Participatory design of sensing networks: strengths and challenges.

Full text

Page 1

eScholarship provides open access, scholarly publishing
services to the University of California and delivers a dynamic
research platform to scholars worldwide.
Center for Embedded Network Sensing
University of California
Peer Reviewed
Title:
Participatory Design of Sensing Networks: Strengths and Challenges
Author:
Shilton, Katie, University of California, Los Angeles
Ramanathan, Nithya, UCLA
Reddy, Sasank
Samanta, Vidyut
Burke, Jeffrey A
Estrin, D
Hansen, Mark
Srivastava, Mani B.
Publication Date:
06-15-2008
Series:
Technical Reports
Publication Info:
Technical Reports, Center for Embedded Network Sensing, UC Los Angeles
Permalink:
http://escholarship.org/uc/item/7bx0g78h
Additional Info:
To appear as an exploratory paper in the Proceedings of the Participatory Design Conference
(PDC '08).
Original Citation:
Shilton, K., Ramanathan, N., Reddy, S., Samanta, V., Burke, J., Estrin, D., Hansen, M.
and Srivastava, M. Participatory Design of Sensing Networks: Strengths and Challenges. In
Proceedings of the Participatory Design Conference 2008 (Bloomington, IN, 2008), ACM Press.
Keywords:
Sensing networks, community-based participatory research
Abstract:
Participatory design (PD) involves users in all phases of design to build systems that fit user needs
while simultaneously helping users understand complex systems. We argue that traditional PD
techniques can benefit participatory sensing: community-based participatory research (CBPR)
projects in which complex technologies, such as sensing networks using mobile phones, are the

Page 2

eScholarship provides open access, scholarly publishing
services to the University of California and delivers a dynamic
research platform to scholars worldwide.
research instruments. Based on our pilot work on CycleSense, a community-based data gathering
system for bicycle commuters, we discuss the benefits and challenges of PD in participatory
sensing settings, and outline a method to integrate PD into the research process.

Page 3

Participatory Design of Sensing Networks:
Strengths and Challenges
K. Shilton, N. Ramanathan, S. Reddy, V. Samanta, J. Burke, D. Estrin, M. Hansen, M. Srivastava
Center for Embedded Networked Sensing (CENS), University of California, Los Angeles
{kshilton, nithya, vids, sasank, jburke, destrin, cocteau, mbs}@ucla.edu
ABSTRACT
Participatory design (PD) involves users in all phases of
design to build systems that fit user needs while
simultaneously helping users understand complex systems.
We argue that traditional PD techniques can benefit
participatory sensing: community-based participatory
research (CBPR) projects in which complex technologies,
such as sensing networks using mobile phones, are the
research instruments. Based on our pilot work on
CycleSense, a community-based data gathering system for
bicycle commuters, we discuss the benefits and challenges
of PD in participatory sensing settings, and outline a
method to integrate PD into the research process.
Keywords
Sensing networks, community-based participatory research
INTRODUCTION
Mobile phone networks provide billions of users with
potential platforms for data collection. Using image, sound,
and location-gathering modalities, phones can collect data
previously too granular, time-consuming, or difficult to
record [1-4]. Mobile phones are also familiar, easy to use,
and widely available. This gives them widespread potential
to serve as tools for community-based participatory
research: methodologies that
members into research projects as co-investigators [5, 6].
When people use mobile sensing systems as instruments in
such research, we call it participatory sensing.
Participatory sensing is inspired by decades of research
promoting community involvement in neighborhood
documentation and representation. By coordinating
increasingly-available devices, participatory sensing offers
automation, scalability, and possibilities for real-time
upload, processing and feedback. These features can
augment traditional CBPR efforts such as participatory
urban planning [7], geographic information systems [8],
and Photovoice initiatives [9].
For example, the Center for Embedded Networked Sensing
(CENS), a science and engineering research center,
recently collaborated with the nonprofit Livable Places1 to
asses the pedestrian and bike-friendliness of two Los



integrate community

1 http://www.livableplaces.org/
Angeles neighborhoods. Teams equipped with GPS and
mobile phones travelled the neighborhood to gather geo-
tagged images and apply tags to document impediments to
pedestrians and bikers. Inspired by this pilot, we are
working with bike commuters to design the “CycleSense”
system, which enables cyclists to plan campaigns—targeted
data collection efforts—to document the safety and quality
of their routes. CycleSense builds on previous work
instrumenting bicycles for data collection [10] to enable
bikers to specify route needs and preferences, collect data
about their existing routes, and learn from other’s data to
discover safer, more comfortable rides.

