►
From YouTube: LPWAN WG Interim Meeting, 2021-01-19
Description
LPWAN WG Interim Meeting, 2021-01-19
A
So
this
is
the
second
train
of
the
fp1
working
group
for
this
year.
2021
welcome
to
the
group.
This
is
an
official
ietf
meeting,
so
the
best
practices
apply
in
particular
best
practice
regarding
patents
and
ipr.
So
if
you're
aware
of
any
ipr
that
is
disclosed
discussed
today,
please
let
us
know
during
the
meeting
or
tell
the
chairs
at
the
end
of
the
meeting.
A
We
are
recording
the
meeting,
so
the
minutes
will
be
published.
They
will
be
available
to
all
so
we'll
so
recording
in
a
minute.
Links
will
be
given
on
the
minutes
with
the
minutes
of
the
meeting
on
the
mailing
list
and
the
attendance
will
be
recorded
from
the
webex
attendee
list.
So
if
you're
attending
sharing
a
desktop
with
somebody-
and
your
name
is
not
sharing
here-
please
let
us
know
and
go
to
the
notes
on
the
link
that
was
given
on
the
chat
and
add
your
name
to
the
blue
ship
list.
A
So
the
original
agenda
was
published
as
this,
we
actually
have
a
request
to
include
a
discussion
on
a
new
document
that
was
just
published,
which
is
a
proposed
rp1
architecture.
So,
as
you
remember,
there
are
a
number
of
times
where
people
said
it
would
be
very
nice
to
have
an
lp1
architecture.
A
So,
finally,
a
draft
zero
was
published
and
alexander
will
discuss
this
class
today,
so
we'll
insert
15
minutes
after
the
sheikh
of
honorawan
discussion
and
before
the
cooperage.
A
A
Lp1
architecture,
we
had
a
to-do
last
time
to
request
a
meeting
for
itf-110,
so
that
was
done,
and
so
please
review
the
way
we
asked
it.
So
basically,
we
asked
for
a
duration
of
one
hour.
A
A
And
then,
because
of
our
attendance
from
americas,
I
also
requested
to
make
it
late,
but
I
guess
that
every
group
with
an
important
population
from
america
will
also
ask
for
something
like
that.
That
won't
help
a
lot,
but
that's
the
way
I
asked
for
this
building
and
the
next
discussion
will
be
a
sheikh
of
our
of
our
one.
B
B
It's
not
easy
for
me
these
days,
but
I'm
still
working
on
it
from
time
to
time
and
I
do
not
send
the
email
regularly,
but
I'm
still
working
on
the
subject.
A
C
C
So
that
was
something
that
we,
as
as
co-chairs,
have
been
discussing
from
the
start
of
the
working
group,
saying
that
at
some
point
of
time,
we'll
need
to,
you
know,
give
this
big
picture
of
how
things
will
be
working
together,
and
you
know
we
we
end
up
publishing
rfc
8724,
so
we
have
a
lot
of
the
the
baseline
principles
and
now
we
think
it's
it's
long
due
to
to
have
this
this
draft
and
we
have
anna's
co-author,
who
has
also
contributed
a
lot
of
ideas
and
and
among
among
other
things.
C
So
that's
that's
really
the
the
first
version
and
of
course
it
will
evolve
over
the
time.
The
goal
of
the
document
is
so
what
we
what
we
have
today
is
we
have
this
the
generic
framework,
so
rfc
28724.
C
We
have
the
technology
specific
documents,
we
have
the
yang
data
model
coming
in
the
oem
and
we
other
other
ideas
on
the
table
like
how
do
we
do
shakeover
ppp?
So
these
are
like
give
different
bricks
and
from
the
co-op
draft
there
were
a
lot
of
a
lot
of
questions
about.
Okay,
how
these
or
how
do
these
all
these
things
work
together.
C
So
how
are
context
provisioned,
you
know,
and
and
during
the
discussions
with
so
there
were
olivia
and
we
who
had
some
questions
about
in
the
working
of
flora
one
you
know
how
do
we,
you
know
reset
some
parameters
or
how
do
we
re
recalculate,
the
the
the
id
and
error
handling
and
all
that
right
and
and
these
things
in
our
opinion,
they
need
to
be
handled
like
in
a
they
need.
