The perfect combination of edge nodes and industrial IoT

Wherever you look these days, you can’t help but hear a lot about the Industrial Internet of Things (IIoT). And, for different industries, this trend manifests itself in different ways. For example, Industry 4.0 is a concept developed specifically for production equipment. In the power grid, IIoT manifests itself as smart grids; in the oil and gas industry, IIoT manifests itself as digital well sites. Although different forms of IIoT have their own specific expressions and processes, the technology and benefits provided by IIoT are largely the same. Although industry leaders are eager to take advantage of IIoT, it is difficult to imagine what it will look like when 50 billion devices are connected by 20201. Experts estimate that half of these new connected devices deployed between 2015 and 2025 will come from the industrial sector2. This means that engineers and scientists will be the drivers of IIoT in factories, test labs, power grids, refineries, and infrastructure.

With IIoT, engineers can expect three main benefits:

•          Increase uptime with predictive maintenance

•          Improve performance with edge control

•          Improve product design and manufacturing through real-world connected data

To realize these benefits of IIoT, design teams must rely on several core technologies. Whether building online monitoring systems, smart manufacturing machines, or testing physical or electromechanical systems, a key commonality is the need for edge intelligence. The more complex the system, the greater the need to make real-time decisions. For example, in structural testing of wind turbine blades, the ability to collect large amounts of high-resolution analog waveform data is critical to understanding the characteristics of blade behavior. At the same time, this data needs to be processed to provide input to the control system that drives the blade to ensure that the test is performed under known conditions. It is no surprise, then, that experts estimate that at least 40% of IoT data will be stored, processed, analyzed, and acted upon at the edge3. To maximize performance and reduce unnecessary data transmission, users must delegate decision-making to edge nodes deployed at or near the equipment.

Figure 1. By 2019, at least 40% of IoT data will be stored, processed, analyzed, and acted upon at the edge.

Over the years, NI has invested in two high-quality control and measurement platforms: CompactRIO and CompactDAQ. Both platforms are flexible and modular with software-defined capabilities. Built-in I/O interfaces and C Series I/O modules provide high-precision I/O and specific measurement signal conditioning, so users can connect any sensor or device through any bus. CompactRIO provides a real-time processor and user-programmable FPGA, which is particularly suitable for high-speed control, while CompactDAQ provides the best software API NI-DAQmx in its class, which is ideal for data acquisition.

However, as we begin to implement these systems, new challenges emerge—especially as the physical size of the systems increases and the number of sensors increases. Using the structural testing example again, to fully understand the performance of a wind turbine blade, we need to equip the entire mechanism with sensors to measure strain, pressure, load, and torque. These sensors all generate analog signals, and to get the most and most useful information, we need high-speed, high-resolution measurements. For large-scale applications such as these, we may need to deploy hundreds or even thousands of sensors throughout the system. While collecting all this data, we also need to be able to process it in real time so that we can provide output control to all the actuators of the control system.

There are several challenges when trying to develop such a system:

•          Synchronize thousands of channels and numerous measurement systems

•          Synchronize the control system so that all operations take place at the right time

•          Synchronize measurement and control systems

These challenges are further exacerbated as systems continue to expand and the number of measurement and control systems used continues to increase. Synchronization between measurement systems and between control systems is not a new challenge. Today, this can often be achieved through a signal-based approach, where physical wiring is used to route a common timebase or signal to distributed nodes. However, this has limitations in terms of distance, scalability, and noise risks. Another option is to leverage protocols based on common standards such as Ethernet. Ethernet offers a high degree of openness and interoperability, but there are no latency limits or bandwidth guarantees. To address this challenge, engineers developed a custom version of Ethernet, often referred to as hard real-time Ethernet. Typical examples include EtherCAT, PROFINET, and EtherNet/IP. These custom versions of Ethernet provide hard real-time performance and best-in-class low latency and control. However, each version requires modifications to the hardware and software of the network infrastructure, which not only increases costs but also means that different devices from different vendors cannot operate on the same network.

New technology is now coming to market to address this synchronization challenge, called Time Sensitive Networking (TSN). TSN is an updated version of standard Ethernet that is not only open and interoperable, but also provides the same low latency and bandwidth guarantees as hard real-time Ethernet. Specifically, TSN provides three key components: time-based synchronization, traffic scheduling, and system configuration. The synchronization function is based on the IEEE 1588 Precision Time Protocol profile, which provides sub-microsecond synchronization over the network. In addition, traffic scheduling and system configuration provide deterministic data communication, so users can schedule and prioritize time-sensitive data (such as control signals) on the network.

An important feature of TSN is the convergence of time-sensitive traffic and other Ethernet traffic. Since TSN is a feature of the Ethernet standard, the two new features of time synchronization and deterministic communication can be supported on all Ethernet communication networks. This means that a single port on a measurement or control system can perform deterministic communication while also remotely updating user interface terminals and supporting file transfers. TSN is a new addition for many industrial applications, such as process and machine control, where low communication latency and minimal jitter are critical to meet closed-loop control requirements. Time-based Ethernet synchronization can also eliminate the wiring required for signal-based synchronization, which significantly reduces wiring requirements compared to traditional monitoring applications and physical system testing (such as structural testing), enabling simpler, cost-effective solutions without sacrificing reliability.

Figure 2. Time-Sensitive Networking is an update to standard Ethernet that includes time-based synchronization, traffic scheduling, and system configuration.

NI products are also increasing their support for TSN, and the latest controllers for the CompactRIO platform are a typical example. Users can add these new controllers to TSN networks and support data synchronization and deterministic communication, making them ideal IIoT edge nodes.

Figure 3. The latest CompactRIO controllers are TSN-enabled, enabling synchronous and deterministic communications.

The introduction of TSN is an important step in solving the challenge of synchronization across the system. Engineers developing these systems are also focusing on how to reduce overall system complexity while maintaining or improving reliability. Because measurement and control are typically independent subsystems, tools, programming environments, and data acquisition mechanisms are independent of each other. Control systems such as PLCs are typically programmed in IEC 61131-3 languages, which operate on single-point data. This type of data is well suited for control applications, but not for extracting information - so we need waveform data. Similarly, measurement systems use waveform data to provide the required information, but it is not suitable for sending single-point control signals or reacting to single-point control signals deterministically.

This feature of measurement and control systems is very straightforward. Over the past few years, the convergence of measurement and control systems has progressed very slowly. Each system has added new features so that more measurement systems can have some control functions, or control systems have some measurement functions. With the release of the latest CompactRIO controllers, we have seen a significant improvement in this convergence. In addition to leveraging real-time processors and FPGAs to implement deterministic control applications, the new controllers can also be programmed using the easy-to-use and powerful NI-DAQmx driver to implement measurement applications. NI-DAQmx is more than just a basic hardware driver. It not only provides configuration and troubleshooting tools, step-by-step configuration tools, but also provides powerful and intuitive APIs that greatly improve work efficiency and performance. Engineers can use the NI-DAQmx API to write custom programs, implement powerful timing and synchronization functions, and perform advanced control and monitoring tasks. For users who need to synchronize high-channel-count systems, develop decision-based recorders, or automate laboratory experiments, hundreds of examples, a vibrant community, and first-class local support can help them quickly transition from concept to deployment. Through this convergence, they can use the same hardware and a single software tool chain to acquire, process, log, and respond to incoming data directly at the edge, ultimately reducing system cost and complexity.