Computer Science Technical Report Archive

USC Computer Science Technical Reports, no. 677 (1998)

(USC DC Other)

USC Computer Science Technical Reports, no. 677 (1998)

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Configuration Challenges for Smart Spaces
John Heidemann, Ramesh Govindan, Deborah Estrin
University of Southern California/Information Sciences Institute
USC Technical Report 98-677
July 15, 1998
ABSTRACT
Smart spaces will be composed of thousands of computing elements
interacting with the physical world. We argue that automatic config-
uration and resource discovery remains an important unsolved prob-
lem in smart-space deployment.
1. INTRODUCTION
Trends in CPU performance and packaging now allow very power-
ful devices to be embedded on single chips. Today a car may have a
half-dozen processors on-board that are monitoring brakes, improv-
ing fuel efficiency, and even regulating climate. We are poised to live
in a world of smart spaces where most things around us (from ban-
dages to books) have computing power and interact with the world.
Until recently smart spaces have been limited by computing power,
price, and energy. Communication between nodes has been ex-
tremely limited. Moore’s law has attacked computing power and
price, a variety of initiatives are attacking energy. Finally, the
growth of wireless networking hardware and explosion of the In-
ternet promises COTS networking hardware and protocols. Or do
they?
Although many challenging impediments to smart spaces have been
overcome, we believe direct application of COTS networking pro-
tocols will not prove satisfactory. Today’s networking infrastruc-
ture has many features important in connecting smart spaces: Inter-
net protocols and implementations are open, widely implemented,
support multiple physical layers, and are relatively efficient and
lightweight. All of these characteristics are important for smart
spaces.
In spite of these advantages, existing networking approaches fail to
address the defining feature of smart spaces: smart spaces have hun-
dreds of heterogeneous, networked computers per person. Smart
spaces have hundreds of computers per person because each piece
of equipment will be tagged, each object visible output will provide
sensor data to the network, and each object with controls will be
manipulatable from the network.
Too frequently, existing protocols require manual intervention to
configure and to use. Smarts spaces will have hundreds of often het-
erogeneous objects which frequently change configuration (as they
move, are deployed, and wear out) and may be physically inac-
cessible. The implication of hundreds of nodes per person under
these conditions is that failure to autoconfigure must be considered
complete failure; failure of self-organization must be considered un-
deployability, and failure of automatic resource discovery must be
considered effective inaccessibility. A corollary of large numbers of
nodes is that smart spaces must employ algorithms which function
in the face of partial operation and heterogeneity.
Although we believe that coping with large numbers of nodes is
the fundamental challenge behind smart spaces, two other issues are
also important. As “spaces” suggests, smart spaces will be part of
the physical world. This implies that they must understand their
place (physical location and electronic neighborhood) and be able
to influence other objects. Also smart-space nodes will often be re-
source limited because of constraints of size, weight, mobility, or
cost.
The remainder of this paper explores these issues further. We ex-
amine the implications of large numbers of nodes for networking
protocols, application design and operation, and security.
2. CHALLENGES IN NETWORK
CONFIGURATION
If the fundamental change of smart spaces is a hundredfold increase
in the number of networked computers per person, then this change
is sure to have effects up and down the network stack.
At the physical layer, many nodes imply a need for very flexible
network connectivity. Wireless technologies are the obvious choice
here. Substantial progress has been made both in COTS wireless
(for example standardization and widespread deployment of IEEE-
802.11 [5]) and in the research community (for example, DARPA’s
GloMo efforts). We believe that flexible wired connections are also
important. Wired connections can provide both high-speed con-
nectivity and (if accompanied by power) can eliminate energy con-
straints.
At the network layer fully automated Internet address configuration
is a requirement. IPv6 holds some promise here with a carefully
considered autoconfiguration [11], but it assumes that failures can
be manually resolved, it bases configuration on globally unique link-
layer addresses, and its protocols require a very large address space.
Approaches to relax these constraints are important if smart spaces
are to apply to a range of link layers and to interoperate with IPv4
systems.
At the transport layer, smart spaces suggest wide use of multi-
cast protocols [1, 6]. Multicast protocols allow groups of devices
or a user and a group of sensors to interact efficiently. Multicast
group addresses decouple the data sender and list of recipients.
With thousands of nodes where weak connections and node fail-
ure may be common, flexible group membership is critical. Finally,
announce/listen-style multicast-based protocols and soft-state sim-
plify failure recovery.
Protocols which self-adapt to congestion are vital for smart spaces.
Floyd argues that it is the use of end-to-end congestion control
and particularly TCP’s additive-increase/multiplicative-decrease al-
gorithms that have allowed the Internet to cope with orders of mag-
nitude in growth [2]. If smart spaces are to adapt from the very high
node densities of a conference room to a relatively sparse city street,
similar protocols will be required.
3. CHALLENGES IN APPLICATION
CONFIGURATION
Successful network autoconfiguration will allow nodes in smart
spaces to talk with each other, and will allow them to do so with-
out overwhelming the network. In addition, application-level ap-
proaches are required to make sense out of this communication.
Since smart spaces involve the physical world, absolute or relative
physical location is an important concept. Node locations must be
automatically configured; manual configuration will be both inac-
curate and time consuming. GPS is an obvious candidate for loca-
tion determination, but several factors argue the need of additional
approaches: GPS accuracy is limited, it requires an antenna which
may be too large for some nodes, reception is limited indoors, and
cost remains a factor. Supplemental protocols are important to off-
set these problems. Accuracy of centimeters or tens of centimeters
