CONTROL FREAKS
Control networks will regulate every
factory, house, and office.
By Deborah Claymon
Picture a factory in the industrial age: a foreman
supervises the entire operation from a glass booth high
above the toiling workers. Now imagine the factory of
the next millennium: the once centrally supervised
manufacturing system is replaced by a control
network--a distributed maze of intelligent devices that
communicate with each other and work together to
monitor and control all the mechanical processes on the
factory floor.
Computerized factories are not a new idea. For more
than 30 years, central control systems have been used to
link machine to machine and process to process
electronically. These mechanical administration systems
require thousands of feet of expensive wiring to connect
"dumb" components to a custom-programmed central
controller similar in function to the watchful foreman.
They used proprietary components and tended to break
down.
In contrast, each component in a "smart" control network
has enough intelligence to function independently as well
as interoperate easily with any other machine on the
network. Such decentralized networks increase
performance by monitoring processes where they take
place, and they are cheaper to create and service than
their dumb predecessors. They range in sophistication
from small networks embedded in a single machine to
large networks with thousands of nodes that monitor all
the functions in an entire building, transportation system,
or manufacturing operation. Eventually everything from a
subway system to a suburban home will probably have
some sort of control network.
Even so, the control network market represents a
fraction of the revenues of data networking, whose
biggest players--Cisco, 3Com, and Bay Networks--are
currently ignoring it. And while niche industry
associations have been working on communication
protocols for specific operations, only Echelon, a private
company based in Palo Alto, has developed a universal
open platform for control networks. No one knows
whether the networking powerhouses will disregard
control networks entirely and risk missing a vast potential
market. But control networks could be the next frontier
for companies struggling to find new product niches not
already dominated by Cisco.
Common sense
Control networks are still less pervasive than their
predecessors, control systems. In essence, control
systems are the unseen engines that run the world,
operating lighting, heating, security, and other process
management systems in factories, buildings, and homes.
These systems use series of sensors and actuators to
measure and regulate their environment. But only when
these components are linked together and given
distributed control and intelligence do they function as a
powerful network. For example, in a control system,
thermometers in a manufacturing process would simply
register the temperature of a machine and relay it to a
central control point, which would then determine the
appropriate reaction. A control network, however,
allows the components that generate heat to sense an
abnormal increase in temperature, then communicate the
problem directly to the building controls, which would
turn on fans, open vents, and otherwise adjust the
environment.
According to Echelon officials, most existing control
systems still use proprietary components. The industry is
reluctant to change, says Echelon CEO M. Kenneth
Oshman, because control systems companies earn 50 to
75 percent of their revenues from servicing their
proprietary technologies and have no incentive to move
their customers over to the new, more efficient networks.
Founded in 1988 by then-Apple Computer director
A.C. "Mike" Markkula, Echelon aimed to change all this
with its LonWorks platform: the first multivendor, open
network architecture--a commonplace, of course, for
computers and telecommunications--designed for control
systems. A LonWorks control network uses LAN
technologies to transmit small bits of data in a constant
stream, reporting on the status of components, like their
temperature and whether they are on or off. (Traditional
LANs are designed to handle large amounts of data in
single bursts.) Echelon's goal is to endow the most
archaic legacy systems in any environment with
communication and control capabilities.
Everybody's getting smart
Building and factory automation have so far earned
Echelon the majority of its revenues, which, Mr. Oshman
says, are substantial enough for the company to predict
that it will become profitable and go public midyear. The
building control market alone topped $7.3 billion in
1995, and, according to Terrence McMahon of the
research firm BCS Partners, Echelon is far and away the
leader in this market. Echelon also shipped more factory
automation applications than any other company in 1996
and 1997, according to the market research firm Venture
Development Group. And yet the market is still
incredibly young. Jim Taylor of Venture Development
projects that the number of nodes (network points)
installed per year will grow from 537,000 in 1997 to 2.7
million by 2002, on its way to an eventual market of
around 30 million new nodes per year.
Mr. Taylor warns that Echelon is not alone in the
industrial automation field. Both Allen-Bradley (a
subsidiary of Rockwell Automation) and Siemens
recently developed open control networks that are
catching up to Echelon's in number of node installations.
Mr. Taylor attributes some of their gains to overall
market growth but also cautions that Echelon has not
solved all the industry's problems with its one-size-fits-all
approach. Echelon will have to develop some
industry-specific applications if it hopes to fend off large
multinational vendors, like Siemens, that have
long-standing relationships with the manufacturing
community.
In addition to factories and offices, Mr. Oshman says
that the market for wired homes could materialize
quickly. Competition emerging from utilities deregulation
is motivating consumers and utilities to demand improved
communication and control. Patrick Hodges of Frost &
Sullivan, another research firm, agrees that in the next
five years utility companies will be under competitive
pressure to be more efficient and thus to rely on
distributed networks for tasks like reading meters. In
response, he expects consumers to begin demanding
home control networks capable of making energy
decisions that reflect fluctuating rates throughout the day.
But what kind of brain?
Despite the growing use of the Internet for all
communications, LonWorks control networks are not, at
the moment, based on IP. But as the Internet becomes
an important network for homes, many argue that
TCP/IP, rather than a dedicated control network, will be
the method to bring communications to every household
device.
Mr. Oshman disagrees, saying that using TCP/IP for
controls has major problems: as a protocol for large files,
TCP/IP is ill suited for the short, urgent messages that
dictate controls, and TCP/IP nodes are ten times more
expensive than LonWorks nodes. "I've talked to people
who want to put a browser in a thermostat," Mr.
Oshman says, "but a thermostat can handle only a very
small piece of software." Internet software is overkill for
most control applications, he argues, and the LonWorks
communication system is a more viable alternative. It
interacts seamlessly with TCP/IP, allowing LonWorks
data to be monitored from any remote location via the
Internet.
But while Echelon officials are positioning LonWorks as
the perfect complement to TCP/IP, the futurist Paul
Saffo warns that "there is a lot of momentum toward
making control networks IP-based, particularly when
IP6 [the next-generation Internet protocol currently in
development] comes online." He says Echelon's plans
could be upset by Internet developments in the same
way that the Web surprised online and interactive TV
developers earlier in this decade.
Does that mean that control networks will meet their
demise as soon as faster, better Internetworking
technology takes hold? Not really. Echelon will more
than likely integrate LonWorks with any emerging
protocol. The company has had the foresight to pioneer
a new frontier for networking technology. Since it is
highly doubtful that today's networking giants will
develop competitive control network technology any
time soon, Echelon stands alone at the beginning of a
phenomenon that will affect every automated aspect of
our homes, businesses, and factories. |
