RAFFI KRIKORIAN holds up a circuit board no bigger than a matchbook, with just enough space for a couple of chips, a few threads of wiring and a socket or two. “It costs maybe three dollars,” he tells me. It’s so cheap and simple, Krikorian’s team at the Massachusetts Institute of Technology’s new Center for Bits and Atoms expect their humble circuit not only to change the way we wire our homes, but transform how we live in them too.
The researchers plan to insert their device into everything from lights, switches and temperature sensors to door locks and shelves. They will then be hooked up to form a network that replaces the usual jumble of cables, switches and other gear you need to wire up and automate a building.
Is this just another over-hyped brick in the wall of the smart home? Certainly not, says Neil Gershenfeld, director of the new MIT centre and head of the project. He and his team are using their tiny circuit board to turn almost anything in a home into a Web server. Give each a unique Web address and these objects can then talk to each other using a simplified form of the standard communications language of the Internet. The result is a single, open-source, simple standard that Gershenfeld has named Internet “zero”.
Advertisement
He claims that the benefit Internet 0 brings will be far greater than the sum of its parts. You get all the advantages of a global Internet – simplicity, reliability and infinite flexibility – yet without the need for computers to control it. “Our idea isn’t to put computers in a building,” says Gershenfeld, “but in a very literal sense to let the building be the computer.”
Extend Internet-like connectivity to even the most rudimentary components of a building and you slash the costs of networking it. You can rearrange and reconnect Internet 0 at will – and you don’t need to know a hub from a handshake to do it. You can link lights, movement sensors, heaters and air conditioning, for example, so the system can tell when rooms are in use and adjust the temperature and lighting accordingly. In case of fire, the network can tell rescuers how many people are in a building and where they are. It can give you remote control of washing machines, cookers, curtains or video players from inside or outside the home. It can bring health monitoring to every room, even reminding you to take your medication if you forget. No matter how and where you want to interact with your space, Internet 0 can help.
If it is to catch on, Internet 0 will certainly need to be something special: there are already plenty of ways to raise the electronic IQ of your home. Take X10, created by a company in Kent, Washington, that also uses electricity cables to form the network. To automate an air conditioning unit, for example, you connect it to an X10 module, then plug the module into an electric socket. Plug a transceiver module into an electric socket elsewhere in the building and it can be used to turn the air conditioning on or off. Wireless technology, such as Bluetooth, simplifies things by enabling computers, phones and other devices to communicate by radio. In larger buildings, these networks need to be more sophisticated, so systems such as Echelon Corporation’s LonWorks use transceivers, control modules and specialist software to manage appliances like boilers and air conditioners.
These kinds of remote command and control systems are proving big business. Driven by the proliferation of computers and a growing desire to improve efficiency and cut costs, the worldwide market for home automation alone is predicted to reach more than $3 billion by 2006, according to New York-based market analysts Allied Business Intelligence. So it’s no surprise that competition for this market is keen.
What gives Internet 0 the edge? The major problem with most of its competitors is that they don’t scale up or down, says Gershenfeld. If you want to use LonWorks simply to turn a light on or off, you still need to invest thousands of dollars in equipment and software capable of handling air-conditioning systems. Meanwhile, systems designed for smaller networks such as X10 run into problems when you try to scale them up. Bandwidth restrictions on electricity cables, for example, mean you can use no more than 255 modules per network. Internet 0 is a reaction against these inefficiencies, says Gershenfeld. “It has learned from their mistakes and is a more mature successor to them. In fact, it could eliminate them.”
Two breakthroughs have enabled the researchers to swap the usual building full of specialised hardware for simple Net circuit boards. The first involved stripping down the Net’s communications system to its bare bones. The second simply slowed the speed at which data travels through the network.
Using a slimmed-down version of the Internet’s own communications system was the real leap. “The Internet has been around for 30 years and already has worked out a lot of its early problems and mistakes,” says Krikorian. It relies on an open protocol that everyone in the world uses. “We realised that we could take advantage of that.”
