Opinion: How should one build a wireless network?

By Control Engineering Staff January 3, 2007

Sven: So, say Ole, do you build a house from the ground up, or start with the roof and work down?

Ole: Why, you start at the ground and work up, of course!

Sven: Dang! Now I have to tear it down and start over.

This old joke illustrates the situation facing companies that are early adopters of today’s wireless process instrumentation and other accessories. Most are starting at the ground and creating point-to-point solutions or small mesh networks to test and validate technology for themselves. Most instrumentation providers are encouraging this activity to make users more comfortable as they consider adopting wireless in more applications.

Emerson Process Management offers starter kits with enough instrumentation devices and supporting equipment to set up and prove a small battery-powered mesh network. Its position is that this will help companies take first steps toward expanding to a more inclusive network. Emerson, Honeywell Process Solutions, Sensicast, Elpro, and others, have been very enthusiastic in their declarations that wireless works, its performance is proven, and it is available right here, right now. Even though standards issues aren’t completely resolved, there is no meaningful technical reason to drag your feet and hold back from adopting it.

That being the case, I was taken aback when I attended Invensys Process System ‘s user group in Dallas in early December. The message presented was low-key to the point of telling customers they shouldn’t get too carried away with wireless as yet, and that wide adoption is probably still at least three, perhaps five years away. Any presenter making such a suggestion at Emerson’s Global User’s Exchange would have been yanked off the stage. Invensys’ message isn’t so much that the technology isn’t ready, but that it shouldn’t be adopted willy-nilly.

Hesh Kagan, Invensys’ director of new ventures is spearheading the wireless product promotion and stated that wireless networks need management, much more than wired networks doing similar functions in the same plant. Consequently, Invensys is taking a top down approach, stressing that networks must be designed and managed from the outset to meet key business objectives. Ad hoc networks, using mixed vendor platforms, and launched by various departments will soon become hopelessly confused and unmanageable, wasting bandwidth. Critical information will get lost. Instrumentation will drop out. Companies that let that happen will incorrectly conclude that wireless just doesn’t work, wasting huge amounts of money in the process. Something that should be a key enabling technology for improved process control will be scrapped because it wasn’t adopted correctly. (Have you seen this happen in real life? If you’ve had an experience along these lines, I am interested in hearing about it. Email me at PWelander@cfemedia.com .)

Wireless solutions, according to Invensys, are largely ready to use, at least from a technology standpoint. The company’s offerings use hardware made by Dust Networks, with network infrastructure designed by Apprion, for integration with Invensys’ larger control systems. If a company can design a system sensibly (and Invensys will help in that critical process), there is no technological reason to wait. Still, it’s better to take it slowly and do the planning the right way, rather than doing something just for the sake of getting a wireless project moving. Decide first how you can best put wireless to work, plan and prioritize the most critical and time-sensitive needs, and then implement carefully following your internal roadmap.

Some might consider this approach excessively cautious, or like building a house from the roof down. On the other hand, planning your strategy carefully is good advice that applies to any vendor. Should it stop you from implementing small networks on an experimental basis? Probably not. Most companies are a long way from clogging their wireless bandwidth. Is Invensys being too conservative? Will they effectively take themselves out of the early adopter wave? Time will tell. Wireless implementation horror stories will certainly receive wide circulation through industry grapevines.

Of course, if you have the right plan, maybe you can build a house from either direction.

— Control Engineering Daily Daily News Desk Peter Welander , process industries editor