Capture & Process
The capture and process flow is essential for a PoB generation, turning sensors data in an anonymized secured dataset.
Capture
Several sources of data are useful to capture and qualify a movement:
- Fused Location:
- Precise location (<10m): GPS
- Coarse location:
- Wi-Fi
- Cellular triangulation
- Motion activity:
- Accelerometer
- Gyroscope
- Barometer
- Battery
The fused location with additional motion activity and battery enable us to detect a trip and its transition but also classify it into transport type more easily.
Privacy guardian
The privacy guardian is an elegant way of protecting the traces' anonymity before publishing them.
It converts positions to instant speeds, losing the starting and end point of the trip e.g. home to work. The data will only contain the number of kilometers made (quantity), when and at which speed to enable transport type classification.
This process can be done locally on the traveller device.
Trust bonus
EcoMobiCoin aims to be universal and easily accessible but without compromising security. We can imagine that a malicious person would want to fake positions and get rewarded for it.
There are different approaches to counter this kind of behavior such as the ones exploring in the following paper:
Stefan Saroiu and Alec Wolman. 2010. I am a sensor, and I approve this message. In Proceedings of the Eleventh Workshop on Mobile Computing Systems & Applications (HotMobile '10). Association for Computing Machinery, New York, NY, USA, 37–42. https://doi.org/10.1145/1734583.1734593
But it requires additional hardware or dedicated software for a trusted environment.
We choose not to make it mandatory but give a trust bonus to people who do so by providing highly secured data without excluding the ones that cannot afford it.
=> More details will be decided on this later on
Trip Transition detector
The trip transition detector allow people to do multi-modal trip and handle it as smaller multiple proofs of behavior rather than an noisy big record.
For example, one can walk from his/her home to the bike station, take his/her bike for 5km and then walk 2 km to the office. The trip transition detector will split this trip in 3 which will at the end enable the generation of 3 proofs (2 with a walking transport type and 1 with a biking transport type).
It leverages all sensors data to detect transport type change and split the steps of the trip.
Process
Once the data is collected, split and anonymized it can be process and classified into transport type.
Classifier
The classifier compute the movement data and outputs a probability vector that will be used in the rewards calculation.
Signature
A classifier can sign its output to certify some properties: version...
Fraud detection
It will later on integrate a fraud detection mechanism to prevent the duplicate submission and cloned data.
References
https://www.nrel.gov/transportation/openpath.html