They
need
to
be
a
store
right.
C
There
need
to
be
some
reference
architecture
in
which
all
these
questions
have
their
their
their
their
place
and
that
we
don't
have
like
half
an
answer
in
that
draft
and
half
an
answer
in
that
other
draft,
and
then
we
end
up
with
a
lot
of
technology
pieces
that
don't
end
up
to
something
working.
C
C
What
interfaces
are
there
and
and
what
what
protocols
can
we
can?
We
imagine
for
that
and
of
course
next
steps
would
be
you
know,
and
that
might
be
even
in
different
documents.
You
know
what
message
flows
and
and
all
that,
so
that
would
be
up
to
to
the
group
to
decide.
C
So
with
no
further
ado,
let's,
let's
go
to
the
first
element,
is
what
we
learned
from
the
shecos
core
and
the
co-op
and
all
the
discussions
with
the
with
the
isg
and
with,
and
we
think
the
need
to
go
further
in
the
security
considerations
right
well.
The
first
thing
is
that,
as
we've
always
been
saying,
shifts
can
be
on
different
levels.
C
So
there
is
the
shake
on
the
network
layer,
click
on
the
application
layer,
but
also
when
you
look
into
oscore,
there
is
like
actu
actually
in
the
application
layer
you
can
have
shake
on
like
on
the
outside
and
in
the
within
the
the
the
compressed
sorry
within
the
encrypted
payload.
C
Well,
that's
an
example
right
for
us
or
very
specific,
for
example,
we
can
apply
shake
before
the
comp
before
the
encryption
and
after
the
encryption,
so
we
have
chic
on
all
these
all
these
places
and
so
that
architecture,
the
way
we
provision
the
context
and
all
that
you
know
that
needs
to
it-
needs
to
fit
these
different
use
cases
right
or
or
we
may
decide.
Well,
you
know
for
one
well
for
one
sheet:
we'll
have
one
architecture
from
the
other
will
have
different
architecture
right,
but
we
need
to
to
spell
this.
C
One
out
also
also
don't
hesitate
to
interrupt
me
if
you
have
any
questions
right.
D
So
one
question
here:
terminology:
is
there
a
difference
between
uppercase,
chick
and
lowercase
shift,
like
you
have
the
the
lowercase
chick
at
the
network
layer
and
upload
this
sheet
of
the
application
here.
C
That
was
taken
from
those
chord
drafts,
so
for
me,
that's,
but
maybe
runner
has
input
on
that.
E
D
Okay,
so
in
this
case,
for
this
presentation
and
architect,
the
overall
architecture,
they
are
now
there's
no
fundamental
pieces.
G
A
It's
not
necessarily
usual
for
people
to
see
lowercase
and
uppercase
mixed
meaning
different
sessions.
So
I'd
suggest
we
homogenize
to
classical
uppercase.
F
C
F
H
Thank
you.
I
have
another
comment
because
I
I
see
this
kind
of
the
the
the
discussion.
H
You
could
have
encryption
before
tsl
and
then
after
tsl
also
have
sick.
So
then,
I
think
that's
the
same
principle,
so
I
don't
think
it
is
only
about
those
quota
is
in
general.
Whatever
you
have
the
encryption,
you
could
have
this
situation
as
well.
C
Yeah,
yes,
yes,
definitely
so
a
good
good
point!
Edgar!
Thank
you
for
for
this
one,
so
we
can
include
it
as
a
use
case
right
to
to
say,
okay
well
what
what
are
the
use
cases
that
need
to
be
covered
by
this
architecture?
Is
it
better
like.
C
Okay,
oh
moved
closer
to
the
wi-fi,
so
so
we
have
right
this
these
different
instances.
So
the
first
thing
is
basically
the
boundaries
are
not
explicitly
on
the
osi
layers
right.
We
can
have
one
cheek
instance
that
is,
for
example,
compressing
ip
and
udp,
plus
the
fragmentation,
then
another
instance
working
on
oscar
outer
and
another
in
oscar
inner.
So
you
see
that
there
could
be
completely
different
distances
on
on
the
same
level
on
the
same
layer
from
the
osi
perspective,
or
it
could
be
the
same.
