Participatory sensing The NoiseTube project Nicolas Maisonneuve, Associate Researcher, SONY CSL Paris

Issue 1: Scarcity of fine-grained data Issue 2: Limited role of citizens Environmental management Change of paradigm in environmental data production Citizens in the loop Environmental/ health Sciences Social /political sciences Participatory sensing Research

Change of paradigm in environmental data production

Citizens in the loop

Data gathering: a general problem Water : limited accurate info on water and treatment systems Food: Agricultural statistics deteriorated over time Health: uncertainty in e xposure assessment [Poor data, weak agencies hamstring U.N. environmental oversight, NY Times, 2009] [ FAO, Audit 2009] [Uncertainty and Data Quality in Exposure Assessment, WHO, 2008] Poor system of monitoring

Why do we need to involve citizens? “ Environmental issues are best handled with the participation of all concerned citizens..” [Principle 10, Rio Declaration, 1992] But, in reality.. no real participation despite international agreements Environmental issues = anthropogenic effect Involvement  change of behavior

Opportunity for Participatory Sensing  New usage of phone phone = personal measurement instrument  Individual level : new user experience Autonomy to measure local phenomena New form of expression/collaboration Collective level: adaptative sensor network with a limited cost Empowerment in the digital world Growing public concern + + Democratization of powerful & sensor-rich phones

 New usage of phone

phone = personal measurement instrument

 Individual level : new user experience

Autonomy to measure local phenomena

New form of expression/collaboration

Collective level: adaptative sensor network with a limited cost

Case study: noise pollution monitoring in EU EU countries Health: 170 Million citizens impacted Economical cost: €12 billion issues Limited modeling Lack of observation Emission vs exposure EU Call for real data! “ Goals for future research include supplying the missing data .” [EC, 2004] “ Every effort should be made to obtain accurate real data “ [EC WG, 2006]  No real exposure data at the individual level Reality Noise map simulation VS.

Limited modeling

Lack of observation

Emission vs exposure

EU Call for real data!

“ Goals for future research include supplying the missing data .” [EC, 2004]

“ Every effort should be made to obtain accurate real data “ [EC WG, 2006]

 No real exposure data at the individual level

Participatory monitoring of noise pollution using mobile phones www.noisetube.net Goal: Enabling citizens to measure and tag their everyday exposure to inform the community by:  supplying real exposure data  building a collaborative exposure map of their environment

How does it work? Human-based sensing No sound sent, only exposure  User Privacy, user experience, Scalability LAeq(1s) computation (SP) Exposure Pattern recognition sensors (microphone, gps) computer + Phone-based sensing free tagging of exposure Diffusion of Raw exposure data Bi-directional service Collaborative exposure map NoiseTube serve Public data commons Data postprocessing (GIS) Gis server

LAeq(1s)

computation (SP)

Exposure Pattern recognition

But, what about the accuracy? Real-world experiment Park Experiment In lab Person equipped with sensors Phone + hand free kit Professional sensors ? = Virtual noise sensor = microphone + software Sound Level Meter (200$) Collaboration with After correction: error 2 db

But, what about the accuracy? ? = Virtual noise sensor = microphone + software Sound Level Meter (200$) Real-world experiment Experiment In lab Phone in Hand Hand free kit Phone in pocket +/- 2,5 db +/- 4,5 db +/- 6,5 db correction in lab phone correction dB dB

And measuring is not enough.. Hazard identification issues Difficult to identify & separate noise sources from exposure recordings O pportunity: tagging exposure which source generated these levels? Noise map Sound level meter map with a semantic layer People are excellent at recognizing noise sources  Using people as sensors to tag the source of their exposures (and thus inform the community)

Real-time collective exposure Simple visualisations for citizens/ police makers Exposure layer Semantic layer Contextual information Real-time Data aggregation by small piece of street

Exposure layer

Semantic layer

Contextual information

Real-time

Connected to the web ecosystem Data commons accessible via public web API (GeoRSS, JSON) (e.g. tracking new data in an area) Connected to the people Real-time s preading of environmental information through social network service  Enhancing the conversation & dissemination (push) Multiple ways to diffusion information (in progress) SMS

Connected to the web ecosystem

Data commons accessible via public web API (GeoRSS, JSON) (e.g. tracking new data in an area)

Connected to the people

Real-time s preading of environmental information through social network service

 Enhancing the conversation & dissemination (push)

Empowerment - case study 1: Subway in Paris Location of the lines reconstructed afterwards (no GPS) (+ PlaceEngine?) No public information about noise in the subway (private data owned by the RATP company) (2008)

Empowerment - case study 2: Mumbai, India (2009) Collaboration with NGO Awaaz Foundation organizing a campaign to raise awareness and map Mumbai, (end of september) (need to adapt the application to the indian phone market) 3rd most noisy city in the world AFP news

Empowerment - case study 2: Mumbai, India (2009)

Collaboration with NGO Awaaz Foundation organizing a campaign to raise awareness and map Mumbai, (end of september)

(need to adapt the application to the indian phone market)

vs Make it real (for real world experimentation) No need new device: Recycling phone 1st project accessible to the public (June) 1st subway noise map Data collected Personal and geolocated exposure data (epidemiological studies) Human-based sensing Visualization & Sharing Data aggregation by urban element (future) pollution information spreaded via social networks (twitter) Related projects for air pollution « sensing atmosphere (taxi driver) Berkeley, 2007 Noisetube “ Mobile Urban Sensing project” Cambridge, 2008 « street sweeper », Berkeley, 2008 Conclusion

Make it real (for real world experimentation)

No need new device: Recycling phone

1st project accessible to the public (June)

1st subway noise map

Data collected

Personal and geolocated exposure data (epidemiological studies)

Human-based sensing

Visualization & Sharing

Data aggregation by urban element

(future) pollution information spreaded via social networks (twitter)

Thanks for your attention Any question? Nicolas Maisonneuve www.noisetube.net

Thanks for your attention

Any question?

Nicolas Maisonneuve