Time Sensitive Networks For Industrial Automation Systems Volker Goller Systems Application Engineer ©2016 Analog Devices, Inc. All rights reserved.

What is TSN? TSN is all about Layer 2 of the OSI model and is an extension to IEEE 802.1 to make Ethernet Deterministic More Robust and Reliable Features are targeted at industrial, automotive, and AVB First and foremost, TSN adds time to standard Ethernet Common understanding of time and clocks TSN uses time to ensure predicable message delivery Traffic Shaping (Bandwidth, TDMA, etc.) TSN adds robustness and reliability with: Seamless Redundancy Preemption 1 Physical Raw Bits On Media 3 Network IP 4 Transport TCP/UDP/.. 5 Session 6 Presentation 7 Application OPC/PROFINET/… 2 Data Link A Data Frame 2 Data Link A Data Frame TSN ©2016 Analog Devices, Inc. All rights reserved.

Where did TSN come from? Bridging was originally defined by IEEE 802.1D When VLANs were added to this specification it became IEEE 802.1Q However, this didn't address the determinism and robustness requirements for industries like Factory Automation Starting in 1999, various industry organizations were formed to address these requirements PNO, ODVA, ETG, etc. developed their own "non-standard" deterministic Ethernet variants In 2006, IEEE began to address determinism with the formation of the Audio Video Bridging (AVB) Task Group in 2006 In 2012, the charter of the AVB Task Group was expanded for a wider range of deterministic Ethernet requirements This is the 802.1 Task Group we now call Time Sensitive Networking (TSN) ©2016 Analog Devices, Inc. All rights reserved.

It provides a common Layer 2 for all upper layer protocols Why is TSN important? It provides a common Layer 2 for all upper layer protocols ©2016 Analog Devices, Inc. All rights reserved.

What are the organizations driving TSN? Purpose Develop standards and recommended practices in the areas of 802 LAN/MAN architecture, internetworking, security, and overall network management A community creating an interoperable ecosystem servicing the precise timing and low latency requirements of diverse applications using open standards through certification Bring together the organizations and technologies necessary to accelerate the growth of the Industrial Internet ©2016 Analog Devices, Inc. All rights reserved.

How does TSN work? A set of 802.1 sub-standards, addressing different needs Not all sub-standard have to be implemented The important sub-standards for Industrial Automation are: 802.1AS (REV) Time Synchronization 802.1Qbv Time Aware Traffic Shaper 802.1Qci Ingress Policing 802.1CB Seamless Redundancy 802.1Qcc Network Management IEEE Alphabet Soup! ©2016 Analog Devices, Inc. All rights reserved.

802.1AS (REV) Time Synchronization Purpose: provide network with accurate, reliable, simple-to-use time How? Define a time synchronization profile of IEEE 1588v2 One or more Grand Masters provide time A Best Master Clock Algorithm (BMCA) is used to select to select the Grand Master Provides for Redundant GMs Synchronize time across the network i.e. the grand master is the master to its slave device, that slave device is a master to its slave, and so on… Peer delay is used to keep time synchronized Uses a one-step process with backward compatibility to two-step Performed periodically up to a 10 s rate Supports working clock and monotonic clock Working clock may jump (DST!) – Good for time stamping events (sequence of events) Monotonic clocks do not jump – Good for continuous processes, including TAS ©2016 Analog Devices, Inc. All rights reserved.

802.1Qbv Time Aware Shaper Purpose: Avoid interfering traffic How? Divide Ethernet traffic into different classes ensure only one traffic class has access to the network create a time-protected “channel” used by that traffic class alone Each egress port is gated, messages queued until scheduled time arrives Send all queued messages until time window closes Builds on 802.1AS so everyone has accurate time in network ©2016 Analog Devices, Inc. All rights reserved.

Preemption, IEEE 802.1Qbu / 802.3br (1 of 2) Purpose: Ensure high priority traffic arrives at a destination with a fixed latency How? Break-up frames into smaller pieces Preemption allows unscheduled traffic to be prioritized to have a fixed latency Without preemption, each bridge has it's own delay and maximum frame delay With preemption, each bridge has it's own delay and a minimum fixed frame delay Even though messages arrive on-time using 802.1Qbv, there can still be a variable latency in which streams arrive at their destination. This is because streams vary widely across the network. So the concept of preemption was introduced in order to provide delivery of messages with a fixed latency – even if the messages are not scheduled. Fixed latency is accomplished by breaking frames up into small, fixed sized pieces. To illustrate the how preemption helps, consider the network in the top part of the diagram. While each switch in the network has a fixed bridge delay, the frame delay is variable due to the variable size of the stream. So as a frame traverses the network, it will encounter fixed bridge delays and variable frame delays. In other words, as a frame traverses the network from talker to listener, all of the bridge delays and all of the variable frame delays accumulate as the frame gets to the listener. This is shown by the collection of delays at the listener. Each time a frame traverses the network it will encounter this total variable delay. Now, consider the network in the bottom part of the diagram. With preemption, each switch now has a fixed bridge delay and a fixed frame delay. As a frame traverses the network, it will encounter fixed bridge delays and fixed frame delays. And each time a frame traverses the network it will encounter this total fixed delay every time.

802.1Qbu Preemption Purpose: Reduce impact of frames interfering with scheduled traffic How? Break-up frames into smaller pieces Scheduled traffic requires a "guard band" so non-scheduled traffic frames don't cause scheduled frames to be late Frame transmission before the start of a reserved time period can extend outside its allocated window, interfering with more critical traffic Guard bands solves this problem, but reduce the transmission efficiency To solve this, preemption breaks interfering frames into smaller “fragments” 1. 2. 3. ©2016 Analog Devices, Inc. All rights reserved.