is important to place a node on one side of a wall or the other. The
use of alternate protocols may be desired in buildings [14]. Finally,
a location-equivalent of the Network Time Protocol [8] is needed to
allow nodes to benefit from nearby GPS receivers.
Although physical location is important, we have argued elsewhere
that logical location is often more important to to real world appli-
cations [4]. (After all, do you care that you’re at latitude 33.97988N,
longitude 118.43994W, or that you’re at work and that you’re in
your office?) Improving mappings between raw physical location
and logical location are key to effective interaction between smart
spaces and the environment.
Resource rendezvous is the process of matching clients and servers
based on physical location and other attributes. Automated ren-
dezvous allows applications to share data at high-levels without
user involvement. Automated rendezvous includes both yellow-
pages services (this light switch controls the northwest bay of lights)
and more complex attribute-based queries (this light switch requests
25% illumination around the podium). Traditional approaches such
as broadcast and expanding-ring search apply to resource discovery
in smart spaces, but new opportunities exist to take advantage of
physical location and device-to-device communication.
In addition to low-level congestion control, coordination at the ap-
plication level is important to controlling network usage. We ex-
pect that devices in smart spaces will self-organize into functional
clusters. Each cluster will performing a higher-level coordinated ac-
tion. For example, devices attached to individual lighting elements
of a light panel may coordinate to dynamically vary their overall in-
tensity based on environmental factors. Self-organization protocols
build on the basic announce/listen protocols; announcements allow
frequent cluster self-organization and reorganization. While existing
clustering protocols [3] can be adapted to smart spaces, two features
distinguish clustering in in this environment:
Clusters for smart spaces frequently overlap. Functionally re-
lated devices (e.g., lighting elements) may not be topologically
contiguous, and different uses may overlap (full room lighting
vs. podium lighting). Classical clustering protocols usually fo-
cus on the formation of disjoint clusters.
Some smart-space devices may be power-constrained. More-
over, these power-constrained devices may communicate with
tethered devices that are not subject to such constraints. Clus-
tering protocols will need to be aware of energy constraints,
where possible, in order to limit communication.
A final implication of large numbers of devices is application het-
erogeneity. Smart spaces will be composed of multiple generations
of smart nodes; applications must cope with many different protocol
versions. Two very different approaches address this problem. On
one hand we can construct nodes to be field-upgradable. Active Net-
works initiatives offer promise here [10]. An alternate approach is
based on the principle of delay and discard. Nodes interact with very
flexible protocols (for example, information buses [9]), delaying the
need for frequent change. Nodes are designed to be cheap enough
that they can simply be discarded when they no longer interoperate.
4. SECURITY CHALLENGES
Just as configuration must scale to support hundreds of nodes per
person, so must security protocols. Existing protocols for key distri-
bution are important, but more understanding is needed for ways to
get good security for thousands of nodes instead of perfect security
for tens of nodes. With many nodes, some will be compromised,
so approaches to identify, isolate, and respond these nodes during
continued operation are important [7, 13]. Finally, for very small
nodes, new lightweight security protocols may be of increasing im-
portance [12].
5. CONCLUSIONS
In this paper we have argued that the key problem facing use of
smart spaces are the implications of configuring and using hundreds
of computers per person. We have examined these issues for net-
work communications, application interaction, and security impli-
cations. Addressing these problems are important if deployment of
smart spaces is to become feasible.
Acknowledgements
The authors would like to thank the members of the Simple Sys-
tems ISAT study (chaired by Deborah Estrin) for their discussions
on issues related to smart spaces.
This paper originally appeared in the DARPA/NIST Smart Spaces
Workshop, July 30-31, 1998, Gaithersburg, Maryland, USA.
References
1. Stephen E. Deering and David R. Cheriton. Multicast routing
in datagram internetworks and extended lans. ACM Transac-
tions on Computer Systems, 8(2):35–110, May 1990.
2. Sally Floyd and Kevin Fall. Promoting the use of end-to-end
congestion control in the internet. Sumibtted to IEEE/ACM
ToN, February 1998.
3. Ramesh Govindan, Cengiz Alaettino˘ glu, and Deborah Estrin.
Self-configuring active network monitoring (SCAN). White
paper, February 1997.
4. John Heidemann and Dhaval Shah. Experiences with user-
configurable, location-aware scheduling. Technical Report 98-
675, University of Southern California, April 1998. submitted
for publication.
5. IEEE. Wireless LAN Medium Access Control (MAC) and
Physical Layer (PHY) Specifications (IEEE 802.11). IEEE,
1997.
6. Van Jacobson. Multimedia conferencing on the internet. Tu-
torial at SIGCOMM ’94; published as a special issue of the
SIGCOMM Newsletter, August 1994.
7. Leslie Lamport, Robert Shostak, and Marshall Pease. The
Byzantine generals problem. ACM Transactions on Program-
ming Languages and Systems, 4(3):382–401, July 1982.
8. Dave L. Mills. Network time protocol (version 2) specification
and implementation. RFC 1119, Internet Request For Com-
ments, September 1989.
9. Brian Oki, Manfred Pfluegl, Alex Siegel, and Dale Skeen. The
information bus—an architecture for extensible distributed
systems. In Proceedings of the 14th Symposium on Operating
Systems Principles, pages 58–68, Asheville, North Carolina,
USC, December 1993. ACM.
10. David L. Tennenhouse, Jonathan M. Smith, W. David Sin-
coskie, David J. Wetherall, and Gary J. Minden. A survey
of active network research. IEEE Communications Magazine,
35(1):80–86, January 1997.
11. S. Thomson and T. Narten. Ipv6 stateless address autoconfig-
uration. RFC 1971, Internet Request For Comments, August
1996.
12. Gene Tsudik and Brian Tung. On constructing optimal one-
time signatures. Submitted for publication, November 1997.
13. Brian Tung. Crisis: Critial resoruce allocation and
intrusion response for survivable information systems.
http://gost.isi.edu/projects/crisis/, January 1998.
14. Andy Ward, Alan Jones, and Andy Hopper. A new location
technique for the active office. IEEE Personal Communica-
tions Magazine, 4(5):8–15, October 1997.

