PEIR: Personal Environmental Impact Report

1
PEIR Personal Environmental Impact Report
Illuminate our individually-made, globally-felt
choices through real-time assessment of personal
environmental impact and exposure, using data
sensed by mobile handsets.

• Jeff Burke, Deborah Estrin, Mark Hansen
• UCLA Center for Embedded Networked Sensing
• UCLA Center for Research in Engineering, Media
  and Performance
• Doug Houston, Andrew Mondschein
• UCLA Institute for Transportation Studies

CENS Urban Sensing is collaborative work of many
faculty, staff, and students Mark Allman, Jeff
Burke, Gong Chen, Dana Cuff, Ryan Dorn, Deborah
Estrin, Mark Hansen, August Joki, William Kaiser,
Jerry Kang, Eitan Mendelowitz, Andrew Parker,
Vern Paxson, Nicolai Munk Petersen,Sasank Reddy,
Vids Samanta, Thomas Schmid, Mani Srivastava,
Fabian Wagmister, Nathan Yau, and others. In
partnership with NSF NeTS-FIND, Cisco, Nokia,
Schematic, Sun, UCLA REMAP, UCLA ITS, Walt Disney
Imagineering RD
2
PEIR Concept

• Existing footprint calculators inform
  long-term choices using coarse-grained models of
  our impact on the environment.

www.footprinter.co.uk
PEIR, a personalized, real-time assessment can
help individuals reduce impact from day-to-day
choices and minimize exposure to local
environmental hazards. As a standalone tool,
PEIR invites investigation of ones own practices
and habits in relationship to others and the
environment, as seen in data and inferred from
models. As a building block, PEIR could enable
other productivity software to provide
integrated feedback about benefits and impacts of
everyday choices in our purchases, commutes, and
other habits.
bp carbon footprint calculator
3
PEIR Precedent

• Environmental Impact Assessment
• The process of identifying, predicting,
  evaluating and mitigating the biophysical,
  social, and other relevant effects of development
  proposals prior to major decisions being taken
  and commitments made.
• International Association for Impact Assessment
  (1997).

Example Air Quality Assessment in
Industry (Norwegian Inst. for Air Research)
Health Impact Assessment Procedures, methods
and tools by which a policy, programme or project
may be judged as to its potential effects on the
health of a population, and the distribution of
those effects within the population. European
Centre for Health Policy, WHO Regional Office for
Europe (1999).
Participatory Appraisal Local people, using the
methods of participatory inquiry, have shown a
greater capacity to observe, diagram and analyse
than most professionals have expected. Whose
Eden An Overview of Community Approaches to
Wildlife Management, IIED and ODA, London (1984).
4
PEIR Prototype System

• Employ the built-in capabilities of modern
  mobile handsets in order to scale to many users
  without specialized hardware.
• Explore model-based analyses possible with
  location traces generated using GPS, cell tower
  and WiFi beaconing, assisted by image and audio
  context.
• Incorporate evocative, accurate data
  visualization and exploration to promote
  individual and community engagement with the
  sensing and modeling process, not just the model
  output.

System components GPS-equipped mobile
handset. Custom handset software for automatic
location time-series collection, robust upload,
over-the-air upgrade/tasking, just-in-time
annotation with voice or text. Server side tools
to analyze individual spatio-temporal patterns
and calculate corresponding impact and exposure
metrics to inform and advise users. Web-based
interfaces informing and advising users, which
provide reports, real-time feedback,
visualizations and exploratory data analysis
tools for non-professional users. (For handsets
and workstations.)
Campaignr
Trace, audio, image
SensorBase
N80, N95


Activity type inference
Server-side classifier
Impact / Exposure Model
5
PEIR Real-time Impact Analysis

• Unrolling usage models
• PEIR monitoring can help account for significant
  driver- and road-specific effects that current
  footprint calculators can only approximate
• Location-specific impacts
• Road conditions, real-time traffic monitoring and
  aggregated driver behaviors in regions can be
  used to estimate localized impacts

Holmen and Neimeier (1997) McCrae (2004).
6
PEIR Exposure Assessment

• Daily activity patterns
• Significant health problems result from complex
  interactions between genetic and environmental
  factors
• Activity patterns can have a significant
  influence on personal exposure to environmental
  risk factors
• PEIR monitoring, combined with micro-environmental
  models could make personal exposure assessment
  possible

Epidemiological research These data might also
replace surveys (the EPAs Consolidated Human
Activity Database) of activity patterns in risk
assessment studies
Harrison, et al. (2002) Klepeis et al, NHAPS
(2001) EPA FERA.
7
PEIR Challenges

• Sensing and classification
• Inference of user actions from location traces.
• Appropriate data protection and privacy
  controls.
• Modeling
• Unrolling, adapting, and refining of exposure
  and impact models.
• Incorporating uncertainty and integrity
  information from sensing/classification.
• Interface
• Accessible, engaging, and accurate end-user
  interfaces to data and the sensing, modeling, and
  inference process.
• Balancing real-time feedback with
  retrospective analysis.
• System
• Robust, scalable, and secure implementation.

