Challenges to Implementing the Internet of Things for Industrial Applications
Building industrial IoT systems is getting easier as both hardware and software building blocks come together to provide robust performance and security.
The Internet of things (IoT) takes on many forms, ranging from a handful of smart home devices linked together and connected through the Internet via a gateway, to networks of hundreds or thousands of sensors and other connected devices in a smart office building, a factory floor, the power grid, or jet engines on a plane. The industrial use of IoT solutions takes on its own identity as the industrial IoT (IIoT), and the IIoT has more stringent performance requirements than consumer-connect solutions. Advanced security, quantified real-time performance, the ability to connect legacy equipment to the network, and the ability to handle huge amounts of data being collected from thousands of end-points are key performance characteristics that differentiate the IIoT from consumer IoT solutions in smart homes.
As explained in a white paper written by several product managers at Moxa, in industrial automation applications, collecting data from field devices will become more important than ever. Temperature, motor speed, start/stop status, or video footage, can be used to gain new insights to increase competitiveness. For example, you can determine how to optimize your energy usage, production line performance, and even when to do preventive maintenance to reduce the amount of downtime. However, these devices often speak different languages: some use proprietary protocols, whereas others use open standard protocols. Whatever the case may be, you will need to find an efficient way to convert back and forth between one or more protocols.
The IIoT thus presents many challenges to designers and system implementers. The first critical issue is to get all the “things” connected so that data and commands can seamlessly flow across the network. Next, all the data being collected must be turned into intelligence (little data from many sensors or endpoints turns into big data, and that data must be analyzed to extract useful intelligence). Often, groups of sensors are collected into a bank and that bank feeds into a gateway that will preprocess the data, reducing the amount of data sent back to the host system. That, in turn, will reduce the network bandwidth requirements, lowering the cost to collect the data since companies could use less-expensive lower-bandwidth interfaces.
Factory-floor applications and many other industrial applications bring with them legacy equipment issues, where equipment that might be 10, 20 or more years old would have to be adapted to communicate over the IIoT. But such equipment might employ many different communication protocols and interfaces, thus making the equipment difficult to configure and integrate. Thus, designers will have to deal with interoperability and scalability challenges to craft large, scalable heterogeneous networks that can operate in harsh environments—whether on the factory floor or dispersed across the country on the power grid.
Once properly set up, the IIoT can improve productivity and make users lives easier. However, an unreliable network would be the bane of any system developer since users would see longer system downtimes, risks from system breaches from hackers, malware, and organized cyber security attacks, and unstable operations as the diagram from Moxa illustrates (Figure 1). This is especially prevalent when wireless connections link portions of the system together.
To deal with all the security and interface issues, Moxa has developed Fieldbus-to-Ethernet gateways with smart functionality to deal with breaches and system performance issues. The company’s MGate gateways not only connect serial devices to Ethernet systems, but they also allow multiple connections and make it easy to employ various Ethernet protocol formats, such as Modbus TCP and Ethernet/IP.
One Starting Point
In addition to all the hardware issues, there are many software aspects that designers must deal with, from the operating system software to application programs that run on the endpoints such as sensor nodes, machines on the factory floor, equipment in the field, or jet engines on a plane. One starting point from Wind River, its free scalable real-time operating system (RTOS) dubbed Rocket, is targeted for 32-bit microcontrollers and is a good fit for sensors, wearable products, industrial actuators, wireless gateways and other resource-constrained devices.
The Rocket RTOS lets designers develop, debug and deploy applications for small, intelligent devices from any browser. Included in Wind River’s Helix App Cloud, a cloud-based software development environment, the Rocket software allows designers to start developing IoT applications in minutes. To get started, designers just have to create an App Cloud account, connect their prototype board or use Wind River’s simulator, and then start writing the application code. A browser interface allows designers to work from anywhere to code and debug the application software, and prototypes can be developed without requiring any hardware purchases.
The kernel is a small footprint kernel designed for use on resource-constrained systems, from simple embedded environmental sensors and LED wearables to sophisticated smart watches and IoT wireless gateways. The software is tuned for memory- and power-constrained devices (as small as 4 Kbytes of memory).