However, turning something as small as a wall plug into a Web server called for a tricky feat of software engineering called “de-layering”. The Internet’s communications system is built up of seven “layers”. The first or physical layer defines the electrical and mechanical specifications for the network’s hardware; the second defines how a computer accesses the network; the seventh provides the interface between software applications such as a Web browser and the network.
According to Gershenfeld, these layers are the software embodiment of human bureaucracy: each one has a sort of management committee to administer it and much of its code exists simply to pass messages on to the next layer. The only way to make this system small enough to fit onto a tiny chip was to strip away redundant or non-essential features from each layer, leaving only the code absolutely necessary to perform a specific task. If you were to fully implement each layer, you might need a megabyte of code. But Krikorian found enough waste and duplication for him to fit the essential functions of all seven layers on a 4-kilobyte memory chip. “Our chip doesn’t have an operating system,” he says. “It doesn’t need to communicate with printers or determine which version of software some other computer is running.” All it needs to do is send and receive simple commands that resemble Web addresses.
In addition to de-layering, turning a building into a hive of Internet sites means drastically slowing the speed at which data flows through the wires. At any junctions in a network, a data bit will create a small electronic reflection. This reflection can interfere with the data and garble the message. Conventional networks such as Ethernet use expensive pieces of hardware called “hubs” to eliminate reflections at corners and junctions. But simply sending the data through more slowly reduces the interference caused by reflections, so in Internet 0, the bit rate is slowed to less than a megabyte of data per second. “That means that we can now build a network the way an electrician wires a room,” says Gershenfeld. “Instead of putting in hubs and special equipment, we can just splice wires together.”
And Internet 0 doesn’t need a central computer to assign an Internet address to each electrical connection. When a device such as a light switch is turned on, its chip picks an address at random and checks that the address hasn’t already been assigned. If not, it will use it permanently. Add a computer to the system or link it to the global Internet so you can control appliances from anywhere in the world, and it will still work.
To prove the concept, the researchers have already built a prototype network using a variety of lights and switches containing Krikorian’s circuit, that connect to each other via a flexible track which serves as the electrical wiring in a house. The lighting track not only brings power to the appliances, it also carries signals between them.
To show me how it works, Krikorian reaches over to a length of track sitting on a table. Two light switches and a small lamp are plugged into it. He flicks each light switch in turn. Nothing. Then he touches a small metal strip on one of the light switches. Next he touches a similar strip on the lamp. When pressed, the strips make the devices broadcast a message along the track, saying “I’m ready to be connected”. Because the two broadcast one after the other, they exchanged network addresses and became “linked”. To prove it, he flicks the switch and the lamp lights up. Even when he unplugs the lamp and moves it elsewhere on the track, the switch still controls it. “Until you reset them, that switch will now control that light no matter where either of them is in the building,” says Krikorian. Swap the switch for a temperature sensor and the lamp for a heater or air-conditioning unit, for example, press their metal strip switches, and instantly you have a temperature control system.
This is possible because the network can do far more than send simple “on” and “off” signals. Krikorian plans to include a programming mode in his sensors and switches. Using this mode it will be possible to set up devices so that they send commands when triggered by a particular stimulus, such as a temperature change, the output from a clock or an infrared sensor.
Krikorian is also working on systems that communicate using high or low-frequency wireless signals. “If you need servers in your basement, a network installer to set up your system, and a programmer to debug it, you’re doomed,” says Gershenfeld. “The network infrastructure has to act just like a building: you put the stuff together and it works.”
Networks based on Internet protocols are the future of smart buildings, agrees Ken Sinclair, a building automation consultant and editor of automatedbuildings.com. Internet 0 could turn out to be a powerful system, he says. But he believes that if the project is to succeed, it will inevitably grow beyond the simple design that the MIT group envisions. Take something as simple as transmitting a temperature setting to a control device, he says. How many decimal places and how many bits of data will you use? Will it be Fahrenheit, Celsius, or just a raw electrical resistance? “They seem to think that all they need to do is give something an IP address. It’s more complicated than that. You need to have a lot of rules in the system and pretty soon their $3 de-layered chip turns into a $10 or $20 chip.”
