Industry Article

Enabling Industrial Network Design with Time-Sensitive Networking

August 12, 2021 by Christian Castel, NXP

OT (operational technology) and IT (informational technology) may have different real-time needs, but they're merged in the TSN (time-sensitive networking) Ethernet-based standards. Learn about the theory and hardware involved in implementing TSN in industrial network design.

Devices in a factory might have very different needs and potentially conflicting goals when communicating over a network. Operational technology (OT) traffic, such as machine control data and sensor value readings, typically requires fixed time delays, low latency, and predictable jitter. Information technology (IT) traffic, on the other hand, is data such as e-mail traffic.

In the IT domain, communication is typically best-effort, and accurate response times are not of utmost importance. Instead, the overall throughput is what typically matters. For OT, missing data at a certain time can lead to failures, and therefore the packets must reach their destination within certain real-time constraints.

Today, there are numerous different industrial protocols used to tackle this problem. However, time-sensitive networking (TSN) was built upon standard Ethernet, and it aims to create a unified standard for real-time communication over Ethernet. It achieves this by merging OT and IT traffic on a single networking cable and adding determinism to Ethernet. The goal is to reduce network delays and lower the latency between endpoints to ensure that certain packets reach their destination on time.

This article discusses TSN, the three essential TSN standards, and their typical use-cases. It also examines three NXP devices (the Layerscape LS1028A, the i.MX RT1170 crossover MCU, and the new i.MX 8M Plus) that allow embedded engineers to design modern connected real-time systems for industrial applications. 

 

What Is TSN?

TSN is not a single standard but rather a family of standards defined by the IEEE. The TSN standards form the foundation of the TSN architecture:


Figure 1. The TSN architecture comprises three layers. The IEEE standards form the foundation. TSN profiles sit on top of the foundation, and protocols make use of the profiles
 

TSN profiles sit on top of the TSN standards, forming the next layer of the architecture. These profiles concretely specify how to parametrize certain TSN features defined in the standards. For example, such a profile can contain parameters that describe how much accuracy in clock ticks is required in an application.

A relatively mature TSN profile is IEC60802, which defines the parameters for industrial applications. However, many other TSN profiles, such as automotive and medical applications, are currently in development. Therefore, the second level of the architecture configures and specifies the features defined in the TSN standards with a specific industry or application in mind. Finally, the top layer contains the protocols themselves.

 

Essential TSN Standards

The 802.1AS standard for timing and synchronization forms the base of TSN. IEEE 802.1AS builds upon the precision time protocol (PTP), allowing multiple devices in a network to synchronize their internal clocks, thus enabling more advanced features such as time-aware scheduling.

The 802.1Qbv standard allows TSN-enabled devices to combine OT and IT traffic and transmit both on a single Ethernet cable. In addition, this sub-standard includes a time-aware shaper, making it possible to create a schedule that states when certain packets can go out on a wire. The devices within the network agree to stick to that schedule, and they reserve time slots for specific packages. These measures lead to minimal and predictable jitter and latency when sending prioritized messages between two end-nodes:

 

 

Note that 802.1AS ensures that all devices on the network share a synchronized time base. Therefore, they all know when to send what type of traffic over the network cables.

802.1CB is another significant standard of TSN. This sub-standard allows system designers to create redundant communication streams over a network to increase the fault tolerance. With this feature enabled, 802.1CB capable networking switches will automatically duplicate specified packages when needed. Furthermore, when a TSN capable switch receives a unique message for the first time, it automatically discards all redundant copies later. Outsourcing these tasks to TSN-capable hardware removes the requirement for complicated software and lightens the load on the main CPU.

Finally 802.Qbu for frame preemption is one of the most important standards for industrial automation. By nature, industrial networks take very special care to a certain real-time approach that requires respect for very strict cycle times. Preemption helps keeping such timing by enabling to split a frame on multiple fragments that will be sent successively, unless an express frame shows up.

All standard frames can be interrupted and fragmented in multiple messages as long as the transmission of each of the messages can finish within a configurable period of time called guardband (802.Qbr). Such systems, used in conjunction with the preemption, prevent that too long or acyclic messages will extend the cycle time.  

The Fundamentals of Time-Sensitive Networking provides more detailed insight into TSN and some of the standards discussed here. 

 

Enabling Time-Sensitive Networking with NXP Devices

The Layerscape LS1028A, the i.MX RT1170, and the i.MX 8M Plus support TSN features to varying degrees. The following table summarizes the TSN standards and which NXP devices implement them:

 

 

TSN-ready hardware is the first step in creating a reliable TSN-enabled Ethernet network. NXP offers extensive software support for their products and a library of SDKs and software examples demonstrating various features.

