Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Semantic Data Layers in Air Quality Monitoring for Smarter Cities

1,651 views

Published on

OpenSense Air Pollution Exposure and Sensing

Published in: Internet

Semantic Data Layers in Air Quality Monitoring for Smarter Cities

  1. 1. Semantic Data Layers in Air Quality Monitoring for Smarter Cities Jean-Paul Calbimonte, Julien Eberle and Karl Aberer LSIR EPFL S4SC Workshop 2015. International Semantic Web Conference ISWC Bethlehem, PA, October 2015 @jpcik
  2. 2. 5% (2.6 million) of all deaths are caused by urban air pollution Global Health Risks, WHO 2014
  3. 3. Health studies show that air pollution increases the risk of cardiovascular mortality (heart attacks) by 5% to 20% at least Health studies have shown the link between pollution and cardiovascular mortality Cardiopulmonary Diseases Cardiovascular Diseases Brook et al, Circulation 2010 Ischemic Heart Diseases
  4. 4. cardiovascular & respiratory morbidity negative effects on nervous system COcarbon monoxide respiratory morbidity airway hyper responsiveness harmful to living organisms decreased lung function lung inflammation Urbain Air Pollutants NOx nitrogendioxide monoxide O3 ozonePMparticulate matter aggravation pulmonary & cardiovascular condition
  5. 5. Air Pollution in a Smart City • Collaborative acquisition • Self-diagnosis • Localized measurements • Heterogeneous data sources • City Health Studies • Open Air Quality Index • Citizen Science • Data Privacy
  6. 6. OpenSense II Sensing the air we breath http://opensense.epfl.ch/
  7. 7. OpenSense2 global concern highly location-dependent time-dependent Crowdsourcing High-Resolution Air Quality Sensing Air Pollution Accurate location-dependent and real-time information on air pollution is needed Integrated air quality measurement platform  Heterogeneous devices and data  Human activity assessment, lifestyle and health data • Link high-quality and low-quality data • Integration of pure statistical models and physical dispersion models • Better coverage through crowdsensing • Incentives for crowd data provision • Finer temporal and spatial resolutions • Utilitarian approach for trade-off between model complexity, privacy and accuracy • Higher accuracy of pollution maps models http://opensense.epfl.ch Institutional stations OpenSense infrastructure Personal mobile sensors CrowdSense
  8. 8. 2 km Governmental station Our measurements Zürich: 10 sensor nodes updated: O3, NO2, CO, UFP, GSM, GPS
  9. 9. EMPA/Decentlab: 2 sensor nodes (Aircubes) : 2x O3, 3x NO2EMPA/Decentlab: comparing Aircubes with reference
  10. 10. Lausanne Deployment Particle sampling module • Ultrafine particle measurements using Naneos Partector • Measures directly lung- deposited surface area Gas sampling module • CO, NO2, O3, CO2, temperature & relative humidity • Hybrid active sniffer/closed chamber sampling operation • Enables absolute concentration mobile measurements Enhanced localization & logger • mounted inside bus • Fused GPS, gyro and vehicle speedpulses • Accurate sample geolocation even in difficult urban landscapes • GPRS communication
  11. 11. Beyond Public Transportation: C0
  12. 12. Crowdsensing • Participatory Sensing • Data Aggregation • Data privacy / utility • Combine with sensing infrastructure Use smartphones to automatically gather the contextual information - location, activity and environment - Data is aggregated anonymously and fused with other sources
  13. 13. Crowdsensing
  14. 14. Reference station Crowd sensing Public transportation Raw Data Acquisition Air Pollutants Time Series Temporal Spatial Aggregations Pollution Maps Pollution Models Air Quality recommendations Health Studies Air Quality Products & Applications From Sensing to Actionable Data
