Making critical connections in the smart factoryFollow article
Over the last two decades, the internet has redefined almost every aspect of life. Consequently, we tend to think of this as the IoT age, where every 'thing' is connected to every 'thing' else. Although reality is much more complex than those three letters suggest, it is convenient to think in this way at a high level.
At a lower level, the internet has enabled much greater interaction between 'imagined' things, like control algorithms, and 'real' things, like switches. The IoT does an excellent job of bringing these together. For example, the smart home uses digital assistants, controlled by our voice, to turn appliances on and off. What these things really do, of course, is control the flow of electricity, through a connected power outlet or plug, to the appliance. What the digital assistant isn't able to tell the homeowner (at least, not yet) is whether the appliance is still plugged in to that outlet and is functioning.
The counterpoints to these smart plugs are the smart sensors. These are sensors that can tell the software what the temperature is inside the house, or if a window has been left open, for example. These are typical examples of IoT applications, but they fall short of providing a fully automated environment, in which the window is closed if the temperature is too low, and the heating only turned up if the windows are, in fact, closed.
Building management systems go some way towards this. But what they really strive to do is close the loop between the control software and the real world. This brings together two aspects: the online – or cyber – aspect of control; and the real – or physical – elements of what is being controlled. When these two halves do come together, they create what is termed a cyber-physical system, or CPS. When several of these systems operate in collaboration, they become a system of systems or CP SoS.
CPS is quietly changing all vertical industries. It is enabled by the internet, but it goes further than the IoT. As such, it has its own advantages and requirements that deserve to be acknowledged. By focusing on these, the supporting industries, such as the electronics market, will be better placed to maximise its potential.
Going beyond the industrial IoT
The differences between the IoT and CPS are also apparent in an industrial environment. Although the terms are often used interchangeably, along with Industry 4.0 and the IIoT, the concept of a system is perhaps more easily understood.
A cyber-physical system can be thought of as being a closed loop that is, perhaps paradoxically, also connected to the internet. We can think of it as a closed-loop because the control elements that are needed to monitor and modify the physical aspects are contained inside the boundaries of that system. Of course, some of those boundaries may be flexible and extend all the way from the edge of the system to the cloud.
By elements, we're really talking about all the sensors and actuators needed to control a system and the algorithmic assessment that bridges the gap between the two. In an industrial environment, this would mean more than just turning a machine on or off; it might extend to sensing whether there is anything on the conveyor belt that feeds the machine, or the rate at which it is using consumables such as cooling fluid.
This combination of sensing, computing and acting really defines a CPS. The data generated will typically fall under the domain of the IT (Information Technology) team, but the way that data is used to implement physical change is very much the domain of OT (Operational Technology). In this way, CPS is instrumental in bringing these two domains closer together, with the aim of increasing productivity.
The cyber-physical system challenge
Although we can think of a CPS as being a closed system, the reality is that modern systems are very much open and subject to the influence of other systems. A good example of this is the autonomous vehicle. In a test environment, autonomous cars work really well. To date, we have yet to make autonomous vehicles that can operate just as well in an open system.
For example, systems of CP systems are designed to operate collaboratively, so that each subsystem knows just what it needs to do and can react to all known conditions. It becomes more difficult if one CPS, such as an autonomous car, comes into contact with another CPS, such as a smart city traffic control system. This introduces an element of unpredictability, but it also highlights the fact that any one CPS may also need to rely on information coming from another, largely unrelated, CPS.
Another important point to appreciate is that people are very much part of the cyber-physical domain. Often, a CPS needs to include people as part of the system. That may mean treating a human as part of the control algorithm, or as part of the system's input or output. Obviously, the system has no direct control over a human, but it may communicate indirectly through a user interface, lights or alarms.
The presence of people is perhaps most apparent in an industrial environment. Despite the continued adoption of automation, human operators still occupy many roles in industrial settings. One of those roles is machine tending, the objective being to maintain a process by, for example, introducing raw materials, removing processed materials or adjusting operating parameters. Robot machine tending is a growing trend but, currently, it is still largely a role occupied by a human operator. It is difficult to integrate a human into a system at this level, but it is something that engineers developing cyber-physical systems need to contend with.
Hard real-time requirements
What may really set CPS apart from the IoT is its time-sensitive nature. In an industrial process, response time is often critical and may be compounded by the relationship between many dependent processes. Unlike the general IoT, where data packets may arrive out of order and without a guaranteed level of service, industrial cyber-physical systems will employ time stamps to ensure the validity of data.
The influence of the IoT continues to redefine CPS. Sensors and actuators are getting smarter, and processing is moving closer to the edge of the network. As a result, cyber-physical systems are increasingly extensible and heterogeneous. They could include a wireless sensor network as a dedicated subsystem, for example, which may need to integrate with legacy subsystems that use wired communications.
