<scribe> scribenick: ted
<scribe> Scribe: Ted
ISO TC204 ITS diagnostics data flow review request
Ted: while this is a plausible flow another maybe more likely than per service incident is owner provides consent once to regularly share data with service provider. not sure whether inquiries to vehicle for additional data points will be permitted from independent service centers
Ulf: do we have support in VSS
for diagnostic data?
... we don't have much, is it sufficient?
MagnusF: is handling diagnostic routines in or out of scope?
Ted: to Ulf's planned to improve diagnostics in VSS, maybe via RPC for MagnusF's question?
MagnusF: they are highly specialized and would be manufacturer specific but to the extent we can generalize
Gunnar: I would start first as
whether we can handle with a regular node and not have to
create additional features
... it can be proprietary branch
MagnusF: reseting DTC
required
... it is a very interactive setup, we should be able to decode
DTC codes and DID data
Gunnar: this can fall into RPC service catalog
MagnusF: it would be a combination of both
Gunnar: agree, want to use VSS as far as possible
MagnusF: sometimes you want to cascade interactions as in this scenario
Gunnar: Franca in YAML style could achieve what you want
MagnusF: I think we have a path
forward
... agree there will be other flows, legislation in US (right to repair) at least
may require access for independent garages
Gunnar: it can considered an example
Ted: thanks all, I'll relay:
* potential for alternate flows in addition to one in diagram,
unlikely to be single, common solution but that could be
influenced by regulators
* diagnostics improvements pending in VSS data model pending in
GENIVI after EV
* diagnostics are usually very vehicle specific, clearing codes
and calling specific routines not simply querying signals
* we have proposed spec work on RPC from JLR and we will include
diagnostics use cases for RPC function catalog work
Per message compression for Web Sockets, alt schemes
Ulf: we were discussing cost of
data transmission and they see uncomperssed JSON as too
expensive
... their use case was wanting to follow a car with high
frequency lat/long and speed
... we should support some compression
... Ted mentioned existing solution in HTTP
MagnusF: I think it should be on
the link layer instead of the protocol itself
... won't work for video data. solution is to do on-board
processing/edge computing that pre-processes data and sends to
backend server
... we manage to compress long/lat down to 1.8 bytes with 10m
resolution
... lower volume/high quality is solution
Ulf: that is absolutely one
solution, you're saying outside of Gen2
... that is one way of doing it and do not think it would hurt
that much
... it exists in most libraries for HTTP
MagnusF: if we don't need to expand our specification extensively
MagnusG: what about Web Sockets? for REST is one thing
Ulf: it has support also
... deflate is also supported in WS, maybe not as ubiquitous
and standard
MagnusG: negotiation on whether it is used will be part of the handshake (content negotiation)
Ulf: sounds like plausible and not dead
MagnusG: if it is in standard HTTP do we need to do anything?
Ted: @@sampling and curve
algorithm - can find a Youtube video from Neil
... HTTP has long supported compression although not used as
widely, offset/tradeoff of increased client and server cpu to
save bandwidth
... we use it for some of our more populare DTD and other
schemata, seldom updated (years) and many millions of
requests/day (also do tarpitting and auto-blocking
logic...)
... @@simple sentence wrt conneg and server may support
MagnusF: I think this needs to be optional. most devices in the car will be underpowered
Ted: Peter relayed they are using gzip in mqtt for vss data
Adnan: for third parties we are
looking at ExtVeh and data points requested
... you can do a search on 'BMW data catalog'
... we are looking at graphql which is showing interesting
results
... some are looking into mqtt
Ted: graphql more on cloud end?
Adnan: yes but aware it is being used/considered elsewhere
Gunnar: I can confirm
... I get pointed to different people for some questions :)
Adnan: we are transparent on what we are doing in GENIVI
Gunnar: mapping to mqtt seems an obvious choice
Adnan: from AutoSAR point of view some are looking at SomeIP...
Gunnar: we have done some initial
analysis and would like to be part of that discussion
... we have an initial implementation of VSS in SomeIP
Adnan: SomeIP story is evolving
Gunnar: scope creep!
... CCS is very similar to ExtVeh
... filling in some details
Ulf: this topic keeps coming up,
PII should not be brought up to the cloud
... some static is easy like VIN to hash
... long/lat is harder. if you de-identify this data you have
two options, you can modify the position from where it actually
was and/or modify the temporal aspect
... if you do that you loose quite a bit of value and eliminate
a number of possible use cases, eg analyzing traffic
congestion
... wondering about another solution and discussed with some
colleagues who see it as feasible
... instead of moving vehicle data up to common repository, we
keep the data in identifiable, raw data at each OEM in
VSS
... then you can make aggregate queries (eg graphql) and
thereby de-identified
... the aggregated results represents multiple vehicles and not
individual ones
... our gut feeling is it would be possible to do this
... de-identification only applied to the aggregated
results
... this may be useful in our graph project and GENIVI CCS
Ted: offboard and protect PII,
use it for calculations and then de-id
... various schemes on [more] safely storing encrypted data
https://www.w3.org/2019/09/13-w3ctransport-minutes#item08
Ulf: yes, need to have access to the raw data in sandbox and before exposing de-identify PII
[adjourned]
This is scribe.perl Revision of Date Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) Succeeded: s/Sebastian/Sebastian_Schildt/ Present: Ted MagnusG Ulf Glenn Thomas_Stimm Adnan Gunnar MagnusF Sebastian_Schildt Found ScribeNick: ted Found Scribe: Ted Inferring ScribeNick: ted WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth WARNING: No date found! Assuming today. (Hint: Specify the W3C IRC log URL, and the date will be determined from that.) Or specify the date like this: <dbooth> Date: 12 Sep 2002 People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option. WARNING: IRC log location not specified! (You can ignore this warning if you do not want the generated minutes to contain a link to the original IRC log.)[End of scribe.perl diagnostic output]