The Layerscape LS1028A typically runs a real-time OS, such as Open Industrial Linux (OpenIL), or a different high-level operating system. The i.MX 8M Plus will also receive support for OpenIL shortly. NXP also provides open-source support for TSN as well as tools to configure it. For OpenIL, NXP offers open-source driver support for PTP. These drivers allow users to control the PTP hardware clock and time stamping. In addition to NXP’s software offerings, engineers can also choose from a range of readily available commercial software stacks.

 

Today’s TSN-enabled Devices

The NXP product portfolio offers a few devices that provide hardware support for time-sensitive networking in industrial settings. Some examples are the Layerscape LS1028A, the i.MX RT1170 crossover MCU, and the i.MX 8M Plus. These devices enable embedded system engineers to design the industrial equipment of the future by combining high processing power with an extensive set of peripherals, security features, and co-processors capable of tackling demanding tasks.

The LS1028A is a well-established applications processor based around two Cortex A72 processing cores. It's primarily intended for the automotive and industrial markets, and it comes with an integrated networking switch that supports various TSN features over four Ethernet ports. The LS1028A also offers a rich set of peripherals (such as a CAN-FD interface), various on-chip co-processors, a dedicated GPU and LCD Controller, and numerous security features. Target applications include networking equipment, industrial HID, and robotics.

 

Figure 2. The LS1028A block diagram. Image Source: NXP product website

 

The i.MX RT1170 family of MCUs utilizes two processing cores. An ARM® Cortex®- M7 core running at up to 1 GHz and a second dedicated Cortex®-M4 processor clocked with up to 400 MHz make these devices among the fastest microcontrollers available on the market today. Its performance and rich portfolio of peripherals and features make the i.MX RT1170 family of MCUs an ideal choice for a wide range of applications. The devices support up to two megabytes of SRAM and up to three Ethernet interfaces.

The i.MX RT1170 crossover MCU also offers a set of modern security and cryptographic functionality. For HMI applications, the devices include a dedicated 2D GPU and 2D accelerator and display interfaces. The i.MX RT1170 is optimized for low-power and low-leakage applications, allowing for efficient, fast, small, and cost-effective designs. 

 

Figure 3. The i.MX RT1170 block diagram. Image Source: NXP product website

 

The i.MX 8M family contains various application processors that target specific markets to fulfill the needs of a particular application. The i.MX 8M Plus is the latest model in the family and includes dedicated hardware for machine vision applications, an NPU unit with 2.3 TOPS for faster AI inferencing, improved LVDS, CAN real-time networking with TSN support, and a 2D/3D graphics accelerator.

Furthermore, the i.MX 8M Plus is currently the only device of the i.MX 8M family that offers multiple CAN-FD interfaces. It also comes with reliability features such as inline ECC for high-reliability industrial applications.

 

Figure 4. The i.MX 8M Plus block diagram. Image Source: NXP product website

 

The Layerscape LS1028A, the i.MX RT1170, and the i.MX 8M Plus are part of the 15-year NXP longevity program, which guarantees that the components will be available for sale for at least 15 years from product launch, which is especially useful for designers that need to go through long enablement or certification phases.

 

TSN Base Solution Example  

In this example, each component communicates with the other by leveraging the various TSN standards described above in order to keep a high level of synchronization and a guaranteed latency independently of the traffic running on the network.

The i.MX 8M Plus is used for image recognition and takes advantage of its ISP and embedded Neural Processing Unit (NPU) for an optimized operation and supports the real-time operations of the manufacturing line.

The i.MX RT1170 is used to guide the robot arm to pick the products off a virtual conveyor belt according to the analysis driven by the i.MX8M Plus.

In between, the Layerscape LS1028 runs a TSN network and relays the frames between the 2 other devices as well as to other potential nodes. TSN is used to ensure data is reliably delivered from the i.MX 8M Plus to the i.MX RT1170. 

In this example, a laptop is also connected in order to simulate best-effort traffic that would exist in any field implementation

Check out the link below for more details on this demo:

Machine Learning and TSN with NXP's i.MX 8M Plus

Industry Articles are a form of content that allows industry partners to share useful news, messages, and technology with All About Circuits readers in a way editorial content is not well suited to. All Industry Articles are subject to strict editorial guidelines with the intention of offering readers useful news, technical expertise, or stories. The viewpoints and opinions expressed in Industry Articles are those of the partner and not necessarily those of All About Circuits or its writers.