee.org/document/7440741
3. http://ieeexplore.ieee.org/servlet/opac?punumber=7553413
4. http://ieeexplore.ieee.org/servlet/opac?punumber=7900319
5. http://ieeexplore.ieee.org/servlet/opac?punumber=7433913
6. http://ieeexplore.ieee.org/servlet/opac?punumber=4579757
7. http://ieeexplore.ieee.org/servlet/opac?punumber=5741896
8. Hirschmann IEEE 1588 White Paper
9. https://1.ieee802.org/tsn/802-1as-rev/
10. http://ieeexplore.ieee.org/servlet/opac?punumber=5375702
11. http://ieeexplore.ieee.org/servlet/opac?punumber=7961301
12. https://1.ieee802.org/tsn/802-1qcr/
13. http://www.ieee802.org/1/files/public/docs2010/ba-boiger-bridge-latency-calculations.pdf
14. http://ieeexplore.ieee.org/servlet/opac?punumber=8064219
15. http://ieeexplore.ieee.org/servlet/opac?punumber=11085
16. https://ieeexplore.ieee.org/document/8091139
17. https://ieeexplore.ieee.org/document/8514112
18. https://1.ieee802.org/tsn/802-1cs/
19. https://1.ieee802.org/tsn/802-1qdd/
About Belden
Belden Inc., a global leader in high quality, end-to-end signal transmission solutions, delivers a comprehensive product
portfolio designed to meet the mission-critical network infrastructure needs of industrial, enterprise and broadcast markets.
With innovative solutions targeted at reliable and secure transmission of rapidly growing amounts of data, audio and video
needed for today’s applications, Belden is at the center of the global transformation to a connected world. Founded in 1902,
the company is headquartered in St. Louis, USA, and has manufacturing capabilities in North and South America, Europe
and Asia.
Belden, Belden Sending All The Right Signals, GarrettCom, Hirschmann, Lumberg Automation, Tofino Security, Tripwire and the Belden logo are trademarks or registered trademarks of Belden Inc. or its affiliated companies in
the United States and other jurisdictions. Belden and other parties may also have trademark rights in other terms used herein.