Abstract (if available)

Linked assets

Computer Science Technical Report Archive

Conceptually similar

PDF

USC Computer Science Technical Reports, no. 692 (1999)

PDF

USC Computer Science Technical Reports, no. 750 (2001)

PDF

USC Computer Science Technical Reports, no. 669 (1998)

PDF

USC Computer Science Technical Reports, no. 631 (1996)

PDF

USC Computer Science Technical Reports, no. 706 (1999)

PDF

USC Computer Science Technical Reports, no. 774 (2002)

PDF

USC Computer Science Technical Reports, no. 723 (2000)

PDF

USC Computer Science Technical Reports, no. 745 (2001)

PDF

USC Computer Science Technical Reports, no. 704 (1999)

PDF

USC Computer Science Technical Reports, no. 603 (1995)

PDF

USC Computer Science Technical Reports, no. 667 (1998)

PDF

USC Computer Science Technical Reports, no. 732 (2000)

PDF

USC Computer Science Technical Reports, no. 697 (1999)

PDF

USC Computer Science Technical Reports, no. 731 (2000)

PDF

USC Computer Science Technical Reports, no. 729 (2000)

PDF

USC Computer Science Technical Reports, no. 675 (1998)

PDF

USC Computer Science Technical Reports, no. 682 (1998)

PDF

USC Computer Science Technical Reports, no. 935 (2013)

PDF

USC Computer Science Technical Reports, no. 741 (2001)

PDF

USC Computer Science Technical Reports, no. 686 (1998)

Description

John Heidemann, Ramesh Govindan, Deborah Estrin. "Configuration challenges for smart spaces." Computer Science Technical Reports (Los Angeles, California, USA: University of Southern California. Department of Computer Science) no. 677 (1998).

Asset Metadata

Creator Estrin, Deborah (author), Govindan, Ramesh (author), Heidemann, John (author)

Core Title USC Computer Science Technical Reports, no. 677 (1998)

Alternative Title Configuration challenges for smart spaces (title)

Publisher Department of Computer Science,USC Viterbi School of Engineering, University of Southern California, 3650 McClintock Avenue, Los Angeles, California, 90089, USA (publisher)

Tag OAI-PMH Harvest

Format 3 pages (extent), technical reports (aat)

Language English

Unique identifier UC16270343

Identifier 98-677 Configuration Challenges for Smart Spaces (filename)

Legacy Identifier usc-cstr-98-677

Format 3 pages (extent),technical reports (aat)

Rights Department of Computer Science (University of Southern California) and the author(s).

Internet Media Type application/pdf

Source 20180426-rozan-cstechreports-shoaf (batch), Computer Science Technical Report Archive (collection), University of Southern California. Department of Computer Science. Technical Reports (series)

Access Conditions The author(s) retain rights to their work according to U.S. copyright law. Electronic access is being provided by the USC Libraries, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name USC Viterbi School of Engineering Department of Computer Science

Repository Location Department of Computer Science. USC Viterbi School of Engineering. Los Angeles\, CA\, 90089

Repository Email csdept@usc.edu

Inherited Values

Title Computer Science Technical Report Archive

Coverage Temporal 1991/2017

Repository Email csdept@usc.edu

Repository Name USC Viterbi School of Engineering Department of Computer Science

Repository Location Department of Computer Science. USC Viterbi School of Engineering. Los Angeles\, CA\, 90089

Publisher Department of Computer Science,USC Viterbi School of Engineering, University of Southern California, 3650 McClintock Avenue, Los Angeles, California, 90089, USA (publisher)