A Middleware Solution for Democratizing Urban Data Sara Hachem Inria Paris-Rocquencourt Joint work with Valerie Issarny, Animesh Pathak, Vivien Mallet,

A Middleware Solution for Democratizing Urban DataSara HachemInria Paris-Rocquencourt

Joint work with Valerie Issarny, Animesh Pathak, Vivien Mallet, Rajiv Bhatia, Alexey Pozdnukhov

May 2, 2014

Data Democratization

- Leveraging a plethora of data sources

- Generating publicly available information about the

environment

- Allowing the cooperation of governments and citizens to

induce policy changes and actions for smarter and healthier

environments

2

Data Democratization: Why?

- 7 out of 10 people in cities by 2050

- Cities should evolve with evolving technologies for citizens’

well-being

- Isolated technocratic institutions to solve urban problems

- No holistic view of the problems and their solutions

- Citizens may have better insights

3

Data Democratization: How?

- Active participatory sensing to complement passive sensing

- Real time learning from streaming data

4

Middleware with hybrid sensing/actuation

Public urban knowledge for citizens and governments

Closing the feedback loop with citizen/government cooperation

But… Challenges remain…

- How to leverage the plethora of available sensors?

- How to assimilate data and produce significant city models?

- How to ensure citizen participation?

- How to integrate all the above in an urban middleware

solution?

5

The Urban Civics middleware

6

Insights

Physical Sensing

Social S

ensing

Ince

ntive

s

Incentives

Insights

Urban Sensing

- Static sensing-Widely available & highly accurate

Require high deployment costs in large city scales

- Mobile sensing-Cheaper but less accurate

-Can complement but not substitute static sensors

! Can have varying precisions according to context (e.g., in

pocket)

- Social sensing-Users’ own perspective

-Data tagging

-Automatic data extraction from social networks

! Can be very subjective

7

http://www.netatmo.com/

Data Assimilation

-Integrate observations from

various data sources with

mathematical simulation models

! New sensors may introduce low benefits in densely

deployed areas

Dynamically configure observation network to task optimal

sensors based on uncertainty reduction

! Manage qualitative data while accounting for subjective

assessments

Convert to quantitative values and compare to other sources

8

http://www.hzg.de

Participatory Sensing

-Proactive user involvement & citizen engagement in data

collection

! Depends on user participation rate and motivation

Provide incentives:

-Financial: e.g., redeemable goods

-Ego-centric: e.g., badges

-Altruistic: e.g., personal satisfaction

-Democratic: e.g., helping the community

9

Early Architecture

10

semanticprobabilistic

Machine learning

Noise: Source of Environmental Pollution

- By-product of urban transport, construction, etc.

- Adverse impacts on physical and mental health

- Environmental management challenge for smart cities

- Exploit Urban Civics to monitor noise using microphones

- Static noise meters, mobile phones, tablets, social networks, user-based input

11

Urban Environmental Use Cases

12

- Air pollution- Static sensors (e.g., NO2)- Mobile wearable sensors- Data assimilation

- Safety- Static/mobile cameras- Criminal reports- Aggregate social variables

Next Steps

13

- Implement the Urban Civics middleware- https://urbancivics.gforge.inria.fr

- Urban-scale Experiments for noise crowd-sensing- In cooperation with the San Francisco environment department

- Exploit outcome to inform further developments and

challenges to investigate

14

THANK YOU!

http://citylab.inria.fr