Minutes IETF103: lwig
Light-Weight Implementation Guidance
||Minutes IETF103: lwig
Wednesday November 7, 2018
LWIG Session at IETF103
Room: Meeting 2
Time: 11:20 - 12:20
Chairs: Mohit Sethi, Zhen Cao (absent)
Note takers: Lijo Thomas, Ari Keränen
Co-chair updates the working group status since IETF102
Suresh, AD : lwig cellular-6 , normative reference and informational reference
to be updated , moving to next level Co-chair : Needs to push at the earliest ,
and let the authors decide the progress Ari: as co-author, can keep it wait for
more time, make sense because RD last call is soon started, if it goes longer,
wait and check the status and update the references if needed. Peter : not
anymore pessimistic about RD, interop coming so that interoperability issues
can be addressed
Co-chair : presented the agenda for todays meeting
# Lwig-tcp-constrained node-netowrks -04 (Carles)
Ilpo Järvinen: will read the draft and come up with more comments
Markku : will review the draft, single MSS issue if delayd ACKs off. When the
data packet arrives, immediately ack , needs to be more clear, unnecessary ACK
needs to fixed. With request / response traffic you immediately ack and send
response. With delayed ACK would piggyback. Not clear if it is advisable to
turn it off. Removed OpenWSN since mostly based on MicroTCP. Rahul : RIOT
option. Hard to convince my people that lwIP can fit to 40kB. Is it possible to
push the references to open source with lwig github Carles : Need to check the
sources and get back Rahul: people are using this now for comparison. Need to
have good sources. Not sure if LWIG github is the right place. Mohit: for the
crypto comparison draft got lots of questions on how are these compiled to
code. configuration for open source so that more people can use and check , a
personal github, or lwig github Carles: some data corresponds to data from
different platforms. Difficult to have fair comparison. Can try to make sure
that references and sources are reflecting what is intended. The code sizes are
not measured by us. From academic papers, from RIOT developers, etc. Have
problem with heterogeneity. Mohit : as per the reference , lwip 2.0 all
protocol stack on embed, how you verify the 40 KB , is it more reliable source
? Carles: we didn't do measurements but based on information from other
references. Carles : WGLC ? MOhit : markku, ilpo and rahul comments to be
addressed and wait for next revision to go for WG LC
# Neighbor management policy (Rahul Jadhav)
Rahul : Neighbor management policy presentation, how do u manage a neighbor
cache in a 6lowpan, reservation and signaling recommendation for neighbor cache
is addressed in the draft, and establish a stable network, Mohit: dont u have
the neighbor cache entry with signal entry received, a specific signal modeling
with your application, Rahul : have a application in place, it may be difficult
to do the signal based network cache, Mohit : not sure about 6lowpan problem,
but make it clear specify with your application and need not be generic
# Fragment forwarding vs per Hop reassembly presentation (Rahul Jadhav)
Carsten: random delay is application delay?
Rahul: in addition to application delay added a random delay
Carsten: what causes the numbers for MAC transmit failures?
Rahul: lots of failures with fragment forwarding; we need to check what causes
the failure in more granular level Carsten: interesting result, requires some
diagnostic work to find out where it is there Mohit: hope to see a presentation
on that next time
# Security classes for Software updates for IoT (Pascal Urien)
18 different classes identified based on set of available security
functionality (available bootloader, OTP, tamper-resistant keys, etc.).
Sometimes easy to re-flash bootloader. That will result in less strong security
than one-time programmable memory. If there's no permanent memory on device,
that can be re-flashed in supply chain process, resulting in attacks.
No way today to securely compute keys by software only. Need tamper resistant
(ran out of time; no time for Q&A)
# Alternative Elliptic Curve Representations (Rene Struik)
Rene presented current status (slide #4).
Draft is actually very simple, most of it is tutorial style to the topic.
Looking forward to more feedback. Also implementation experiences.
For reviews look at mostly the non-appendix parts of the draft.
Mohit: need to close meeting, looking for group to provide feedback.
Carsten: thank you Rene (for doing this)!
Mohit: need a few more reviews to progress the document.