As the Internet extends its reach, peo-ple want to do more and more withgadgets attached to it: check e-mailmessages from the airport, monitorhome security systems, and set roomtemperatures on the way home fromwork. The motivation is natural: youwant to keep in touch—and in con-trol—from wherever you are. Andwho hasn’t had a nagging feeling, afterleaving on a long trip, of having left aback door open or the stove on?
Systems that support these kinds offunctionality are starting to emerge. For instance, thefirst products using Bluetooth wireless technology arescheduled for market next year (http://www.blue-tooth.com). However, current prototypes and prod-uct visions seem quite ad hoc. Applications are essen-tially modeled in a disjointed manner that loses thenatural connections among them. User interfacestend to be superficial, and changing locations isabout the only subtlety underlying design ideas.
The common theme among these upcomingapplications is “action at a distance.” Let me proposea simple idea for modeling this theme computation-ally, so that where an action originates and where theaction occurs can be related systematically.
Spacing OutLet’s begin with some basic assumptions. We aregiven a world—a rich world populated with usersand devices and humming with activity. Let’s refer toa portion of the world, including its users anddevices, as a space. Important characteristics of aspace are the kinds of devices associated with it and
the quality of its network connection.Thanks to devices and connectivity,users can observe (and control) spacesfrom other spaces.
Let’s refer to the placing of spacesin the world as embedding. Each spacehas its own embedding, which revealssome interesting aspects of the space.What makes spaces interesting is thatthey can be based on more than justgeography. When the space is basedon a location, the embedding may tellus its geographical coordinates and
also facts pertaining to the location. For example, inthe home location, the facts would be about thesecurity system, temperature settings, and whether aback door is open. When the space is based on aninteraction—say, a dialog or e-mail exchange—theembedding gives us the user’s progress in it. Whenthe space is based on an ongoing activity, theembedding tells us things like the name anddescription of the activity, how far along the user isin it, and what the user plans to do.
What a user sees depends not only on the spacehe or she observes, but also on the space fromwhich he or she observes. Let’s refer to the view of aspace from a space as a representation. Essentially therepresenting space is a window into the world andfilters out some details of the viewed spaces. Forexample, using a home location embedding, youmight be able to call in to set your room tempera-ture, but not to disable the security system.
Spaces ComputationallyAn embedding of a space governs how much of itcan be observed or controlled from elsewhere.Embeddings are based on the devices that exist in aspace and how they are accessed. Think of embed-dings as computational capabilities. Likewise, a repre-sentation governs how much of an embedding canbe observed or controlled from the user’s currentspace. Think of representations as computationalauthorizations. Representations are based on thedevices that exist in a space and how their credentialsare set up. Pass the given portion of the worldthrough an embedding-representation path to com-pute what the user can see and control.
When you can represent a space which itself repre-
6 SEPTEMBER • OCTOBER 2000 ht tp://computer.org/internet/ IEEE INTERNET COMPUTING
From the Editor-in-Chief . . .
Beyond GeographyMunindar P. Singh • [email protected]
BEI
NG
INTE
RA
CTIV
E
An embedding of a spacegoverns how much of it
can be observed or controlledfrom elsewhere.
sents a space, you effectively tunnel through one space toget to another. From my workstation, I can view all labfiles and read my e-mail. From my home computer, Ican’t do either directly. But when I telnet to my worksta-tion, I can view all lab files and read my e-mail, effective-ly acquiring the authorizations of my office space.Tunneling helps us hack around the limitations of howthe authorizations and capabilities are set up. Tunnelingis a cheap way to achieve application integration. But it isnot always desirable. For example, in the near future, Imight leave my e-wallet on a server, change its settingsfrom my desktop (wider authorizations), and use it foronly small purchases from my phone (limited authoriza-tions). Of course, tunneling from the phone to the servervia the desktop would be a security risk, not to mentionconfusing to the user. Thus, tunneling may sometimesbe essential, but would often not be desirable. However,minimizing tunneling presupposes a clean applicationdesign that includes precisely the embeddings and repre-sentations that users need.
HabitationThe space metaphor provides a useful way to partitionthe design requirements of applications where users arefundamentally acting at a distance. Partitioning helps usbuild applications that keep track of where a user is. Andwhere a user is is not just a question of geography. �
7IEEE INTERNET COMPUTING
IEEE INTERNET COMPUTING
IEEE Computer Society Publications Office10662 Los Vaqueros Circle, PO Box 3014
Los Alamitos, CA 90720-1314
EDITOR-IN-CHIEF
Munindar P. Singh • [email protected]
ASSOCIATE EDITOR-IN-CHIEF
Robert Filman • [email protected]
EDITORIAL BOARD
Salah Aidarous • [email protected](IEEE Communications Society Liaison)
Miroslav Benda • [email protected]
Scott Bradner • [email protected]
Fred Douglis • [email protected](Liaison to IEEE CS Technical Committee on the Internet)
Stuart I. Feldman • [email protected]
Li Gong • [email protected]
Seif Haridi • [email protected]
Michael N. Huhns • [email protected]
Ray W. Johnson • [email protected]
Gail E. Kaiser • [email protected]
Leonard Kleinrock • [email protected]
Doug Lea • [email protected]
Frank Maurer • [email protected]
Charles E. Perkins • [email protected]
Charles J. Petrie • [email protected](EIC emeritus)
Agostino Poggi • [email protected]
William Regli • [email protected]
Helmuth Ritzer • [email protected]
Anthony Michael Rutkowski • [email protected]
Ravi Sandhu • [email protected]
Brian Thomas • [email protected]
STAFF
Managing Editor: Linda [email protected]
Associate Editor: Joan [email protected]
Staff Editor/IC Online Design: Steve [email protected]
Magazine Assistant: Hazel [email protected]
Art Director: Joseph Daigle
Cover Design: Joseph Daigle
Graphic Artist: Ken Duckworth
Contributing Editors: David Clark, Keri Schreiner, Martin Zacks
Publisher: Angela Burgess
Membership/Circulation Marketing Manager: Georgann Carter
Advertising Supervisor: Marian Anderson
CS MAGAZINE OPERATIONS COMMITTEE
Sorel Reisman (Chair), William Everett (Vice Chair),James H. Aylor, Jean Bacon, Wushow Chou,
George Cybenko, William I. Grosky, Steve McConnell, Daniel E. O’Leary, Ken Sakamura, Munindar P. Singh, James J. Thomas, Michael R. Williams, Yervant Zorian
CS PUBLICATIONS BOARD
Sallie Sheppard (Chair), Jake Aggarwal, Laxmi Bhuyan, Jon Butler, Lori Clarke, Alberto del Bimbo, Laurel Kaleda,
Rangachar Kasturi, Mike T. Liu, Sorrel Reisman, Mike Williams, Zhiwei Xu
computer.org/e-News
Available FREEto members.
Good news for your in-box.
Be alerted to• articles and special issues
• conference news• submission and
registrationdeadlines
• interactive forums
Sign Up Today forthe IEEEComputerSociety’se-News
Sign Up Today forthe IEEEComputerSociety’se-News
