Data tf
Data TF
W3C Automotive task force on automotive big data
meets bi-weekly on Tuesdays
8 PST / 11 EST / 16 GMT / 17 CEST / 00 KST & JST
Draw list
Efforts to align or draw upon
- W3C VISS, RSI
- Sensoris
- ISO20078
- Neutral Vehicle
- Caruso
- Automotive Ontology Community Group at W3C (schema.org)
- EURECOM / BMW ontology based on VSS
- W3C Web of Things
Areas of exploration
Data sampling/edge computing
Base on standard API for accessing signals on-board (W3C VISS, RSI)
Data quality and degradation
- VISS may drop subscriptions or degrade frequency due to server load (discussed but not in spec)
- Intermittent connectivity
- Unable to control when IVI system is powered down, potential loss of cached data being analyzed before transmission
ACLs
- signals discovery, what can an application see? VISS getMetadata
Data contract
samping methodology (regular polling interval, Peucker algorithm etc) needs to be described and sent to the cloud
introductory conversation @tf call 2018-05-31
We are encouraged to look at W3C PROV for ideas.
Transmission
Group feels this should be out of the initial scope for this task force since there are many custom, one off solutions being used. Experiences should be gathered and shared, could lead to guidelines instead of standardization.
Sending data to the cloud
Compresssion and efficiency in sending off board to cloud
W3C EXI is used by OEMs and in millions of production vehicles.
An EXI solution provider, Agile Delta, who also contributed extensively to the EXI spec will be sharing their experiences in automotive to the tf in July 2018.
Data marketplace
Business considerations
The group feels this is out of scope although has potential for some aspects to be commonly defined and sent to cloud, eg cpu time consumed
- reasonable and non-discriminatory
- mechanism to assess costs associated with the collection of data and recuperation
- onboard cpu value for edge computing, look at cloud computing model - Jeremy
- bandwidth, 4/5G and leveraging home/business/other wifi hotspots
Ontology
demo server providing SPARQL endpoints
Privacy Policy
on-board versus cloud usage
Ted to invite input from W3C Privacy Interest Group and regulators suspect we might define guidelines instead of formal standards on privacy in general. consent capture does have potential for standardization
Signals sensitivity
develop privacy guidelines for which signals are considered sensitive, could be used to profile (fingerprint)
use cases outlining which information is sought by what types of third parties
signals OEM may be required to provide certain classes of applications as deemed by regulators
Data Governance
Consent model
capturing demonstrable and enforceable consent from vehicle owner and/or user on which signals can be shared with which third parties, see Caruso presentation at F2F and follow-up meetings 2018-06-14 2018-06-21
With the industry moving towards more usage based models, for consent and multitude of other use cases personal profiles stored in the cloud that can be accessed by vehicles after authentication for a trip may be needed.
ORDL Vocabulary may be relevant for representation of granular signal consent to third parties and to be able to communicate it to different data providers.
Retention
Duration data can be retained and handling of deletion requests.