W3C

Automotive Working Group Teleconference

04 Feb 2020

Agenda

Attendees

Present
Peter, Adnan, Ulf, Ted, Magnus, Gunnar, Glenn, Harjot, Benjamin
Regrets
Chair
Peter
Scribe
Ted

Contents


Gen2 access restriction tagging of signals (issue 322)

Ulf: I haven't received much feedback since proposing this
... some expressed interest in exploring other options for access restriction including WAMP
... we can discuss this and inform that separate study
... I started off with requirements for what we need
... it should be possible to read/write without a token (no auth) for some small number of nodes, if the implementer wishes
... it is possible to require access for write but read wide open
... and other nodes require a token for both read and write

Peter: will all be tagged?

Ulf: all individual nodes, some or branch and inheritance for them

Ted: I wonder if want to have granular read, whether for instance if you can share that data outside the vehicle for instance

Ulf: hard to control what someone does with the data after

Ted: you can spot bad behavior and have recourse

Ulf: since you are obtaining the tokens from authorization server in the cloud

Ted: I'll write something on the issue thread describing this

Ulf: not so easy to control
... flexibility comes with complexity
... simpler often better in security
... different clients will be provided different parts of the tree
... you can provide different client apps different data views
... please add your ideas to this issue thread
... tags can be added in deployment phase, not in VSS YAML but added after by OEM
... i chose 'validate' as the attribute name for the tags
... we are using JSON web tokens
... we have three possibilities, no tag, read or read/write
... a resource inherits from its parents unless otherwise specified

Adnan: what is the benefit for adding this?

Ulf: a client would obtain explicit approval from authorization server

Adnan: you said some can be accessed without a tag

Ulf: the guard would be able to check is there is a tag and enforce it, no tag means it will be served

Adnan: let's look at speed, and let's say it is read
... what does that change in this process?

Ulf: in case of sensor, read is the only option

Adnan: I understand the token part but if this reinventing the Oauth method

Ulf: this is using Oauth

Adnan: tags are not part of Oauth

Ulf: no but it allows extensions
... this is also influenced from what we inherit from ViWi
... in ViWi everything is read/write protected

Adnan: if we have five clients, each with different signal access rules
... for door lock status, read to get state and write to unlock for remote access
... how do we differential with tags?

Ulf: the authorization server will set the access per client

Adnan: still not seeing the benefit, maybe an illustration for a given leaf and different clients

Ted: i can see scenarios will a given app needs to be able to handle different access across vehicle manufacturers or perhaps in same vehicle if eg valet instead of owner. tags would be part of layering topic Gunnar will hopefully lead us on next week

Ulf: other thing I wanted to discuss is authorizationLevel 0-9
... this is put in tree as validate is
... added as access token in field, when client goes to authorization server in cloud it can also request higher authorization level
... 0 being the bare minimum signals generally made available whereas 9 is highest including eg engine data
... this proposal meets our requirements and alternatives being looked at should do the same

[Resume in two weeks after layering discussion]

F2F agenda

Ted: I'll open registration where people can make suggestions

https://www.w3.org/auto/wg/wiki/Auto-f2f-mar-2020

VSS Collaboration project proposal

Glenn: I have a conceptual idea on how to fast track this project
... the objective is to have vehicle data in VSS format on a public server so we can do demonstrations and reference projects
... we have a dog food program at Geotab where employees give permission to use their data for research projects
... we can extend that so some Geotab employees will be able to provide data for this W3C project. vehicle manufacturers involved in this group are welcome to do the same
... Harjot worked on a specific consent draft with our legal team and can provide it for review
... we can probably have a document in a week or two that can satisfy this narrow consent

Ted: happy to review it and get additional eyes from our attorney

Glenn: I will send to W3C Members

Peter: I would need to see how we would feel about it

Glenn: participants in this group would be able to participate and provide data under these terms with our Go device

https://www.w3.org/auto/wg/wiki/Vss_data

Ted: we can perhaps provide additional motivation in this doc so it will make sense to other readers

Harjot: that hasn't been done yet

Glenn: I am hearing encouragement and look forward to comments
... there was considerable excitement at our previous f2f to make this available
... this early step has been a bit of a legal hurdle
... we need a standard for anonymizing and see that as highly customized
... Nicole has provided me 8-10 documents on aggregation and anonymizing protocols. I will review and share pertinent subset
... we will use ourselves on this server at MIT
... we can perhaps start on dummy data in graph server at f2f

Peter: we should be able to demo the Gen2 open source project as well at f2f

[adjourned]

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2020/02/04 20:08:52 $