With competing standards beginning to settle and technology becoming more reliable, wireless networking is making great waves in industry. But is it right for every environment?
Wireless networking is making major inroads into industry, but with it comes the inevitable issue of how to integrate such a 'game changing' technology with existing wired networks. End-users who are still undecided over whether to take the plunge into the wireless networks route are looking for assurance in four principal areas - standards, implementation, management and security.
On the standards side, matters have improved in recent months. Although there is as yet no unified standard, there is a clearer polarisation beginning to emerge.
Sivakumar Narayanaswamy, industry analyst for the Industrial Automation & Process Control group at consultancy Frost & Sullivan explains: 'Strong end-user demand is driving the implementation of wireless in the automation industry, but the progress made in adopting a unified standard is far from over.
'The ISA's ratification of ISA-100.11a in July 2009, and WirelessHart's approval by the IEC in March 2010 (as IEC 6259), goes to show that a consensus is yet to arrive. So we still have two major groups of supporters - ISA and WirelessHart - holding on to their positions to fight it out, in a broad parallel with the 'Fieldbus Wars' of the 1990s.'
He concedes though that, given the openness of the two standards, end-users don't need to worry too much anyway; standards on the wired side are now mature and stable to a large extent.
This point is echoed by Robert Jackson, senior product manager, Industrial and Embedded, at National Instruments. 'A lot of the technologies here are very mature - for example, standard Ethernet, Profibus, Devicenet, Flexnet and so on - and to build on that, wireless is pretty much a given too,' he says.
His colleague Nicholas Butler, wireless sensor products manager, adds, 'The standards landscape is still very fragmented, and will remain so until some sort of converging standard emerges, but as far as customers are concerned the products are there. The focus, though, should not be on standards but on the application in question.
'Each protocol has its own pros and cons, so what we would like to see is convergence on a protocol, particularly with respect to IP-based networks,' he says.
As to the benefits of using wireless for control applications, opinions vary. Narayanaswamy says, 'In applications where implementation and maintenance of the traditional wired network is difficult and expensive, the advantages of wireless can be realised in a big way. Also, better control in difficult applications in any industrial environment can be realised with wireless control networks.'
However, NI takes a different view. Butler says, 'Wireless is still okay for process applications, as its inherent [high] latencies are not a problem for the process industry. It's good for measurement but still not there yet for anything other than simple control - in fact I don't know if we'll ever have deterministic closed-loop wireless control.'
Emerson Process Management, which supports WirelessHart, is more specific. Mike Ferris, business manager at Emerson's Rosemount Division, says, 'The big driver for wireless is its ability to provide cost-effective solutions to measurement problems - that's the key issue.'
But he adds, 'Essentially you can use wireless for numerous applications in the process industry. Many of our systems have been used for process monitoring, but recently we have started to deploy wireless in control applications.'
So for wireless, at least at the moment, think process and measurement/monitoring, with some scope for control. WirelessHart is unsuitable for discrete manufacturing, which needs far lower latencies, according to Frost & Sullivan, while ISA 100.11a is not (as yet) specified for critical monitoring and control.
For suitable applications though, certain steps need to be taken before integration. 'The key component here is the software,' NI's Butler says, 'so first you need to understand how all the applications work in software. Compatibility is a key issue.'
Jeremy Fearn, wireless specialist at Emerson, agrees. 'Before integration, you need to clarify the capability of your existing host system and the available integration protocols, such as Modbus or OPC for example,' he says.
Next, as Butler points out, implementation of wireless devices is more than likely to be in a harsh physical environment and one that's electromagnetically cluttered, particularly in the 2.4GHz part of the spectrum. 'So you need to carry out a pre-installation site survey to determine what other wireless stuff is out there that you have to contend with, as well as determine which channels you can use, signal strengths and so on,' he says.
'Then there's the technical aspect - issues such as what to measure, at what frequency and how many channels to use. Then, and this is more at the back end, there's the issue of data management - do you send data to a database or to more of an enterprise-wide system? People these days want data access from anywhere via the Internet, so browser-based systems are becoming more important.
'In some respects these are the same questions as those you'd ask for in a wired system, but with wireless there's a new and extra set you have to ask as well. There are extra challenges to overcome with wireless - power supply, for example, when to change batteries and where to put the devices for adequate signal strength.'