Figure 1: Participants use mobile phones to update and
improve a map interface
In systems such as CycleSense, research decisions about
what data to collect, and at what granularity, affect design
of the collection instruments. For example, how can
designers equip mobile phones to document data of interest
to cyclists, such as potholes, without inconvenience or
safety risk to the rider? Traditional CBPR tackles these
challenges by involving participants in the design of paper-
based research instruments such as maps or surveys [11].
Cooperation to create surveys
straightforward process; cooperation to design sensing
networks may be less so. Participatory sensing data
collection instruments include a network of mobile phones,
a central database, and web- and phone-based user
interfaces. Creating complex technologies for data
gathering in a participatory research environment requires
incorporating participatory design (PD) as a component of
participatory sensing.
is a relatively

Page 4

Participatory design involves a technology’s intended users
throughout the design process. Traditional CBPR methods
do not explicitly address technology design. But in research
settings that rely upon technology for data gathering, PD
can ensure participation in specifying and designing
research tools. In the following section, we outline benefits
of PD for participatory sensing. We then discuss the
challenges of applying PD in participatory sensing, and
propose methods to meet those challenges.
BENEFITS OF PD FOR PARTICIPATORY SENSING
Harnessing known strengths of PD will not only improve
the functionality of participatory sensing technologies, it
will also enable community participants to use these
technologies more effectively. Advantages of PD for
sensing network development include methods to target
local knowledge and address relevant community
challenges; techniques to foster participant understanding
of sensing systems and consequently improve systems with
participant feedback; and processes to mitigate privacy
concerns through engagement and participant control.
Gathering Local Knowledge
In participatory sensing projects, participants are experts on
their surroundings and the associated challenges of their
environments. Fully involving community members in the
design process can therefore help researchers develop
sensing technologies that will gather not just data, but local
knowledge. Local knowledge is the understanding gained
by living within a particular setting and social group [7].
An integral part of community-based participatory research
projects, local knowledge is notoriously difficult for
outsiders to access [12, 13]. Participatory design methods
are intended to incorporate local knowledge within the
technology design process [14]. Adding techniques such as
participant observation [15], storytelling, and cooperative
prototyping [16] to the design of sensing networks can
ensure that participatory sensing efforts collect data that
both reflects and contributes to local knowledge.
Local knowledge can benefit participatory sensing in two
ways. Designing sensing technologies with the cooperation
of community partners enables projects to target the
documentation of phenomena known to local residents, but
difficult to prove to decision-makers and authorities. In
CycleSense, for example, riders may know of poor bike
path surfaces neglected by city authorities. Prior
knowledge of poor surfaces can prompt design of sensing
networks that safely and easily record time-stamped, geo-
tagged images of road hazards. Automatically aggregating
these images can document the safety risks posed to bikers.
Participatory sensing systems can also help participants
establish the credibility of their data. To make a case for
change in a community, participants must be able to defend
data validity. Working with participants to define validity
needs and threats, we can increase validity with a variety of
measures. These include infrastructure-based verification
of data, in which participants use trusted networks (e.g.
wireless networks managed by project leaders) to upload
data. The network adds time and location tags to data. An
attestation service can check these tags for consistency with
metadata attached by the phone. Organizers could also ask
participants to verify other’s contributions. In CycleSense,
for example, the system might send an image and tag
uploaded by Participant A to Participant B, who takes a
similar route to work. The system would ask B to overlap
with A’s route to verify the image and associated tag.
There is an epistemological benefit of incorporating and
verifying local knowledge during participatory sensing.
Local knowledge recognizes local problems. Identification
of local problems can spark new community research
projects that outside researchers might otherwise miss. The
design cycle for sensing technology is long and the process
can be expensive. Focusing design on the realities of
participating communities will help researchers target
systems to relevant community challenges.
Understanding and Improving Systems
Sensing networks provide accurate and granular data, but
they also produce more data than a human can easily parse.
CycleSense, for example, might collect hours of latitude
and longitude readings to tie to complex models of traffic
density or air quality. Because copious sensing data often
must be aggregated or visualized to be legible,
participatory sensing systems require sophistication and
experience to analyze research results. For community
members to participate effectively in participatory sensing
projects, they will need to understand the system’s data
flow and analysis process.
researchers and community members to design data
analysis and visualization interfaces can make complicated
data legible to communities relying on that data.
Cooperating to decide what data is collected, where the
data is sent and stored, and how the data is processed,
interpreted, and displayed will additionally help participant
designers be effective researchers and contribute to their
understanding of the accuracy and reliability of that data.
Participants who understand precisely where their data
comes from, how it is verified, and what it means will be
able to understand and argue for the validity of the
evidence they collect. They will also be better able to
identify data that indicates breakdowns in the sensing
system.
Feedback from participants will also help designers
improve sensing systems. Iterative work with Livable
Places’ volunteers helped us refine our first implementation
of a mobile-phone based image capture and tagging
system. Participants who had trouble using the system were
able to work directly with designers to improve the
system’s interface, which improved participant technology
literacy while concretely benefitting sensing system
design.
Cooperation between