802.1Qci Ingress Policing Purpose: Prevent traffic overload conditions (DDoS, erroneous delivery) from affecting the receiving node How? Filtering traffic on a per stream basis by providing an input gate for each stream Input gate serves to enforce a "contract" between the talker and listener Contract functions could be: Pass/no-pass "Leaky bucket" policing Time/bandwidth-based Threshold counter Burst sizes Packet sizes Misuse of labels, etc. Packet processing Port ©2016 Analog Devices, Inc. All rights reserved.

802.1CB Seamless Redundancy Purpose: Provide lightweight redundancy for reliable delivery of traffic streams How? Frame replication and elimination Send two copies of a message along maximally disjoint path to ensure delivery Use of redundant paths minimize packet loss due to Link or device failures Congestion Discard duplicate frames upon reception ©2016 Analog Devices, Inc. All rights reserved.

Centralized Configuration, IEEE 802.1Qcc Stream A Stream B Qbv AS Stream A Stream B Qbv AS Qbu Qbu The name of the Qcc specification is actually not “Centralized Configuration”, but for the TSN industrial segment, this name fits and is more appropriate to the concept of what we are trying to accomplish. Each of the TSN features we have discussed require configuration parameters to setup the operation of the network. Qcc provides a way to set parameters that answer questions like: who will be the 802.1AS grandmaster and who will be the backup grandmaster? What is a talker’s Qbv schedule time for Streams A and B? What path should these streams take? What redundant paths should these streams take if the network is supporting 802.1CB? Which switches need to support preemption? What ingress policing policy will be used so talkers and listeners are not overwhelmed by certain types of streams on the network? It is the job of the Centralized User Configuration, or CUC, and the Centralized Network Configuration, or CNC, to get the answers to these questions. The CUC and CNC are new terms that simply identify their function generically. In terms of today’s systems, the CUC function is typically a PLC, and the CNC function is typically software residing in an infrastructure switch. The reason for making these terms generic is to allow the function to reside anywhere in the system and not in a fixed place like a PLC or an infrastructure switch. The same holds true for the terms “Talker” and “Listener”. These terms identify their function generically, and most of the time a talker can also be a listener. In today’s industrial systems, talkers and listeners are field devices that communicate to a PLC – like a Remote IO unit in factory automation. Qci Qci CB CB

Centralized Configuration, IEEE 802.1Qcc CUC Gets End Station capabilities Determines stream requirements for Talkers and Listeners Tells the CNC the stream requirements CNC gives CUC End Station network configuration CUC distributes network configuration CNC Determines network topology Gets End Station stream requirements from CUC Calculates network schedules Determines Stream paths Policing requirements When preemption is required Provides CUC with End Station network configuration This slide provides a brief synopsis of the role a CUC and CNC play in a TSN network. Starting with the CUC, keep in mind that it is the CUC’s job to use the network along with the talkers and listeners to perform some operation. This may be as simple as a PLC communicating to devices in a conveyor belt that controls the movement of product, or as complex as a body welding cell in an automotive plant. The CUC gets the capabilities of the end stations and determines the stream requirements for the talkers and listeners. It then tells the CNC these requirements and the CNC, in return, gives the CUC network configuration for all the end stations. The CUC then distributes the network configuration to the end stations and prepares the system for operation. The CNC first determines the network topology so it has an understanding on how to route streams from talkers to listeners. After receiving the end station stream requirements it performs the schedule calculations of the streams over the network. From there, it must also take into account the stream paths and factor in redundancy. It also determines the policing requirements for the end devices and whether preemption is required in order to meet network timing. Finally, the CNC provides the CUC with the network configuration for the end stations. What the CNC and CUC do in a TSN network in not unlike what the Industrial Ethernet protocols already do today. It is just that TSN is an attempt to do these functions in a standard way using a standard set of parameters.

Where are we now? AVNU Industrial Segment IIC TSN Testbed Theory of operations complete Goal is to certify devices in 2017 IIC TSN Testbed Network operational with initial set of vendors Goal is to expand as use of TSN grows and to explore the technology Standardization bodies (PI, ODVA, ETG, OPC UA, etc.) across the industry working on integrating TSN into their specifications IEEE is in the process to finalize the remaining TSN sub-standards ©2016 Analog Devices, Inc. All rights reserved.

TSN and its future in Industrial Automation TSN is an opportunity to converged on a common Layer 2 technology for all Industrial Ethernet protocols It has the power to provide interoperability at Layer 2 in order to achieve a true converged network TSN can be adopted by existing industry standards like PROFINET or EtherNet/IP to gain Better and broader hardware support Converged real-time model at Layer 2 A path to Gigabit Ethernet ©2016 Analog Devices, Inc. All rights reserved.

For further information IEEE 802.1 TSN Working Group http://www.ieee802.org/1/pages/tsn.html Avnu Alliance http://avnu.org/ Industrial Internet Consortium http://www.iiconsortium.org/ IIC TSN Testbed http://www.iiconsortium.org/time-sensitive-networks.htm And finally, if you would like to know more about any of these organizations or more about the TSN specifications, use the links provided here to learn more. Pause the webinar now if you would like to write down any of these links. [[[wait a few seconds before closing]]]

THANK YOU! Questions? ©2016 Analog Devices, Inc. All rights reserved.