C
Shake
instance
doing
all
that
right
so
that
that's,
but
let's
say
that
okay,
they
can
be
under
different
layers,
and
each
of
them
have
potentially
a
set
of
rules
that
are
different
and
they
can
could
be
managed
by
different
entities.
So,
typically,
if
we
have
the
network
there,
it
can
be
managed
by
one
by
one
entity
with
one
set
of
rules
and
the
the
application
layer
check
can
be
managed
and
there
could
be
rules
managed
by
handled
by
different
entities.
C
So
you
know
that's
something
that
I
think
that
we
need
to
have
in
mind
when
we,
when
we
design
the,
when
we
define
the
architecture
and
let's
start
really
simple
right,
let's
start
really
simple
and
say:
okay,
what
are
the
entities?
Well,
we
have
the
the
device
side
and
then
we
have
the
the
core
network
side.
Here,
I'm
not.
C
We
are
not
talking
specifically
about
you,
know
this,
which
gateway
on
which
level
right,
we
just
say,
okay.
Well,
there
is
the
device.
There
is
the
the
core,
the
application,
and
then
you
have
the
compression
and
fragmentation
and
reassembly
and
and
and
they
compared
and
there
are
the
set
of
rules.
So
the
r
is
the
set
of
rules
and-
and
you
know
like
the
compressors
and
like
the
check
instances,
are
reading
the
rules
from
the
from
the
rule
set.
C
Oh
yeah,
so
that's
a
good
question.
It's
just
because
that's
like
a
very
high
level
representation,
so
it
it's
this.
Basically
that
could
that
should
be
considered,
or
so
it's
the
device
talking
to
the
core
like
to
the
network
gateway
or
it
could
be
the
application
on
the
device.
So
let's
say
the
the
os
core,
the
co-op
score
talking
to
the
application
that
is
in
the
cloud.
C
So
probably
you
know
it's
for
lack
of
just
saying
a
and
b
like
at
least
like
the
two
end
points
talking
to
each
other
that
be
on
a
different.
Let's
see,
layers,
and
so
questions
are
like
how
do
these
two
discover
each
other?
How
do
they
stay
in
think?
You.
H
C
So
these
are
things
that
we
need
to
answer
in
this
architecture
that
the
architecture
needs
to
to
to
cover
right,
that
it
needs
to
to
make
them
to
make
it
possible.
C
Now,
if
we
do
a
a
zoom
on
that,
we
can
say
that
the
the
minimal
and
very
simple
way
of
actually
doing
that
is
when
we
have
the
the
compressor
the
sheet
compressor,
the
compressor
in
the
lower.
Do
you
see
my
pointer
actually
if
I
move
my
pointer
like
this.
C
Okay,
let
me
highlight
her
okay,
so
here
the
of
course
the
the
the
this
is,
the
shape
of
the
compressor
right.
It
reads
the
rules
from
the
from
the
rule
set,
and
then
there
is
this
entity
that
we
may
say
that's
a
rule
manager
and
that
rule
manager,
you
know,
has
it
it.
It
is
the
one
that
this
entity
that
goes
and
and
reads
and
creates
and
updates
or
deletes
rules
from
the
rule
set
that
are
that
are
in
the
in
the
rule,
store
and
full
manager.
C
Then
can
it
can
have
these
interface,
this
network
interface
or
this
management
interface,
that
it
supports
this
management
protocol,
and
here
the
the
most
natural
way
of
looking
at
that
is
using
core
conf,
which
is
of
the
family
of
net
conf
and
and
restaurants.
C
So
this
protocol
is
used
to
communicate
with
the
rule
manager
and
then
the
room
manager,
you
know,
performs
the
operations
to
write
in
the
google
door
and
the
the
the
shake
instance.
You
know
just
consult.
C
It
so
the
rule
manager
in
that
point.
In
that
perspective,
so
two
rule
managers.
They
they
work
in
pairs.
So
if
we
say
that
the
device
and
the
core
or
device
in
in
the
in
the
cloud
you
know
there
are
rule
managers
on
both
ends
and
high
level
requirements
or
high
level
design
choices,
are
you
know
we?
We
have
the
the
the
shake
model
representation
done
in
yang
in
the
protocol
from
the
protocol
selection,
we
can
have
many
different
protocols.