Page 5

Mitigating Privacy Concerns
Because participatory sensing systems use mobile phones
to capture data at unprecedented resolution, the technology
is inherently intrusive. In CycleSense, for example, bikers
might contribute GPS traces that identify their home, work,
and daily routines. People’s willingness to have devices
record, share, and retain such personal data depends upon a
host of personal and social factors, including at what
granularity data is captured, where data is captured, who
will have access to the data, and for how long [17].
Because people’s willingness to share their personal data is
variable and highly contextual [18], system privacy design
must respect this variability.
Participatory sensing attempts to address this variability by
encouraging participants to discuss privacy and make
choices throughout design and use of the system [19].
Participant research decisions include weighing the
sensitivity of a campaign’s data against research benefits.
Throughout the system design process, PD methods can
encourage community members towards understanding of,
and consensus on, system defaults and user choices for data
granularity, data sharing, lengths of time for data retention,
and permissions for data reuse.
CHALLENGES FOR PD AND PARTICIPATORY SENSING
Participatory design methods bring a number of strengths
to participatory sensing, but these projects also pose
difficult problems that remain open challenges in the PD
literature.
Working with Diverse User Populations
Participatory design projects have traditionally focused on
workplace technologies with discrete user groups [16, 20-
22]. Participatory sensing necessitates partnership with
organizations representing diverse demographic groups.
Sensing participants will have different levels of
technology experience, speak multiple languages, and be
variously comfortable taking on design and research roles.
Working with loosely-affiliated communities not united
under a work organization is an unsolved PD challenge
[16].
Compounding the problem of diverse user groups is the
newness of participatory sensing technologies. Our
introductory experience
participatory projects with community groups points to a
major challenge often faced in participatory design, and
exacerbated by moving PD into the context of data
gathering and CBPR. Not unlike the challenges of
introducing internet-based workplace technologies a
decade ago [22], communicating the possibilities and limits
of sensing systems, and in turn learning from community
groups, is hampered by the newness of the technology.
Developing methods to describe the possibilities of sensing
technologies without imposing creative limits on
participant groups is a major challenge for participatory
sensing projects. To address this challenge, we are
exploring PD methods such as scenario construction and
broaching and planning
design
discussions and scenarios focused on data (“what
information do you need to know?”) towards systems
(“what could help you find that out?”)
Diverse user populations also present challenges to system
scalability. Though some participants are actively involved
in design and implementation, the privacy and ethical
preferences of this subset may not be indicative of needs as
wider populations adopt the technologies. Designing
customizable systems for new user populations (for
instance enabling flexible
customizable levels for data sharing) can help address this
challenge.
games, which encourage movement from
privacy defaults and
Institutional Challenges
CENS faces internal cultural challenges that will affect
participatory design projects. Foremost among the internal
challenges may be reliance on a distributed design process.
As a research institution composed of faculty, graduate
students, undergraduates, and staff, CENS disperses
authority for design and implementation throughout the
organization. A project may involve a dozen designers
responsible for different aspects of planning, coordination,
and code. How will the center manage the logistics of PD
while cooperating with a large number of community
members and a large number of designers? Pekkola et al’s
definition of “mediators”—a
betweens—may prove useful in the CENS design
environment [23].
A further challenge will arise if participatory design
techniques slow sensing system implementation. In an
academic environment where researchers must build
models, test systems, and publish results, how do we justify
a slower, stickier design process? There is some evidence
that a prolonged design process may enable new nodes of
innovation [14, 20]. We hope to find that a slower, user-
centered design process prompts advances to answer many
of the challenges inherent to participatory sensing systems.
few designated go-
LINKING PD AND PARTICIPATORY SENSING
Participatory sensing is a research process: people gather
data to learn about a phenomenon of interest and come to
new understandings. Participatory sensing participants will
formulate research questions; plan campaigns and design
instruments to answer those questions; deploy campaigns;
and analyze and present results. During this process,
participants form flexible communities of practice: groups
bound by common purpose that develop a shared store of
knowledge and set of accepted work behaviors [24].
Designers can observe and participate in the formation of
these research groups to find spaces for participatory
design decisions and interventions. Building on PD
methodologies such as visioning [25], scenario-based
design [26], software prototyping [23], and design games
[25], sensing researchers can actively incorporate design
elements into the participatory research process.