Both Emerson and NI say this process is comparatively simple. Ferris says: 'As far as we're concerned it is very easy to implement and integrate wireless technology into legacy systems. There is no fundamental difference between integrating the data from a wireless measurement system and that from a wired measurement system.'
Jackson concurs, saying, 'For us it's all about providing wired and wireless options - real-world solutions. We don't see any discrete separation between wired and wireless, and there's no such thing these days as an application being wired or wireless. It's inevitable now that you'll have a solution that contains both.'
Post-integration management is also seen as straightforward. Narayanaswamy says, 'Although the network needs to be maintained and managed, there will be minimal changes after deployment. Network management is determined and devised by the process control engineers, and is usually done with a set of tools or applications and devices that they recommend.'
Fearn has a similar view, saying, 'The integration should be seamless so there are no real issues to tackle following the completion of integration.'
The final area to address is security, and any extra precautions that need to be taken when adding a wireless networks. While Narayanaswamy has warnings in this respect, Emerson and NI are quick to forestall concerns from prospective end-users.
If not addressed, says Narayanaswamy, security issues will naturally lead to information leaks that could pose grave risks to the plant. 'Remote access, the use of laptops or PDAs and creating a clone of a wireless access point are reasons that threaten security here. But these can be addressed by using a VPN or installing a firewall or intrusion detection system. Segmenting the network into zones or cells also helps to prevent unauthorised access,' he adds.
Emerson's Ferris, however, says, 'This isn't an issue. WirelessHart has a multi-layered approach to security with military-standard encryption, authentication, verification and anti-jamming. The security is always 'on' by default with WirelessHart, you cannot simply turn it off as with some wireless systems.'
NI views wireless networking in terms of the underlying IEEE standards, 802.11 and 805.15.4, and from this perspective Butler says, 'Authentication and encryption in the standards are acceptable for most customers - say, 95 per cent. It's not a concern, and we've even had customers who don't care about security, as they're only using wireless to send non-critical data.'
Jackson adds: 'Wireless is not inherently insecure - in fact you could argue that it is more secure than wired networks because so much money has been invested in this aspect of the technology.'
According to Fearn, there's a rapidly growing number of end-users for whom none of this will come as a surprise. 'Basically, the technology is becoming mainstream. Adoption is snowballing,' he says.
Assuming he's right, does that mean wired networks will be completely replaced by wireless? According to Butler, 'It might happen one day, but then this question reminds me of a quote: 'What is wired one day becomes wireless the next - and what is wireless becomes wired'. As an illustration of this, just look at how telephones have gone from wired landlines to wireless cellphones, and how a lot of TV systems have gone from using an aerial to being fed by cable.
'On balance though, I think there'll always be a place for wired systems.'
Wireless monitoring and measurement applications are not confined to the process industry of course. For example,'UK researchers at the University of Sheffield have used wireless technology to monitor the structural health of the Tamar Bridge, which carries road traffic between Devon and Cornwall in south west England.
The bridge had already been monitored to measure its dynamic and quasistatic behaviour, but the Sheffield team also wanted to measure the bridge deck's longitudinal displacement at the expansion joint. Analogue sensors known as extensometers were used to collect this data, which needed to be stored on a Web-accessible PC to allow it to be remotely queried from Sheffield. The distance between the expansion joint and the PC was 450m, making a wired solution too expensive and time-consuming.
The team opted for the Wireless Sensor Network (WSN) platform from National Instruments. A measurement node digitised the analogue inputs from the three sensors and transmitted them using IEEE 802.15.4 radio at 2.4GHz to a gateway node connected to the PC via Ethernet.
Given the project's 450m transmission distance - greater than the 300m range of a single NI WSN node - the team used an intermediate measurement node as a router, configured in software, after a series of tests using different antennae.
The team then developed an interface using NI's LabView to acquire and store the received data on the PC. The interface, which was accessible from the team's Sheffield base, displayed the readings from the three sensors. It also monitored the quality of the link between the two communication hops in the wireless network and the power supply to the measurement and router nodes.
Data from the extensometers was sampled every five seconds, and was then stored in a binary file on location every 24 hours and downloaded to the team's database when required. The data was then processed in Sheffield to investigate how the bridge deck's longitudinal movement correlated with environmental changes and different traffic patterns, helping the researchers gain a better understanding of the structure's performance.