The proliferation of IoT/IIoT devices brings with it significant security challenges. Hackers as well as cybercriminals can often find a way to enter the networks and wreak havoc by compromising system functions, holding system data for ransom, or siphoning off valuable system data for sale to other criminals. Both hardware and software measures to protect the networks are needed to prevent or minimize network intrusions and alert companies when an intrusion is taking place. To that end, microcontroller (MCU) vendors now include random-number generators as well as full encryption/decryption blocks on their chips to provide real-time encryption/decryption. Previous generation MCUs typically used software encryption/decryption and the software overhead slowed system throughput.
In addition to the RTOS software challenges, the ability to handle hundreds to thousands of sensor or endpoint inputs often requires the use of gateways or edge devices that can aggregate the data coming from the sensors or endpoints, potentially preprocessing the data and then forwarding the data to the host system. These Internet gateways will often handle multiple communication protocols such as WiFi, Bluetooth, ZigBee, LoRa (long-range wireless), and still others. The gateway will translate all the inputs into a common communication protocol, typically WiFi or wired Ethernet.
Situated between the sensors and the gateways, sensor hubs usually perform some degree of data reduction, extracting the key information from the sensor data streams to reduce the amount of data sent back to the host system (Figure 2). There are many vendors that offer gateways for IoT applications. Offering various reference designs for system gateways, Intel solutions, for example, span the range of simple gateways based on its low-end Quark processor, the X1000 system-on-a-chip, or its higher performance Atom processors such as the E3826. The gateway products provide connectivity from the sensors all the way up to the cloud and enterprise systems. The more intelligent gateways can preprocess and filter data to deliver selective results to the host. Local decision makes it easier for gateways to connect with legacy systems, and a hardware root-of-trust along with hardware supported data encryption can provide end-to-end security.
Thus designers can scale processor performance based on the processing requirements in the gateway. Additionally, the processors support multiple operating systems—Wind River, Microsoft, Ubuntu, and others. Robust security provided by McAfee (now part of Intel) embedded control security technologies tightly integrate with on-chip hardware based security features to provide seamless secure data flow from the edge devices to the cloud, protecting the data while in flight or at rest. Additionally, pre-integrated manageability options provide developers with an out-of-the-box offering they can build upon and customize.
The sensors that collect all the data have come down in price—from tens of dollars to just a dollar or two today—thanks to the large volumes consumed by the mobile communications and compute products such as smartphones and tablets. They are also getting more highly integrated, offering anywhere from a single-axis accelerometer to a nine-axis multi-sensor (three-axis accelerometer, three-axis gyroscope and three-axis compass) solution in a single surface-mount package such as offered by Invensense. And along with the sensors, some vendors also include some data processing functions on the sensor chip to preprocess the raw data stream, thus reducing the amount of data sent to the gateway. The low cost of the today’s multiaxis sensors allows designers to use them throughout the system to monitor many more aspects of system performance than in previous system generations.
The software and hardware building blocks to build a robust IIoT system are now readily available from multiple suppliers. Development tools now let designers prototype systems in minutes to a few hours. But there are still many system aspects that designers must address—deciding which communications protocol standard (as mentioned earlier) will be key to system scalability and performance. Additionally, for many systems the choice of either an RTOS or other operating system can be critical, but there are no standards that designers must follow. The key issue will be to determine if the system will be “closed” and only allow hardware from one vendor to do all the work, or be “open” and allow products from different suppliers to interoperate.
For smart homes, the Thread Group has developed an approach to connect and control products in the home. Built on open standards and using the IPv6 and 6LoWPAN protocols, Thread provides a secure and reliable mesh network with no single point of failure as well as simple connectivity and low power consumption. On the industrial side, the Industrial Internet Consortium is trying to define standards to encourage interoperability. The IIC is a worldwide not-for-profit, open membership organization that was formed to accelerate the development, adoption, and widespread use of interconnected machines and devices, intelligent analytics, and people at work. It helps achieve this by identifying the system and subsystem requirements for open interoperability standards and defining common architectures to connect smart devices, machines, people, and processes that will help to accelerate more reliable access to big data and unlock business value.