System Components
Running the PEIR application on your mobile phone records a location trace, which is a time-stamped measurement of your position. This data is collected once every 30 seconds. The software then automatically transfers this data to the PEIR servers, and it is used to generate feedback about your impact and exposure.
You can learn more about how we calculate impact and exposure here.
The PEIR application has several key components:
- Location trace collection:
Location traces are gathered from mobile phones using GPS,Cell Tower, and WiFi beaconing.
Trace correction and annotation: Where possible, the error prone, under-sampled location traces are corrected and annotated using techniques such as map matching with road network and building parcel data. Map matching is a technique we employ to determine if a PEIR user is on a freeway or not. This helps us to determine whether they are engaged in a driving activity. Our map matching techniques use geospatial queries to determine road segments within a distance of 100 meters from a given location point and find the nearest road that is a match. The system also attempts to determine if the user is on a freeway.
Activity and location classification: The corrected and annotated location traces are automatically classified (e.g., estimating time spent traveling by car vs. on foot for a given day) to provide a first level of refinement to the data for a given person on a given day. Our activity classification is used to infer if a user is ’still’, ‘walking’ or ‘driving’. Each data point has a speed value from GPS and a freeway attribute from the map matching module. Based on speed distribution and freeway information, we convert raw data points into six different observations states. We employ a Hidden Markov Model (HMM) to build the activity classifier.
Context lookup: Both the corrected, fine-grained location data and the classified data are used as input to web-based information sources on weather, road conditions, real-time traffic monitoring, aggregated driver behaviors, and zoning/planning data.
Exposure/Impact Calculation: Finally, the fine-grained, classified, and derived data are used as input to geospatial data sets and micro-environment models that in turn are used to provide an individual’s personalized estimates and documentation of localized impacts.
