Architecture of the Industrial IoT: Questions and Answers
The Internet of Things is creating tremendous hype, and there is real power and value to the concept, but it requires a new approach to make it work. A now-archived webcast on the topic, "Architecture of the Industrial IoT" with Mike Fahrion, B&B Electronics director IoT product and strategy, had extra questions beyond those answered verbally and recorded. Answers follow on industrial Internet of things (IoT) topics including industrial network size, overlay systems, IPv6, cooperation between IT and operations, and security, among others. Click here to access the IoT webcast, sponsored by B&B Electronics.
Webcast question: In an industrial environment, a discrete manufacturing facility, what are the average sizes of networks that normally would be considered?
Fahrion answer: Number of nodes is going to be dependent on what return can be had on the investment. For example, if a plant has 500 high-power motors where a theoretical 20% improvement in energy consumption would create a positive financial return, then there is a motivation to add 500 new network endpoints. In any plant you could do an audit of "important stuff" that may be important because of energy/resource consumption, or the high-cost loss of productivity if it fails, or some other reason. That count of "important stuff" probably relates closely to the number of endpoints appropriate for that plant.
Q: What is the state of the art for energy harvesting for sensors without hardwired power?
Fahrion: We’re studying this quite a bit now. Outdoor solar is easy, and indoor solar is manageable with low-power end nodes (think of low power in the realm of 300 microwatts). Applications with thermal and vibration properties [for energy harvesting to power the sensors] are both manageable as well, but are more application specific.
There’s a nuance to energy harvesting that needs to be thought through as well. Most harvesting applications require a storage element, presumably battery or capacitor, and those storage elements have a life expectancy of their own. Once we get low enough power consumption to be a practical harvester, we’re also likely at a point where a primary battery scheme could last 5 years or more. There’s a return on investment (ROI) decision to make on the value of the harvesting mechanism, which likely extends the service cycle from (perhaps) 5 years to 10 or 12 years, not to infinite, though.
Q: Is IPv6 going to become widely adopted sometime soon?
Fahrion: It doesn’t appear that way, at least not in the industrial IoT arena, where gateways and private networks are the norm, along with product lifecycles of 10 years or more.
Q: Do you have any recommendations or best practices on how to foster the cooperation between skill sets [of personnel in IT and operations areas]?
Fahrion: In the industrial world I think we’ve all run into some friction between automation and IT. My perception is that is improving all the time. The industrial IoT is really combination of operations technology (OT) and information technology (IT). Any companies that are struggling with that cooperation are at risk. Like any change management, finding a common goal to get teams united around and push for a "quick win" is a good strategy. My tip: Lots of people connect personally with the theme of energy conservation. Most every industrial plant has an abundance of opportunity to reduce consumption of energy or natural resources. Once you get cross-functional teams to get behind a common goal, you’ll likely see good things happen.
Q: The overlay network idea implies network architecture with excessive bandwidth and enabled QoS features to me. I didn’t hear any comments on either of those concepts. Shouldn’t people be cautioned to implement more than they need today to support overlay network traffic in the future?
Fahrion: The overlay networks we’re seeing so far are not sharing the same IP network; they’re "drop-in" networks, typically of gateway-attached wireless sensor networks or IP devices connected through an independent network behind a cellular router. In time perhaps we’ll see more willingness to leverage existing IP networks; and you’re correct that building today’s network with tomorrow in mind is a good practice!
Q: Broker concept appears as though it might be recursive to me (for example, process brokers feeding area brokers feeding a plantwide broker to support enterprise brokers). What determines when and where to establish a broker?
Fahrion: Yes, you could have a hierarchy of brokers organized as needed to support the application needs. For example, an application which required many client-to-client transactions may use a different hierarchy than one where many clients are typically publishing data up to a central location.
Q: Security is expensive for end devices. What are your thoughts on where to implement security layers? What are the most prevalent standards being developed towards a unified IoT language (protocol)? Is Europe further along standardization than the U.S., and if so, should we look there for guidance when considering application development?
Fahrion: Security is a topic that requires a lot of attention and diligence. I’m not sure that there is anything "new" about IoT in comparison to other forms of web traffic (such as all of our financial transactions) that already traverses our networks. However, the diversity of applications and types of networked devices is exploding.
Q: Who makes overlay systems?
Fahrion: Any overlay network is any means to extract information from an existing process without touching the existing system. In some cases, applying an overlay system means installing additional sensing devices and communications, a wireless sensing mesh network, that can feed information to an application through a gateway.
Q: Do you see customers hesitant about locating control routines into the cloud versus at the edge?
Fahrion: Yes, process control will remain a local function for some time. But collecting and analyzing process data from one or multitudes of remote processes is an excellent application for "cloud" (public or private). To abuse a home-automation example, I see no real benefit of moving the control function of my thermostat to the cloud. But the ability to collect, analyze, and correlate environmental and energy consumption data to identify opportunities for efficiency gains, and perhaps feed forward-looking supply/demand information to the energy producers could be an excellent application.
Q: Since you mentioned industrial sensors and actuators, would wireless play a big role in industrial IoT in just monitoring, or how about control?
Fahrion: The near future is monitoring, providing information about resource consumption, machine condition, productivity, and such up to reporting, alarm/alert, and analytics applications. Next could come supervisory control, where a condition report is provided, and if no human response is provided, then some pre-scripted failsafe action taken. Industrial IoT applications today aren’t talking about real-time type control.
Q: You say there’s not a huge capital expense. There is still a big hurdle around convincing budget-holders to implement IoT solutions. Not really a question, more of a statement I guess.
Fahrion: Very true. In the industrial world the early adopters are those with the biggest potential savings. I’m seeing a pattern of early adopters in the area of energy efficiency and machine conditioning. Both of those, depending on the size/cost/productivity of the machines, have large "low-hanging fruit" opportunities for quick ROI with low risk. An industry segment that’s much further along is Telematics. That industry has had solid ROI (at least for commercial fleet vehicles) for a decade or more.
– Mike Fahrion is B&B Electronics director IoT product and strategy. Edited by Mark T. Hoske, content manager, CFE Media, Control Engineering, firstname.lastname@example.org.
View the archived webcast: Architecture of the Industrial IoT
– See other stories by Mike Fahrion about Ethernet, wireless, and Industrial IoT below.