A world in which computers and networks disappear into the background of smart rooms and buildings is as viable a vision today as it was when the concept of ubiquitous computing was first articulated 15 years ago. But translating this vision to commercial reality has been slow, and much R&D work remains, say researchers in the field.
Mark Weiser, a researcher in the Computer Science Lab at Xerox Palo Alto Research Center, first put forward the notion of ubiquitous computing in 1988, as information technology's next wave after the mainframe and PC. In this new world, what Weiser called "calm technology" will reside around us, interacting with users in natural ways to anticipate their needs (see www.ubiq.com/hypertext/weiser/UbiHome.html).
The concept initially sparked work in mobile tablets and software agents. Today's efforts have morphed to pursue intelligent buildings packed with wireless sensor networks and displays, where information follows users wherever they go. "As computers proliferate, the kinds of relationships we have with them will be less based on input and output and more on sense and response," said Joe McCarthy, a senior researcher for Intel Corp.'s University of Washington lab in Seattle.
For example, while working at Accenture Technology Labs, McCarthy developed an application that would determine the optimal mix of background music to play in the company's exercise room, based on information from sensor badges detailing who was working out at the time.
McCarthy will chair the fifth annual Ubicomp conference Oct. 12-15 in Seattle. His goal for the conference is to get researchers to demonstrate as many working systems as possible. "We are nearing a stage where there are enough apps out there that we can go beyond Powerpoint presentations and actually experience [ubiquitous computing]," he said.
But other researchers said that "ubicomp," as the discipline is known, is not ready to emerge from the lab just yet. "There's still a lot of work in how to build the distributed computing systems needed for ubiquitous computing," said Steven Shafer, a senior researcher at Microsoft Research. "There are a lot of systems research issues in creating reconfigurable, stable and ubiquitous environments. And underlying sensor research [is still] needed."
But perhaps the biggest job is to develop standards or platforms for this emerging field, Shafer said. "One of the ironies of ubiquitous computing
is [that] almost no ubiquitous-computing systems work ubiquitously, or even in two places," he said. "Most demos for a smart room or building work only in that one place."
Microsoft's Shafer: Calls for standards, platforms.
Shafer hopes to address that issue with a universal location tool kit that could act as an underlying software interface for location data from any room, building, campus or city. The software plumbing could help tie services and applications from one environment to another as users roam from place to place, he believes.
"I want to whip out my PocketPC in a mall and say, 'Where is store X?' and 'Where is the nearest cash machine?' and get a map and directions. I could also take my mobile device to the grocery store and ask, 'Where are the raisins?' " said Shafer. The researcher hopes to complete a basic prototype system within a month.
Ubiquitous computing raises the debate between data security and data privacy to new levels with its ability to share user information across public and private systems. Beyond that, he said, researchers are still trying to define the commercial rationale for ubicomp.
"No one has discovered the killer apps that will make the smart room or building take off," he said. "It's a field of solutions searching for a problem, and the problems are very difficult to articulate."
On a smaller scale, IBM Corp. has gained traction with a less-ambitious ubicomp spin it calls "pervasive computing." The goal is to enhance and tie servers together in ways that begin to build on the ubicomp vision. "We want to deliver data across various devices and networks in the most effective way," said Todd Moore, program director for architecture and technology strategy in IBM's pervasive-computing group.
IBM's efforts have been based on versions of its Websphere Internet middleware and its ViaVoice voice recognition package. One Websphere variant helps corporate users roam between cellular, 802.11 and other wireless nets. Another, now in trials with set-top-box and residential-gateway makers, aims at helping link those systems to the Web and to home and business networks. Yet another version has real-time capabilities aimed at automation for automakers.
The pervasive group is working on standards for integrated voice and data. Such a standard would let users create a Web site that doubles as a telephone voice response system. "I can see the day where you will be able to pick up your cell phone and access your e-mail," said Moore.
Perhaps IBM's biggest step toward ubicomp was last fall's announcement that some 2003-model Honda Accords carry an embedded version of ViaVoice. Touching a "talk" button on the steering wheel, drivers use a 150-word English vocabulary to request a car temperature change or ask directions, which an on-board navigation system "speaks" via the car's audio.
It may not exactly be the future of ubiquitous computing, but it's pretty cool, say Honda executives.