A lack of standards is holding back wireless automation.
Late last autumn, automation end-users in the US were asked in an Internet poll which new technologies they think will have the greatest impact for their businesses in the next few years. Far and away the leading contender cited by about 20 per cent of respondents was the adoption of wireless, with cable-free I/O, Ethernet, control, networking and Fieldbus among the specific examples.
This is an explicit indication of industry's growing appetite for wireless, which is increasingly being seen as a strategic enabling technology: doing away with wires offers greater workforce mobility, better network flexibility and access to remote or difficult locations. There are also potentially huge savings in cabling costs a key issue for the process industry, for example, which has been one of the earliest adopters of wireless technology.
On the face of it, wireless offers a win-win solution, yet the widespread take-up of the technology is being held back by a number of factors, perhaps the most pressing of which is the paucity of common standards.
According to a report last year by consultancy Frost & Sullivan, 'The future trend of the adoption of wireless industrial automation solutions will depend on their seamless integration into the existing plant infrastructure and other control systems. The key assumption for growth is that wireless devices from multiple suppliers need to be compatible and a uniform standard applied wherever possible. This is what end-users are clearly looking for in their solutions.'
WirelessHart and ISA100.11a
So far, only the process industry has anything approaching an accepted standard, although even here different vendor groups are backing one of two contenders WirelessHart and ISA100.11a neither of which yet has worldwide approval from the International Electrotechnical Commission (IEC).
As its name indicates, WirelessHart is the wireless extension to the Hart protocol, an early implementation of Fieldbus which now has a worldwide installed base estimated at 25-30 million field devices, putting WirelessHart in a strong position as ade factostandard for the process industry. Developed as a multi-vendor, interoperable standard, it's been on the market since 2007 and is backed by the HARTE Communication Foundation, a consortium of companies including ABB, Emerson and Siemens Emerson was the first to ship WirelessHart-enabled devices in 2008.
By contrast, ISA100.11a still has some way to go. It's the first sector-specific standard in the ISA100 family, the rest of which are at the nascent stage, and is backed by companies including ExxonMobil, Honeywell, Rockwell and Yokogawa.
It too is an 'open' standard, but wasn't released by the International Society of Automation (ISA) until last autumn. It's broader in scope than WirelessHart, however, offering plant-wide and multi-protocol support, as well as the facility for video feeds, for example, and a platform for integrating control rooms.
In championing their respective causes, Emerson and Honeywell are perhaps the staunchest supporters of WirelessHart and ISA100.11a respectively. Putting the WirelessHart case, vice president of wireless for the Rosemount Measurement Division of Emerson Process Management, Bob Karschnia, says, 'WirelessHart has already reached a critical installed base, with more than 1,000 end-user installations, and has become ade facto standard in the [process] industry. The IEC-approved standard will be IEC62591, which is based on WirelessHart, and is expected to be finalised early this spring.
'ISA100.11a is still in its infancy, however. It must first be approved by ANSI [the American National Standards Institute] before it can begin any work with the IEC, and ANSI has noted some inconsistencies in the way the ISA developed its documents, which must be resolved prior to the work beginning with the IEC. After resolution of all the ANSI-noted inconsistencies, it will be at least 18 months before ISA100.11a is approved by the IEC.
'So in the meantime, there is only one IEC wireless standard, and it is 62591 (WirelessHart),' he says.
Honeywell's argument is that WirelessHart supports only Hart devices, whereas ISA100.11a supports a range of current and legacy protocols, helping to protect existing investment. The standard also allows more than 1,000 devices in a network compared with about 250 with WirelessHart.
Go for IEEE 805.15.4?
Eventually though, much of this dispute will make little difference to the end-user as these are both open standards, designed to accommodate products and technologies from different vendors. And Karschnia, for example, concedes that, although WirelessHart is not yet part of the ISA100 family, a fresh specification, ISA100.12, is working towards that end.
But for now, which path should you choose? Karthikeyan Balasubramaniyam, senior research analyst and team leader in the industrial automation and process control group at consultancy Frost & Sullivan, says, 'Deciding on which to adopt is usually a function of the user's existing installed base and who its current suppliers are.
'One important factor is why a user wants wireless. If it's to replace cabling, then they're more likely go for WirelessHart, but if it's to add the facility for some new applications, then it's probably best to wait.'
National Instruments, by contrast, takes a more fundamental view on standards here. 'Wireless today is all about two key IEEE standards 802.11 and 805.15.4. It's here where the more important work is being done, so you need to look here [at the Physical layer of the OSI model] rather than the higher levels,' says Robert Jackson, NI's senior product manager, wireless.
IEEE 802.11 (or Wi-Fi) is the set of standards for wireless LANs many home wireless broadband systems are now based on it while 805.15.4 is the basis for a wide range of protocols including ZigBee (perhaps best known at the moment for its application in home automation) which are aimed at delivering low-power and long-distance wireless. 'Both WirelessHart and ISA100.11a are based on 805.15.4, so it's premature to argue over ade factostandard before considering the underlying IEEE standard it's better to comply with 805.15.4 than WirelessHart or 100.11a,' Jackson says.
