Maximum tolerable downloading time
If the user specifies a maximum tolera-
ble downloading time, the system will
negotiate the content version automatically
to meet this specification. In this case, the
content might require segmenting—for
example, cropping the large HTML text
and moving the residue to another page
linked by the “next” anchor. In Figure 7a,
the specified maximum tolerable down-
loading time decreases from left to right.
Color versus downloading time
Figure 7b demonstrates the effect of the
negotiation algorithm trying to strike a
balance between the conflicting factors of
color and downloading time. With de-
creasing weights on the user’s perception
of color, the system responds with content
versions of 256 colors, 16 colors, and 2
colors for the PDF document, to minimize
downloading time over a slow network.
We tested the system’s adaptability to dif-
ferent network characteristics by varying
bandwidth while keeping the other factors
constant. As Figure 7c shows, the system
switches modality to suit the connection’s
current bandwidth to keep downloading
time within the allowed tolerance.
To test the system’s ability to handle het-
erogeneous devices, we adjusted the device’s
memory buffer size to see whether the sys-
tem will automatically return the optimal
content version. The results were similar to
those for the network characteristics and
closely agreed with our expectations.
e plan to extend the nego-
tiation model to apply
transcoding to the tasks a
user wishes to perform on a
device, rather than just to the delivered
content. By “tasks,” we mean programs or
program functions. In the future, users will
be able to download program versions on
demand on the basis of their need and the
context. For example, an e-mail program
could have several versions: read only, read
and compose, or text only. Such function
adaptation is much more complex than
content adaptation. We also plan to extend
our prototype’s document model to include
images so that the quality domains could
also include histograms on color use, seg-
mentations, and so on.
In an ongoing project related to this one,
we study the trade-off between dynamic and
static adaptation.12To produce content ver-
sions, the decision engine can exploit static
preadaptation (sacrificing I/O performance
for CPU performance) or dynamic real-time
adaptation (sacrificing CPU performance
for I/O performance). A mixed approach
can yield a productive balance between
these two modes, leading to the most cost-
effective methodology for content synthe-
sis, with the decision engine’s guidance.
We are grateful to the reviewers for their useful
1.T.L. Pham et al., “Composite Devices Com-
puting Environment: A Framework for Sit-
uated Interaction Using Small Screen
Devices,” Personal and Ubiquitous Com-
puting, vol. 5, no. 1, 2001, pp. 25–28.
2.T.W. Bickmore and B.N. Schilit, “Digestor:
Device-Independent Access to the World
Wide Web,” Proc. 6th World Wide Web
Conf. (WWW 6), 1997, pp. 655–663;
3.S. Saha, M. Jamtgaard, and J. Villasenor,
“Bringing the Wireless Internet to Mobile
Devices,” Computer, vol. 34, no. 6, June
2001, pp. 54–58.
4.R. Comerford, “Handhelds Duke It Out for
the Internet,” IEEE Spectrum, vol. 37, no.
8, Aug. 2000, pp. 35–41.
5. C.E. Ortiz and E. Giguere, Mobile Informa-
tion Device Profile for Java 2 Micro Edition
(J2ME): Professional Developer’s Guide,
John Wiley & Sons, New York, 2001.
6.J.C. Mogul, “Server-Directed Transcod-
ing,” Computer Comm., vol. 24, no. 2, Feb.
2001, pp. 155–162.
7.A. Fox and E.A. Brewer, “Reducing WWW
Latency and Bandwidth Requirements by
Real-Time Distillation,” Proc. 5th Int’l
World Wide Web Conf.(WWW 5), 1996,
pp. 1445–1456; http://www5conf.inria.
8.R. Han, V. Perret, and M. Naghshineh,
“WebSplitter: A Unified XML Framework
for Multi-Device Collaborative Web Brows-
ing,” Proc. Computer Supported Cooper-
ative Work 2000 (CSCW 2000), ACM
Press, New York, 2000, pp. 221–230.
9.C.-S. Li, R. Mohan, and J.R. Smith, “Mul-
timedia Content Description in the Info-
Pyramid,” Proc. IEEE Int’l Conf. Acoustics
Speech and Signal Processing(ICASSP 98),
IEEE Press, Piscataway, N.J., 1998.
10.R. Mohan, J.R. Smith, and C.-S. Li,
“Adapting Multimedia Internet Content for
Universal Access,” IEEE Trans. Multime-
dia, vol. 1, no. 1, Mar. 1999, pp. 104–114.
11. J.-P. Richter and H. de Meer, “Towards For-
mal Semantics for QoS Support,” Proc. 17th
Ann. Joint Conf. IEEE Computer and Com-
munications Societies(INFOCOM 98), vol.
2, IEEE Press, Piscataway, N.J., pp. 472–479.
12.W.Y. Lum and F.C.M. Lau, “On Balancing
between Transcoding Overhead and Spa-
tial Consumption in Content Adaptation,”
to appear in Proc. Mobicom 2002, ACM
Press, New York, 2002.
Wai Yip Lumis an M.Phil.
candidate in the University of
Hong Kong’s Computer Sci-
ence and Information Sys-
tems Department. His re-
search interest is in the design
of content adaptation ser-
vices for mobile and pervasive
computing. He received his B.Eng. in computer
engineering from the University of Hong Kong.
Contact him at the Univ. of Hong Kong, Pokfu-
lam Rd., Hong Kong; firstname.lastname@example.org.
Francis C.M. Lauis the
head of the University of
Hong Kong’s Department of
Computer Science and
Information Systems. His
research interests are in par-
allel and distributed
computing, operating sys-
tems, and mobile computing. He received his
B.Sc. from Acadia University and his M.Math.
and PhD from the University of Waterloo, all in
computer science. He is a member of the IEEE
and served as a vice president of the IEEE Com-
puter Society in 1999. Contact him at the Dept.
of Computer Science and Information Systems,
Univ. of Hong Kong, Pokfulam Rd., Hong Kong;