C
The
core
conf
is
the
one
that
seems
like
the
native
one,
because
it's
young
native
it
uses
co-op.
It
uses
a
c-bor,
so
it's
really
quite
efficient,
but
there
be
other
there
could
be
others
like,
as
we
have
the
example
with
the
pvp
draft
there.
You
know
we
have
the
the
the
the
protocol
for
session
negotiation
that
that's
used
between
the
two
entities
that
could
be
looked
at
as
as
rule
session
managers
like
rule
managers.
C
So
from
from
that
perspective
you
know
we
have
the
rule
managers.
We
have
the
interfaces
that
the
two
of
them
you
know,
communicate
and
then
going
a
little
bit
further
in
the
definition
of
that.
Of
that
architecture
is
rule
managers
there
are
they
have
ip
addresses
by
default,
of
course.
C
So
if
we
are
working
on
a
network
level
check,
these
can
be
a
local
ipvc
address
of
v6
addresses
or
if
we
are
looking
on
an
application
level
check,
then
that
can
be
global,
ipv6
addresses
and
in
that
way,
of
course
they
can
be.
You
know
they
can
communicate
over
ip
over
the
the
shaky
instances
and
just
a
short
note.
In
some
cases
these
could
be
collocated.
They
can
can
reuse
the
application
level
protocol.
C
So,
for
example,
if
you
have
a
application
level
shift
that
compresses
co-op
traffic,
you
know
for
some
co-op
applications
that
could
be.
You
know
that
there
could
be,
it
could
be,
the
the
rule,
manager
could
reuse
the
same
co-op
server,
and
so
it
can
have
like
a
logistical
endpoint
that
can
be
that
can
be
used
as
as
communication
between
the
two
and
on
the
secure
security
wise.
So
this
is
how
both
rule
managers
are
able
to
communicate
to
each
other
and
on
the
security
on
a
security
wise.
C
Well,
there
is
a
choice,
of
course,
but
we
can
simplify
the
things
and
say
well
on
the
network
level
chick.
If
the
layer
2
encryption
and
the
layer
2
security
is
deemed
sufficient,
then
maybe
we
don't
need
additional
security
or
you
know
we
can
say
well.
We
need
end-to-end
encryption
or
end-to-end
security
for
application.
As
for
application
level
shake
as
with
you
know,
oscar.
C
If
we
are
doing
between
in
core
conf,
it
could
be
core
conf
over
co-app
or
gtls,
or
it
can
be
core
con
or
score
collapse
so
that
that's
up
to
to
the
final
design.
C
It's
well
that's
about
the
it's
about
the
protocol.
Let's
say
it's
about
the
protocol
that
allows
the
two
rule:
managers
to
communicate:
okay,
so
to
secure
them
to
secure
that
traffic.
Okay,
then
there
could
be
additional
yeah,
but
here
is
it
this
part,
and-
and
in
that
last
point
I
think
I
already
mentioned
you
know
the
rule.
C
The
measure
protocol
goes
through
the
sheet
compression
and
the
compression
could
go
through
the
shake
compressing
the
compressor-
and
let
me
just
give
you
an
example,
so
this
is
an
example
of
a
network
level
shift.
So
here
we
have
both
ends.
Let's
say
that
the
left
hand
side
is
the
device
and
the
right
hand.
Side
right
is
the
core.
C
Here
we
have
the
sheet
compressors
and
the
compressors,
and
here
we
have
the
lpn
link,
so
the
application
is
talking
with
through
the
ip
stack
through
the
chic
instance
here
over
to
the
other
side
through
the
ip
stack
to
the
other
application
site.
Well,
that's
really
nice,
where
the
sheet
compressor
and
the
compressor
you
know,
uses
the
rules
that
are
in
the
rule
store
and
now
what
we
have
is
that
the
rule
managers
can
communicate
through
their
management
protocol.
C
For
example,
core
con
over
earth
core
can
communicate
to
each
other
and
that
communication
actually
could
go
through.
You
know,
as
a
as
I
said
before,
you
know
if
they
are
on
the
same,
so
if
the
network
level,
so
that
they
could
be
using
link
local
ipv6
addresses,
so
each
of
them
has
a
link
locally.