  15. 15. Localization: GNSS fusioned with odometry GPRS • packet parser • system logging • database server • GPS interpolation • advanced filtering • fault detection • system health monitor • automatic reporting 10busesinLausanne CO, NO2, O3, CO2, UFP, temperature, humidity The Lausanne Deployment
  16. 16. GSN: Global Sensor Networks https://github.com/LSIR/gsn
  17. 17. GSN: Accessing the Data 18 • Public Data Access • Online Processing • Basic Plotting/Filtering • Further Processing
  18. 18. Data Formats: CSV on the Web 19 { "name": "type_event","virtual": true, "aboutUrl": "#obs-{_row}", "propertyUrl": "rdf:type", "valueUrl": "opensense:CO_Observation" }, { "name": "unit","virtual": true, "aboutUrl": "#obs-{_row}", "propertyUrl": "qu:unit", "valueUrl": "unit:mgm3" }] } } { "@context": ["http://www.w3.org/ns/csvw", {"@language": "en"}], "url": "opensense.csv", "tableSchema": { "columns": [{ "name": "time","titles": "Time", "aboutUrl": "#obs-{_row}", "propertyUrl": "ssn:observationResultTime" "datatype": {"base": "datetime","format": "yyyy-MM- ddTHH:mm" }, }, { "name": "station","titles": "Bus sensor", "aboutUrl": "#obs-{_row}", "propertyUrl": "ssn:observedBy" }, { "name": "co","titles": "CO concentration", "aboutUrl": "#obs-{_row}", "propertyUrl": "ssn:observationResult" } http://www.w3.org/2013/csvw/
  19. 19. Example: SSN Ontology 20
  20. 20. SSN Ontology 21 ssn:Sensor ssn:Platform ssn:FeatureOfInterest ssn:Deployment ssn:Property cf-prop:air_temperature ssn:observes ssn:onPlatform dul:Place dul:hasLocation ssn:SensingDevicessn:inDeployment ssn:MeasurementCapability ssn:MeasurementProperty geo:lat, geo:lng xsd:double ssn:hasMeasurementProperty ssn:Accuracy ssn:ofFeature aws:TemperatureSensor aws:Thermistor ssn:Latency dim:Temperature qu:QuantityKind cf-prop:soil_temperature cf-feat:Wind cf-feat:Surface cf-feat:Medium cf-feat:air cf-feat:soil dim:VelocityOrSpeed cf-prop:wind_speed cf-prop:rainfall_rate aws:CapacitiveBead … … …
  21. 21. t1 t2 t3 t4 aggregated Provenance Raw data Spatio-temporal aggregation Event Annotation Temporal Segmentation Spatial Processing Data Granularity & Annotation
  22. 22. From Activity Recognition To Exposure • Accelerometer / Location Data • Online Activity Detection • Online Breathing / intake estimation • Estimation of Air Pollutant Exposure • Personalized recommendations
  23. 23. TinyGSN 24 • Android background application • Front-end to change parameters and select the sensors to use. • Based on the same principles as GSN: • wrapper, • virtual sensors • streamElements. • Scheduler, optimized for gathering continuous location without depleting the battery, • Manage Android Services and Alarms • Allow the device to sleep between the measurements.
  24. 24. Goal: estimate the health effects of long-term exposure to air pollution Health Studies in OpenSense
  25. 25. Health Studies in OpenSense
  26. 26. Combine with Pollution Models
  27. 27. Data filtering & calibration Data validation LUR model Pollution map Raw data Processing steps: Map validation 2 km Governmental station Our measurements From raw measurements to fine-grained pollution maps Winter (January - March)Spring (April – June)Summer (July - September)Fall (October – December)
  28. 28. OpenSense: Air Pollution in Smart cities • Data quality and curation • Failure and Noise handling • Semantic Layers of Data for different purposes • Privacy protection • Air qualitymodels and data fusion • Incentives for participatory sensing • Personalized health recommendations
  29. 29. Muchas gracias! Jean-Paul Calbimonte LSIR EPFL @jpcik

×