8
PEIR Preliminary Timeline
Alpha version of handset software. Trial data collection. Initial model selection and end-user interface concept design. Alpha version of server-side location trace activity type classifier with overlay visualization on map. Beta handset capture software incl. feedback from server. Alpha version of exposure/impact calculations and visualization. Refined exposure/impact models based on pilot data. Testing of integrated data gathering, inference, and calculations with ground truth data collection for verification. Alpha version of user advising and web interface suite. Private integrated beta testing, evaluation. Public beta release. Complete Ongoing Jun 07 - Jul 07Aug 07 Sep 07 Sep 07 Jan 08 Jan 08 - Feb 08 Mar 08 Apr 08 - May 08 May 08
9
PEIR Context Participatory Urban Sensing

• Heterogeneous, multiscale, human-in-the-loop
  sensing systems enhanced by mobility.
• Leverage the installed base of mobile phones and
  their on-board sensors, in combination with
  server-based geospatial data and models.
• For PEIR Sensing GPS, Cell Tower ID, WiFI
  beaconing, plus image and audio as context.
• Models Demographics, transportation, air
  quality, energy use, etc.
• Participatory sensing systems as personal and
  social tools to increase the legibility of
  everyday life.

10
Urban Sensing Campaign Model

• Distributed data gathering challenges as
  Campaigns -
• Spatially and temporally constrained systematic
  data collection operations.
• Exploring a single hypothesis, phenomena or
  theme.
• Using human-in-the loop sensing to gather data.
• With automatic and manual classification,
  auditing, and analysis.
• Phases - Define, Recruit, Gather, Audit/Upload,
  Analyse, Publish
• Precedent - Community-Based Participatory
  Research

Citizen Science World Water Quality Day
PhotoVoice Caroline Wang, 1996
Participatory GIS Ctr for Neighborhood Knowledge
Civic Participation Video the Vote
Citizen Science Cornell e-Bird
11
Urban Sensing Research Challenges

• Onboard processing.
• Adaptive collection protocols.
• Imager and microphone as sensors.
• Scaling and credibility.
• Coordinated, opportunistic sampling.
• Network attestation and verification of
  location, time, and other context.
• Encouraging sharing.
• Reputation, incentive, and authoring
  frameworks.
• Data protection and selective,
  resolution-controlled dissemination.
• Anonymous and pseudonymous participation.
• Spatial interfaces to data and authoring.
• Finding, visualizing, and analyzing data.
• Data stream naming, privacy-respecting
  discovery, and signal search.
• Server-side signal processing for data
  processing, browsing, and auditing.

12
Urban Sensing Other Pilot Campaigns
DietSense - Ongoing
Sidewalk Walkability - Ongoing
Those in walkable neighborhoods were more
likely to know their neighbors, participate
politically, trust others, and be socially
engaged.
Enhance dietary intake and food purchase choice
monitoring, and reduce self-reporting bias.
Pilot of NIH Proposal with W. McCarthy, School of
Public Health, K. Watson, et al., David Geffen
School of Medicine.
Leyden KM. Social Capital and the Built
Environment The Importance of Walkable
Neighborhoods. American Journal of Public
Health, Sept 2003 93(9).
Autonomous image capture and upload, quarantine
and auditing of data, privacy-enhancing image
processing, tools for image tagging, clustering,
browsing and search.
Coordinated opportunistic sampling of image,
audio, location, user-triggered sampling with
background location trace, reputation and
incentive management.
L.A. Noisemap - Summer
Remapping L.A. - Summer
Distributed, geotagged photodocumentary by
Angelenos for public experience at the Los
Angeles State Historic Park.
Sound level and frequency characteristic as a
quality of life metric and proxy for
transportation (traffic!) impact.
With the Center for Neighborhood Knowledge,
Dept. of Urban Planning, UCLA.
With the Center for Research in Engineering,
Media and Performance, California State Parks,
Disney Imagineering RD, and others.
Coordinated opportunistic sampling, client-side
and server-side signal processing, sensor
calibration, context resolution control,
reputation management.
User-driven geotagged image and audio collection,
privacy protections and anonymous participation,
tools for image tagging, clustering, browsing and
search.
13
Urban Sensing Existing and New Tools
ImageScape
SensorBase
Campus/Downtown WiFi
Campaignr, StarScape
14
PEIR Personal Environmental Impact Report
Illuminate our individually-made, globally-felt
choices through real-time assessment of personal
environmental impact and exposure, using data
sensed by mobile handsets.

• Jeff Burke, Deborah Estrin, Mark Hansen
• UCLA Center for Embedded Networked Sensing
• UCLA Center for Research in Engineering, Media
  and Performance
• Doug Houston, Andrew Mondschein
• UCLA Institute for Transportation Studies

CENS Urban Sensing is collaborative work of many
faculty, staff, and students Mark Allman, Jeff
Burke, Gong Chen, Dana Cuff, Ryan Dorn, Deborah
Estrin, Mark Hansen, August Joki, William Kaiser,
Jerry Kang, Eitan Mendelowitz, Andrew Parker,
Vern Paxson, Nicolai Munk Petersen, Sasank
Reddy, Vids Samanta, Thomas Schmid, Mani
Srivastava, Fabian Wagmister, Nathan Yau, and
others. In partnership with NSF NeTS-FIND,
Cisco, Nokia, Schematic, Sun, UCLA REMAP, UCLA
ITS, Walt Disney Imagineering RD