But IEEE 805.15.4 is not yet a silver bullet for industrial automation. Increasingly, industry is connecting its control and automation infrastructures to enterprise-wide IT systems and, beyond that, the Internet. And as Jackson points out, 805.15.4 doesn't support IP addressing yet, so the IP for Smart Objects (IPSO) Alliance was formed in 2008 by vendors including NI, Cisco, Ericsson, Fujitsu and Intel, to promote the use of IP on such networks. In this respect it supports the work of organisations including the Internet Engineering Task Force in developing standards.
These are clearly still early days for wireless. Despite the process industry's early lead here, the technology is still confined to monitoring and measurement applications in wireless sensor networks (WSNs) ISA100.11a for example is specified for 'non-critical monitoring and control applications'.
On top of that, other standards are emerging. 'There are at least two other standards currently being developed beyond WirelessHart and 100.11a an industrial version of ZigBee and a Chinese standard,' says Karschnia.
'If you consider what a standard is, it is an insurance policy for a customer that they can be sure of a broad supply base no one vendor makes everything and a sustainable base of suppliers that will ensure continuity over time. There is an old saying: 'The great thing about standards is that there are so many to choose from'. The point though is that while there will always be multiple standards to choose from, one will typically be the 'gorilla',' he says.
With so much dispute already raging over a standard for the process industry, engineers in the discrete manufacturing sector which is far behind in the standards-setting process, with bodies like the ISA and the EU's ETSI having only recently set up working groups could be forgiven for looking on in dismay.
At the moment the ISA is focusing on two areas. As with the process industry, the first is to cut maintenance costs by using wireless in place of cables, although the applications here include robotic arms, where the constant flexing through their axes of movement causes the cabling to break eventually.
The second area is to address the specific need for low wireless latencies in factory machine applications. WSNs in the process industry can often tolerate latencies in the 100-500ms range, but discrete manufacturers need them to be far lower, 10ms widely being regarded as the minimum acceptable.
While WirelessHart will not figure in this sector 'it's no good for discrete', says Balasubramaniyam ' ISA100.11a could be revamped to embrace the latency needs of discrete manufacturing. That said, however, IEEE 802.11 and existing industrial Ethernet protocols such as EtherNet/IP also hold promise. In fact Jackson, for one, thinks 802.11 will dominate.
But setting a wireless standard or standards for discrete manufacturing should go more smoothly than in the process industry, according to Balasubramaniyam. 'Given the disputes over 100.11a, to some extent the same will happen with future parts of the ISA100 standard,' he says. 'But once the process industry standard is out, people will have learned from initial mistakes with it, and, as the broader standard becomes more complete, some parts of it will become common across the sectors.'
While he says the lack of a common standard is a major factor holding back some end-users from adopting wireless, Balasubramaniyam also says it isn't the only factor. 'A key challenge remains the technology's operational reliability, its ability to work in harsh environments this is the biggest concern for end-users. Things like mesh networks and their 'self-healing' capacity [on which these standards rely] help a lot but users are still resistant to the technology and remain to be convinced.'
Jackson agrees that it isn't just the fault of lack of standards. 'Users need three things,' he says. 'They need to define the broad product information, by sourcing products from established vendors; they need education in the use of wireless; and they need to see how the products fit the application. This last point is where standards come in, and the standards are always changing, so users need a modular approach to the architecture of their infrastructure to account for that.'
Balasubramaniyam says some of this is already happening, as vendors try to work closer with end-users by setting up pilot systems for non-critical applications.
Security is another issue, he says one where wireless exacerbates the problem of hacking. But Jackson is less worried. 'Security is a major concern with wireless but then major vendors such as Intel and Cisco have invested more money in Wi-Fi security than anything else. The best security ever has been driven by the Internet,' he says.
In Balasubramaniyam's view, ironically, both these concerns come back to a matter of standards. 'To an extent, the issues of wireless reliability and security would be taken care of if we had a unified standard,' he says.
So when can industry expect a unified wireless automation standard what Jackson calls 'the million-dollar question'? Understandably, opinions vary.
'One of biggest factors will be vendors making a consolidated effort to agree on the standard which is a nice idea in theory,' says Balasubramaniyam. 'When will we have a single standard? I would love to say in three years or so from now but I won't be surprised if it takes longer.'
Jackson won't be drawn on a timetable but he does say, 'Overall, the process will happen in the same way as it did with the adoption of technology such as Ethernet, which came into industry from other sectors and had industry-specific layers added on top of it. I also think things will go the IP way once IP is on top of 805.15.4, then a lot of protocols will fit on top of that.'
In the meantime, for users who cannot or don't want to wait for a unified standard, it looks to be a case of 'You pay your money, and takes your choice'.