At
six
address,
these
ipv6
addresses
can
be
discovered,
can
be
provisioned
in
advance
can
be
defined.
You
know
to
be
generated
in
in
a
static
in
a
static
way.
C
You
know
I
mean
the
the
link
offices.
Of
course
they
are
known,
but
the
the
destination
port,
and
all
that
you
know
all
the
they
run.
The
viewpoint
can
be
fixed,
and
so
you
know
the
first
rule
manager
coming
x
over
the
ip
stack.
Then
it
goes
over
the
shake
instance
all
the
way
to
the
other
ip
stack
and
then
up
to
the
room
manager.
C
So
they
have
like
this
indication
the
protocol
exchange
between
the
two
rule
managers,
and
then
I
mean
core
conflicts
like
it's
like
netcom.
You
know
you
can
have
all
the
configuration
all
the
all
the
operations
that
are
necessary
there
and
then
the
rule
managers
can
perform
cred
operations
to
the
to
the
rule
stores
on
both
on
both
sides.
C
You
know
to
ensure
that
they're
they're
synchronized,
so
I
think
that's
so
that
that
these
are
a
couple
of
examples
and
next,
so
there
are
a
couple
of
points.
Next
there
are
some
we
need
to
find,
which
other
use
cases
we
would
like
to
to
cover
by
this
architecture.
C
And
and
then
we
can
go
into
like
the
the
different
stages
of
of
the
network,
message
flows
and
all
that
so
and
probably
others.
So
that's
in
a
couple
of
of
words.
C
And
yes,
if
you
have
any
any
other
questions
with
anna
and
pascal
we're
happy
to.
A
A
A
So
in
one
example,
which
is
the
device
which
is
shipping
with
you
know
all
the
rules,
provisioned
and
basically
the
device
number
or
something
will
give
you
what
rules
it
has,
and
maybe
it
will
not
communicate
for
a
long
time
with
its
old
manager.
A
Then
it
might
be
that
the
the
communication
on
the
left
is
not
synchronous
versus
the
communication
on
the
right
could
be
synchronous
with
the
room
manager
right,
but
it
might
be
that
the
room
edge
at
the
device
is
actually
done
early
right.
C
So,
communication
on
the
left
you
mean
the
rule.
F
A
C
Yeah
so
actually
yeah.
I
I
see
yeah,
because
I
added
this.
This
schema
the
last
moment
here.
Actually,
the
the
this
is
the
device.
C
C
Exactly
exactly,
and
and
yes,
yes-
and
this
is
where
this
is
where
we
were
discussing-
and
this
is
where
the
point
that
you
know
maybe
when
it
shows
up
the
device
it
can
send
an
operation
to
the
other
manager
to
say
hey.
Well,
you
know,
go,
do
a
dns
operation
or
you
know
fetch
the
the
the
rules
from
from
from
from
that
url.
A
I
can
see,
oh
so,
but
we
need
to
describe
that
because
otherwise
does
not
meet
the
eye.
The
other
thing
that
that
the
other
case,
which
is
yet
another
variation
of
what
we
see
here,
is
the
chic
of
rpp
draft.
H
B
A
The
compressor
will
find
somehow
a
url
url
of
the
room
manager
where
the
room
edger
are
positioned.
So
it's
not
really
the
room
manager,
it's
the
roof
itself
on
the
url.
So
so
there
will
be
a
relation
between
a
server
which
provides
the
rules
and
the
rule
managers
on
both
sides,
which
you
know
basically
the
left
side,
discovers
the
device.
D
A
Fetches
the
rules
from
the
url
from
the
rule
manager
and
then
fetches
the
rules
themselves
from
the
url,
and
then
it
tells
you
know,
as
it
creates
the
ppp
session
in
in
the
ppp
headers,
where
I
describe
the
compression
it
tells
where
the
url
is
for
the
compression
rules,
so
that
the
other
side
can
go
and
fetch
the
same
rules
at
the
same
url,
and
the
expectation
is
probably
that
your
hull
can
be
local.
For
instance,
if
you
have
a
control
loop,
you
know
inside
the
factory
network,
not
necessarily
connect
that
to
the
internet.
F
A
C
Rules?
