One of the things I hate about meetings is the information transfer rate is usually pretty low. In other words, lots of talk, little data. The CSIA meeting has historically had an impressive information transfer rate, and this morning's sessions are no different.
"Bill and Bill Electronics"
Gerry Niemi of B&B Electronics led off after a small commercial in which he admitted that the Bs in his company name were really "Bill." "We're a small company that used to be in a cornfield 90 miles from Chicago. Now we're a small company in a Chicago suburb. We didn't move, the corn fields did."
Gerry proceeded to give a one hour "Mesh Networking for Dummies" presentation that was as detailed and authoritative as any I've seen.
Industrial wiring costs (according to Frost and Sullivan) range from $10/ft for pipeline monitoring to $1000/ft for nuclear plants. Most process plants are in the $12/ft range, with discrete automation plants in the $90/ft range. It so happens that I disagree with Frost and Sullivan's numbers. I believe them to be on the order of one order of magnitude too low. It costs lots of money to wire up one single sensor, and costs more the farther that sensor is from the control room.
Wireless, as can be seen from the April CONTROL poll, is growing in acceptance, and Gerry presented numbers indicating growth of greater than 34% per year.
But there are issues. The biggest holding back the spread of wireless in industrial process applications are reliability and security.
There are three basic kinds of wireless networks, and most of them trade distance for bandwidth and redundancy. Wi-MAX can go thousands of feet, WiFi can go hundreds of feet, and the mesh networks and Bluetooth can go only tens of feet.
In reality, mesh networking is about sensor networking. Mesh networks have existed for years. The wired telephone switchnet is a mesh network. So, too, is the cellular net, and, above all, the Internet is itself a mesh network. What the Zigbee protocols (IEEE Standard 802.15.4) do is to begin to adapt the theory of mesh networking to wireless communications.
It is hard to talk to a technical audience with mixed knowledge states. Sometimes you start so far back at the beginning that you lose people. Some people get impatient and want you to jump ahead. Some people go to sleep until you say something interesting, and some people are grateful for the underpinnings that "everybody should know already."
Basically, the drive to mesh networking, according to Gerry (and I agree with him completely) is due to the limitations of other wireless protocols, specifically 802.11x. He believes that eventually, 802.11g (and whatever succeeds it) will be used for decision/data networks like the MES and ERP system, and for standard enterprise networking. He believes that mesh networking will replace wires on the plant floor.
WiFi is of limited use on the plant floor because it ain't as easy as it looks. Interference, blockage, RF noise, and other issues cause unexpected (if you didn't do a good old fashioned site survey) lapses of communication, IP address conflicts, and security concerns.
Mesh networks were designed to deal with this issue.
Gerry tells the story of a plant manager who explains that it takes him nearly 30 minutes to find the closest operator, and get that operator to a problem area to correct the problem. What he wants, Gerry notes, is to be able to bring the information to wherever the relevant person is at the time, rather than trying to bring the person to the information.
In mesh networks, every node is a router, and every node is a master. Thus there are no single points of failure. Does this mean that eventually even SIS systems may go wireless? I suspect it does.
Gerry believes that mesh networking is a disruptive technology. It is designed to produce a wired world, and a wired process plant. As the cost per node reduces (it is at about $250 now) it will be possible to wire more sensors than ever before, to do things like optimize the HVAC system in the plant (one early adopter is controlling the well-being of his cows in Arizona using a mesh network), to actually do real predictive maintenance on motors and drive trains throughout the plant, and to do realtime process control and indication on the flattest data hierarchy possible.
The military believes that mesh networking is the key to the battlefield of the future, with disposable mesh-enabled sensors being scattered over the entire battlefield to produce an electronic picture of the battle unavailable by any means now. This is intended to reduce to nothing the "fog of war."
Mesh networks are self-configuring, self-healing, and self-reconfiguring. If a sensor cannot reach one node, it can reach another, and information can be delivered through short multihops rather than increased power. Very low power consumption is at the heart of the Zigbee standard.
There are downsides, of course, to mesh network technology as it exists. Gerry notes that the technology is immature, the standards are new and sketchy, and because the standard is designed for commercial use rather than industrial, there is no interoperability between companies because of the quasi-proprietary add-ins necessary to make systems industrial grade. And it isn't cheap enough yet.
So, where does the system integrator fit? The integrator supplies the interface for the owner/operator to be able to use the technology. The integrator will provide the knowledge base to be able to site sensors, nodes, gateways, and make sure that all this talks to the control system and the enterprise, just like the integrator does now, with wired systems.
RFID: Hype or Revolution?
Certainly, as I've talked about before, the congruence of mesh networking and RFID on the plant floor is causing many of us to rethink how process plants are run.
Siemens' Alex Steubler , Business Manager for Factory Automation Sensors, based in Norcross, Ga., says that RFID is the "oldest emerging technology," since companies like Siemens have been doing it for 25 years already.
In fact he's right, since the first time I ran into RFID technology was about 15 years ago when I had my dog chipped for identification.
RFID is one of the three hottest mainstream and techie media topics, along with WiFi and VoIP telephony.
Part of this is the Wal-Mart effect, as Wal-Mart mandated first its 100 biggest suppliers, and now all suppliers to conform and RFID mark case lots and manage their consigned inventory for Wal-Mart based on RFID-enabled inventory.
Part of this is the DOD"¦war theater wastage is a horrendous burden, and by 2004, the DOD had managed over 90% RFID tagging with corresponding increases in efficiency in getting supplies to the battlefield, and reductions in waste and surplusage.
And now the FDA is getting into the act, mandating no drug untagged by 2007.
What will a RFID enabled world give us?
First, increased efficiency. I've talked before about the belief that RFID will transform manufacturing and warehousing, in fact the entire supply chain.
Second, increased security, as everything is tagged, even us.
This will create a databased world, for manufacturing, distribution and retailing.
It is also going to create a marketing world in which we really do create "an army of one"¦"
What privacy? Do you think you have privacy now?
Comments?
--Walt Boyes