Since Internet 0 can interface with the Internet outside, concerns about security also loom large. You don’t want hackers to turn off your alarm system in the middle of the night, says Doug Johnson, manager of standards strategy for Sun Microsystems. “In my judgment, Internet 0 won’t fully address this issue for some time.” Gershenfeld and Krikorian will have to include some interim solutions until there’s a better understanding of exactly what forms of security are needed, he suggests.
Gershenfeld has most of the answers. There are various firewalls and encryption techniques that can protect you from hackers, he says. But he admits that at the moment there are limits to the amount of information that Internet 0 can handle. It’s simple to encode something like a temperature-setting in a message, he says, but the system wouldn’t have the capacity to send documents between PCs, for example. Ultimately, he argues, the limits of Internet 0’s capabilities will be determined by the community that uses it.
If the people who automate buildings still have their doubts about Internet 0, the people who build them don’t. “When we started this work, we found that the construction industry was passionate about it,” says Gershenfeld.
Because the system combines the supply of power with networking, it could change the way that buildings are constructed. As much as 80 per cent of the cost of erecting a building is labour. Rather than having to hard-wire switches to particular sockets, it means that the entire building can be made programmable. That could often save the time and cost of hiring electricians whenever you want to swap light switches, for example, or hiring network specialists to install hubs and other dedicated network paraphernalia.
“Builders love this,” says Krikorian. Say a builder puts up an office block wired with Internet 0 components. A buyer looks over a unit and says, “I’ll take it but I want this wall taken out.” With Internet 0, the builder won’t always have to hire an electrician to rewire the entire place. He can just push a few buttons and re-route lights and sockets automatically. The potential economic impact is enormous. There should be savings in heating bills too. For example, the air supplied to a room can be adjusted locally, based not just on a thermostat setting but on data from motion sensors indicating the number of people in the room, from switches showing whether the windows are open and from the outside air temperature. The room’s air-conditioning unit can receive information directly from those devices and act appropriately.
Internet 0 will obviously be adopted by large commercial buildings first. And its first test should come soon, at MIT itself. The team hopes to have battle-hardened devices ready for production in time to try them out in the Media Lab’s new building, which is already under construction.
Gershenfeld’s group has worked with the architects to ensure that the new networking technology is an integral part of the plan. “That means that Internet 0 hardware will be in some form of commercial production in about a year,” Krikorian says. “Our goal has been to get this out of the lab and onto the market as fast as possible.” So far about a dozen companies, including Sun Microsystems, Panasonic and Intel, are sufficiently intrigued by Internet 0 to have pitched in to help Gershenfeld and his team think it through.
But Johnson warns that it will take at least another year to bring specifications for the devices to the point where manufacturers feel comfortable enough to mass-produce them. “It’s a long road from where Internet 0 stands right now until the time when you can go down to the electronics store and buy an Internet 0-ready device.”
Krikorian is already thinking about other things, including writing the software code that will enable cheap devices to go to work far beyond the walls of a building. He has talked with a researcher who wants to deploy a fleet of these devices in the oceans, where they will gather data on currents and water temperature, for example. Gershenfeld is working with herders in Lapland who want to use the cheap gadgets to track sheep and reindeer grazing in the mountains of northern Scandinavia. It could be useful in India, to help improve the efficiency of the electricity supply system. Krikorian’s students are even using Internet 0 to create an interactive bathroom shelf that reminds people to take the correct dose of their medications. “That’s what’s important about Internet 0,” Gershenfeld says. “It’s not just about making buildings intelligent. It’s about bringing the benefits of the Internet out into the physical world.”
