An Infrastructure Approach to
Context-Aware Computing
Jason I. Hong and James A. Landay
University of California at Berkeley
ABSTRACT
The Context Toolkit (Dey, Abowd, and Salber, 2001 [this special issue]) is
only one of many possible architectures for supporting context-aware applica-
tions. In this essay, we look at the tradeoffs involved with a service infrastruc-
ture approach to context-aware computing. We describe the advantages that a
service infrastructure for context awareness has over other approaches, outline
some of the core technical challenges that must be addressed before such an in-
frastructure can be built, and point out promising research directions for over-
coming these challenges.
1. INTRODUCTION
People have always used the context of the situation to get things done. We
use our understanding of current circumstances to structure activities, navi-
HUMAN-COMPUTER INTERACTION, 2001, Volume 16, pp. 287–303
Copyright © 2001, Lawrence Erlbaum Associates, Inc.
Jason Hong is a computer scientist with an interest in context-aware computing
and multimodal interfaces; he is a PhD candidate in the Group for User Inter-
face Research at University of California at Berkeley. James Landay is a com-
puter scientist with interests in informal user interfaces, multimodal interaction,
and design tools; he is an Assistant Professor in the Computer Science Division
at University of California at Berkeley and co-leader of the Group for User In-
terface Research.

gate the world around us, organize information, and adapt to conditions. For
example, when we’re holding a conversation in a noisy place, we talk louder
so that the other person can hear. But when we’re in a meeting, we whisper so
as not to disturb other people.
Context awareness has also been an integral part of computing. Even sim-
ple forms of context, such as time and identity, have been used in a number of
meaningful ways. For example, by being aware of the current time, computers
can give us reminders of calendar events. By being aware of our identity
through logins, computers can personalize the look and feel of our user inter-
face. Different kinds of context can also be used together. For example, com-
puters can tag files with both time and identity, giving us many ways of
organizing and finding information created in the past.
Strides in miniaturization, wireless networking, and sensor technologies are
enabling computers to be used in more places and to have a greater awareness
of the dynamic world they are a part of. In fact, over the past few years, a new
class of context-aware applications that make use of these technologies has
been developed, showing how computers can leverage even elementary no-
tions of location, identity, proximity, and activity to great effect (e.g., Active
Badges, Want, Hopper, Falcao, & Gibbons, 1992; ParcTabs, Want et al., 1995;
and Cyberguide, Abowd et al., 1997). The key to these context-aware applica-
288 HONG AND LANDAY
CONTENTS
1. INTRODUCTION
2. LIBRARIES, FRAMEWORKS, TOOLKITS, AND
INFRASTRUCTURES
3. ADVANTAGES TO AN INFRASTRUCTURE APPROACH
3.1. Independence From Hardware, Operating System, and Programming
Language
3.2. Improved Capabilities for Maintenance and Evolution
3.3. Sharing of Sensors, Processing Power, Data, and Services
4. CHALLENGES TO BUILDING A CONTEXT-AWARE
INFRASTRUCTURE
4.1. Defining Standard Data Formats and Protocols
4.2. Building Basic Infrastructure Services
Automatic Path Creation
Proximity-Based Discovery
4.3. Apportioning Responsibilities
4.4. Scoping and Access of Sensor and Context Data
4.5. Scaling Up the Infrastructure
5. SUMMARY