As the value of CPS increases, its flexibility becomes crucial. In an industrial application, the cost of replacement can be prohibitive. However, industrial applications can benefit massively from the adoption of cyber-physical systems. This has created a landscape that includes many different types of communication technologies.
Sustaining this diversity is aided in part by the use of common protocols. At the physical layer, Ethernet is the single most deployed wired technology. Wireless technologies include Wi-Fi (the wireless equivalent of Ethernet), Bluetooth, Zigbee and other physical layers. In order to better serve the industrial sector, Ethernet was ruggedized, making it suitable for use in harsh environments.
However, industrial Ethernet has its own limitations. Firstly, connections are often based on modified versions of standard Ethernet, which has introduced incompatibilities at the protocol level. Secondly, the industrialisation of Ethernet has led to cables and connectors that are more robust but, also, larger and more expensive. This means the deployment of industrial Ethernet is restricted to the factory floor and relatively large pieces of equipment.
As cyber-physical systems continue to push the intelligence close to the edge, into the sensors and actuators themselves, there is a need to provide high bandwidth, reliable connectivity to the smallest device. As industrial Ethernet has been unable to penetrate this far, it has been left to other solutions. To date, that has mostly involved the use of the wireless technologies mentioned above. That may be about to change, as a new Ethernet technology promises to unify the IIoT and offer connectivity at a new level.
Single-pair Ethernet extends to the field level (Source: SPE Industrial Partner Network)
Will single-pair Ethernet unify the IIoT?
The headline here is that single-pair Ethernet (SPE) uses just one pair of conductors, as opposed to the four pairs needed in standard Ethernet. This has significant implications for data density; the amount of bandwidth that can be fitted into a given area. For small, smart devices like sensors and actuators operating in a cyber-physical system, this means they can now be directly connected to the same Ethernet backbone that runs through the entire system.
In doing so, the IT and OT domains are brought even closer together, using a physical layer that can support up to 1Gbit/s bandwidth while also employing technologies that are critical in an industrial environment, such as TSN (Time Sensitive Network).
Because it uses TCP/IP, data is packetized inside the smart device itself. The SPE Industrial Partner Network is actively promoting SPE and driving standardisation. This includes the standard IEC 63171-6, which was published in early 2020. It was the very first standard for SPE in industrial applications and has been adopted by all of the leading standards bodies. Standards for the cabling and cable components have also been published. The Ethernet protocol standard, IEEE 802.3bw, covers the implementation that guarantees interoperability.
By removing all but one pair of conductors, the cable and cable assembly is simpler, smaller, lighter and cheaper. However, there are some important differences to note. Primarily, the frequency of operation has increased from 100 MHz to 600 MHz. This leads to a reduction in cable length, from 100 metres for 8-wire Ethernet to 40 metres for SPE using shielded twisted pair (Type B), and just 15 metres for unshielded twisted pair (Type A). However, the transmission distance is arguably a secondary consideration for SPE, as it will enable short-range, high-bandwidth communication for a new generation of smart sensors and smart actuators.
For these reasons SPE isn't expected to replace Ethernet as we know it, but it will augment it in applications such as cyber-physical systems. Semiconductor manufacturers will be enabled to create integrated solutions that include Gbit Ethernet, for smaller and smarter devices sitting at the network's edge.
For example, cameras are increasingly being used as part of the feedback loop in closed systems, typical of CPS. The amount of data generated by these high-resolution image sensors is only increasing, which is putting pressure on the traditional backhaul solutions used. Without an alternative, traditional industrial Ethernet would need to be used here, which has all of the limitations mentioned earlier. We can expect SPE to open up this and many other applications like it, to enable more high-bandwidth sensors to be used in more places around the system.
Another important feature offered by SPE is power. Specifically, similar to Power-over-Ethernet, or PoE, SPE offers Power over Data Line, or PoDL. This modulates power and data over the same two conductors, delivering up to 50 W to the endpoint. This is enough power for many of the target applications, but the industry is also looking at ways to provide two dedicated power conductors in SPE cables, which could extend this to other devices that require more power.
At a high level, the industrial IoT represents a range of technologies that aim to increase productivity and provide greater insights. At a deeper level, cyber-physical systems describe the actual implementation of that vision.
As more verticals embrace the CPS approach, we can expect new technologies to emerge. Many of these technologies will have a role in the industrial sector. The introduction of SPE could be instrumental in the way CPS develops in the future and be fundamental in the way Ethernet continues to dominate high-speed communications.
For more on single-pair Ethernet refer to this article from Connector Geek