Slow and Stupid Networks Often Win Race

By Justin Pope  |  Posted 2003-09-08 Print this article Print

"Faster and smarter" is the mantra in the wireless world, but Robert Poor has a different vision: He dreams of networks that are slow and stupid.

BOSTON (AP)—"Faster and smarter" is the mantra in the wireless world, but Robert Poor has a different vision: He dreams of networks that are slow and stupid. For years, wireless has been all about developing standards like Bluetooth and the still-more-powerful 802.11 that can hurl data through the air rapidly enough to fuel complex applications like streaming video on laptops or cell phones.
But not everybody needs streaming video. Sometimes slow and steady—but light and cheap—packs more punch.
Imagine a municipal light department trying to track broken street lamps. It would need wireless outposts—known as nodes—in thousands of lamps, so the nodes must be inexpensive. Their only job would be to occasionally relay a simple piece of information: whether a light is on, or off. "A street lamp doesnt have a lot to say," said Poor, chief technology officer at Ember, a Boston start-up developing low-power wireless networks. "They dont have to talk very fast. They dont have to talk very often." Techno-visionaries have a long list of plans for low-power, low-speed networks, from thermostats on factory floors and in residential buildings to tiny battlefield sensors for the Pentagon and more efficient irrigation systems. The idea is to build motes—tiny computers that broadcast a radio signal—that are cheap enough to deploy everywhere but just smart enough to "self-organize" into powerful networks that can sense and convey information like whether milk is spoiled or a bookshelf overloaded. Dozens of companies are working on it, but there could be an increase by the end of the year with the adoption of a standard called Zigbee, being negotiated by a consortium of companies that includes Ember and bigger players like Honeywell, Motorola, Philips and Samsung. The standard gets its name from an idea engineers are hoping to emulate: the zigging and zagging of bees, which are individually simple organisms that work together to tackle complex tasks. Consortium members have agreed on initial portions of the standard, which governs how data is broadcast from a wireless node, and products that incorporate it are starting to trickle out. But the real advances wont come until the next portion is endorsed. Thats expected some time this year. On Zigbee, data crawls along as slow as 20 kilobits per second, about one-fiftieth the speed of Bluetooth. And the 802.11 family of wireless standards, better known as Wi-Fi, can be many times faster than Bluetooth. But slow and steady has its virtues. With more complex networks, each "node" on the network has to be within range of a wired, expensive hub. With "mesh" networks like Ember, both the messages and the components are simple enough that each node can act as a relay station. In theory, the price would allow each node to be placed within broadcast range (about 100 feet) of another. A message could simply jump from lamp post to lamp post, all the way to headquarters, with no need for expensive hubs or wires. Ember already can make its own nodes talk to each other, and other companies can do the same. The emergence of Zigbee as a standard would let nodes made by different companies talk to each other, expanding their potential exponentially. Its the premise of "Metcalfes Law," which states that the key to a powerful network isnt its sophistication, but its number of users. In fact, the laws namesake, Robert Metcalfe, inventor of the Ethernet networking standard, is one of Embers backers. "Sometimes you want to go far, sometimes you want to go fast. Sometimes you want to have low power, sometimes you have all the power in the world," he said. "The 747 does not obsolete the bicycle." The Wireless Data Research Group has estimated the market for low-power, low-speed data networks will hit $8 billion by 2007, though its likely to start in industrial areas like factory automation and petrochemical processing. But Zigbees impact remains to be seen, said Ian McPherson, an analyst with Wireless Data Research. Some companies may not want to make their products talk to others, or they may decide they can better solve clients problems with their own standards. Ericsson, for instance, has expressed interest in a different standard some call "Bluetooth light." "Zigbees not the beginning and the end of the market," McPherson said. The consumer side may be even shakier. Many of the most obvious uses are in home automation, a market that has yet to take off. "My take is the actual market for this kind of home automation application is not even on the horizon," said Charles Golvin, an analyst with Forrester. "Sure, Bill Gates house can turn on the lights when people walk in and out of the room. But that kind of reach and demand for average users and average homes is way off in the distance." Still, Kristen Law, the Zigbee point person at Motorola, said the company plans to integrate Zigbee into a full range of products, from cable boxes to cell phones and handheld computers. Its already producing samples of semiconductors it will sell to other manufacturers, and expects to begin selling its own devices within a year or two. "Its very possible that by the end of 2004 you could go to Home Depot and buy a Zigbee light kit," she said. For companies involved in Zigbee, the economics are tricky. Law says some have overhyped how inexpensive it will be off the bat, and Ember has found the market among the industrial customers its targeting unexpectedly complicated. One company wants to monitor chemical flow five times a second, another wants to read a gas meter once a day for 10 years. Ember decided to stop selling complete systems and now sells chips and software licenses, then helps customers build the technology into their own systems. But Poor, a musician who went back to school at age 42 to pursue a Ph.D at the Massachusetts Institute of Technologys Media Lab, still envisions a future in which tiny wireless networks go about their business unnoticed, cheap and puny in their individual components but powerful when working together. "I cant tell you any more than I could in 1985 what the intelligence will be," he said. "In a sense, I hope there wont be a killer app. There will just be lots of things quietly doing their thing."

Submit a Comment

Loading Comments...
Manage your Newsletters: Login   Register My Newsletters

Rocket Fuel