Industrial Ethernet / Wireless

Reliable Device Networks Are Crucial for the Internet of Things

A badly designed network is often the weakest link in the system.

By Ian Verhappen

IoT's Weakest Link

A common conversation among many industrial networking specialists these days revolves around whether we should be distinguishing between wired and wireless networks. WINA, of which I am chairman, is one organization taking a closer look at this with a webinar on the topic to be presented in February.

Today's control systems typically include analog field devices (sensors and final control elements) that, in most cases, if not fully fieldbus-enabled, are capable of hybrid digital communications via HART. Such connections are commonly referred to as the field sensor network. The wired or wireless field sensor network typically connects to some form of I/O device that traditionally has been part of the control system as a dedicated interface device, but is now migrating to more generic forms including gateways and remote I/O or data concentrator units. From this level onward, communications tend to be based on Ethernet packets and internet protocol (IP) communications. Once the message has been converted to an IP packet, the options for moving it from one device to another are almost limitless, hence,all the discussion underway about the Internet of Things (IoT).

See Also: Missing Links: The Current State of Manufacturing Technology

As we have just shown, IoT depends on reliable, integrated, IP-based networks, and reliability is exactly what we as automation and controls engineers know well. Unfortunately, not everyone understands networks and network design as well as they should, and as a result, the network can often be the weakest link. My fellow columnist John Rezabek has demonstrated many times how the physical layer, including such simple things as grounding, can lead to loss of communications to a device, or at a minimum, an incorrect reading which, because it is often undetected, can be worse than receiving no signal at all.

Power is one of the biggest concerns end users have with wireless networks, and for this same reason full IoT is still going to be some time in the future.

I, too, have mentioned in previous columns how the physical layer, which for wired systems normally provides power and signal, is critical to and the basic foundation on which control systems are built.

Power is often one of the biggest concerns that end users have with wireless networks, and I believe that it is for this reason that Ethernet to field devices, and hence full IoT, is still going to be some time in the future.

Power over Ethernet (PoE) for industrial settings has two hurdles to overcome: the IEEE 802.3af that defines the output voltage between 40 and 57 Vdc while field devices expect to see 24 Vdc, and the 100-meter limitation for copper Ethernet cable between nodes. PoE does, however, have plenty of power, providing between 4 and 15 W, with work underway to possibly increase that further. Unfortunately for automation systems, Ohm's law and the thin wires in a CAT 5E or CAT 6 cable mean very high losses, which will reduce the maximum distance from the field device to the next node to less than the 100-meter Ethernet limitation. Of course, that field junction box will now be a PoE switch or a similar device to gather all the data from multiple IP-enabled devices, and the PoE switch will require a reliable power supply, both of which will have to be suitably electrically classified for the location in which they are being installed.

Despite the above, based on the number of articles published and interest from the consumer market, IoT appears to be the next big thing in the world of automation and will be a hot topic later in the month at the ARC Forum. I believe that IoT is presently starting up the slope of the Gartner hype curve, “Peak of Inflated Expectations.” (For more on the hype curve, refer to my column in Industrial Networking "Five Phases in the Adoption of a Technology"). IoT is being driven by the consumer market, and I am sure that just as with Ethernet technology, there will be crossover and pollination, but not to the extent being discussed in the press today.

These are some of the reasons that I believe that for the foreseeable future at least device networks will continue to be I/O or controller networks, though I could be wrong. One thing that will happen is the use of digital signals will continue to increase with combined wired and wireless networks as the mechanism for transporting all the bits between the controllers, enabling a more distributed control system than we have at present and likely one with higher reliability as a result.