Page 6

CycleSense will be our first systematic exploration of
participatory design of sensing systems. We are recruiting
self-identified members of the bike commuter community
through the Los Angeles Bike Coalition,2 blog outreach,
and word-of-mouth. Design methods will include leading
visioning workshops with bikers during research planning
(“In an ideal world…”) [25]; discussing incomplete
scenarios to spark debate and iteration on potential designs
[26]; prototyping systems
observational data and explicit responses [23]; and asking
users to keep journals during system pilots to gather ideas,
criticism and feedback. The CycleSense PD plan takes
advantage of the existing structures of the research process
to incorporate design of the research instruments.
with users to gather
FUTURE WORK
As researchers continue participatory sensing projects,
future work will explore instances of creativity and
understanding that arise in the cooperative design process.
Case studies of PD will also illuminate new challenges for
incorporating PD into participatory sensing and research.
ACKNOWLEDGEMENTS
The authors would like to thank Livable Places’ team and
Dorothy Kieu Lê of the Los Angeles Bike Coalition.
Thanks also to anonymous reviewers and colleagues in
UCLA’s Department of Information Studies for comments
on earlier versions of the paper.
REFERENCES
1. Paulos, E., et al. Sensing Atmosphere, in Proceedings of
SenSys (Sydney, Australia, 2007), ACM.
2. Burke, J., et al. Participatory sensing, in Proceedings of
Sensys (Boulder, CO, November 2006), ACM.
3. Eisenman, et al. MetroSense Project: People-Centric
Sensing at Scale, in Proceedings of SenSys (Boulder,
CO, November, 2006), ACM.
4. Eagle, N. et al. Reality mining: sensing complex social
systems. Pers Ubi Comp (2006), 255-268.
5. Cargo, M. et al. The value and challenges of
participatory research: strengthening its practice. Annu
Rev Publ Health, 29 (2008).
6. Byrne, E. et al. Questions of ethics: Participatory
information systems research in community settings, in
Proceedings of SAICSIT (Cape Winelands, SA, 2006).
7. Corburn, J. Bringing
environmental decision making: Improving urban
planning for communities at risk. J Plan Edu Res, 22
(2003), 120-133.
8. Elwood, S. Critical issues in participatory GIS:
Deconstructions, reconstructions, and new research
directions. T GIS, 10, 5 (2006), 693-708.
local knowledge into

2 http://la-bike.org/
9. Wang, C., et al. Chinese village women as visual
anthropologists: a participatory approach to reaching
policymakers. Soc Sci Med, 42, 10 (1996), 1391-1400.
10. Eisenman, S. B., et al. The BikeNet mobile sensing
system for cyclist experience mapping , in Proceedings
of SenSys (Sydney, Australia, 2007), ACM.
11. Shallwani, S. et al. Community-based participatory
research: a training manual for community-based
researchers. Aga Khan
Development Programme, 2007.
12. Srinivasan, R. Reconstituting the Urban through
Community-Articulated Digital Environments. J Urban
Technol, 11, 2 (2004), 93-111.
13. Srinivasan, R. et al. The South Asian Web: An
Emerging Community Information System in the South
Asian Diaspora, in Proceedings of PDC '06 (Trento,
Italy, 2006), ACM.
14. Gregory, J. Scandinavian Approaches to Participatory
Design. Int J Eng Educ, 19, 1 (2003).
15. Crabtree, A. Ethnography in Participatory Design, in
Proceedings of PDC '98 (Seattle, WA, 1998), ACM.
16. Muller, M. J. Participatory design: the third space in
HCI. Handbook of HCI. Erlbaum, Mahway, NJ, 2003.
17. Waldo, J., et al. Engaging privacy and information
technology in a digital age. The National Academies
Press, Washington, D.C., 2007.
18. Nissenbaum, H. Privacy as contextual integrity. Wash
Law Rev, 79, 1 (2004), 119–158.
19. Shilton, K., et al. Participatory privacy in urban sensing,
in Proceedings of MODUS '08 (St. Louis, MI, 2008).
20. Pilemalm, S. et al. Third generation participatory design
in health informatics. J Biomed Inform (2007).
21. Dredger, S. M., et al. Using participatory design to
develop (public) health decision support systems
through GIS. Int J Health Geo, 6, 53 (2007).
22. Kensing, F. et al. Participatory design: issues and
concerns. Computer Supported Cooperative Work, 7, 3
(1998), 167-185.
23. Pekkola, S., et al. Towards formalized end-user
participation in information systems development
process, in Proceedings of PDC '06, (Trento, Italy,
2006) ACM.
24. Davenport, E. et al. Organizational Knowledge and
Communities of Practice. Annu Rev Inform Sci, 36
(2002), 171-227.
25. Brandt, E. Designing exploratory design games: a
framework for participation in Participatory Design? in
Proceedings of PDC '06 (Trento, Italy, 2006), ACM.
26. Carroll, J. M. Making Use: Scenario-Based Design of
Human-Computer Interactions. The MIT Press,
Cambridge, MA 2000.
University Human