Yes,
yes,
yes,
so
so,
basically,
that's
what
what
that
means
is
you
understand
correctly,
and
I
think
that
works
really
really
well
is
to
have
like
a
third
rule
manager
actually
configures.
Both
rule
managers
on
the
device
here,
on
the
left
hand,
side
and
on
the
core.
A
A
There
is
this
expectation
that
there
are
two
entities.
One
is
really
the
logic
which
says
you
know
you
pick
those
rules
and
the
other
piece
is
just
the
main
storage
of
these
rules.
Like
the
repository
of
the
books,
it's
kind
of
separate
and
the
cool
thing
about
having
your
house
for
the
rule.
Is
that
your
hand
that
you
know
it's
unique
and-
and
you
want
to
be
sure
that
both
aliens
will
use
the
same
data
for
compression
and
decompression,
and
they
don't,
you
know,
use
the
different
rules.
A
C
C
Exactly
exactly
so
so
yeah
I
mean
we
really
wanted
to
get
the
the
ball
rolling
because,
as
I
said,
we'll
be
talking
about
this-
and
we
got
a
lot
of
comments
from
the
for
the
co-op
shook
over
co-op
sorry,
co-op
oversight,
draft
that
were
basically
saying,
hey
guys.
How
do
you
do
that?
How
do
you
do
that?
C
And-
and
I
and
that's
right-
the
the
first
stone
of
this
one
and,
as
pascal
said,
you
know
covered
at
least
say
if
you
have
other
cases
or
you
know
how?
How
do
you
feel
about
about
this
approach?.
D
So
what
one
general
comment
I
would
have
is
that
now,
after
hearing
your
presentation,
it
cleared
my
mind,
I
was
a
little
confused
with
the
title
and
I'm
afraid
that
it
could
happen
to
other
people,
because
in
rfc
8724
the
generic
framework,
we
do
have
a
section
three
that
is
called
lt1
architecture.
D
So
in
this
case
a
sheet
architecture-
and
I
was
a
little
confused
about
what
exactly
we
were
saying,
because
we
had
already
presented
an
architecture.
Now,
it's
clear
to
me
that
it's
two
different
things,
but
probably
there
should
be
some
some
explanation
or
some
mapping
explaining
how
that
was
a
basic
arc
network
architecture,
and
this
one
is
expanding.
C
Oh
yes,
yes,
but
we'll
need
to
to
to
clarify
the
the
title
and
and,
as
I
said,
it's
a
first
version
and
the
point
was
to
to
get
the
discussion
going
and
to
say
to
see
what
you
know
is
it
going
in
the
right
direction
and.
C
A
It
it
really
depends
on
what
we
want
to
put
in
this
document,
but
I
believe
we
we
need
architecture
and
framework.
It's
usual
now,
I'm
seeing
that
in
many
places,
but
you
need
the
architecture
to
position.
The
objects
like
you
need
to
describe
what
this
rule
manager
is
about.
If
we
have
a
rule
store,
we
need
to
explain
what
it
is.
Then
we
need
to
position
which
object
talk
to
which
object
and
when
and
all
this
is
architecture.
A
Yeah,
yes,
and
then
once
we've
done
that
we
can
explain
those
things
in
more
details
which
could
be
framework.
But
there
is
an
architecture
piece,
that's
missing
and
I
was
talking
to
alexander
and
if
you
remember,
or
maybe
you've
seen
it.
Maybe
not.
We
published
a
document
in
the
ieee
and
both
ideotriple
unity
have
journal
pretty
much
interval
about
the
time
when
we
formed
this
working
group
and
we
actually
described
the
architecture
in
there.
G
A
I
believe
that
the
text
or
the
general
you
know
we
need
to
adapt
it
to
because
things
have
changed,
but
we
need
to
to
bring
that
back.
We
need
to
explain,
for
instance,
the
discussion
about
where
you
have
a
crypto:
you've
got
a
separation
of
endpoints
and
you've
got
separation
of
rules,
etc,
etc.
This
is
architecture.
This
is
not
framework.
A
So
we
we
need
that
and
it's
not
in
87
24
I
mean
that's,
that's
missing
and
that
really
caused.
You
know
a
lot
of
hassle,
in
particular
for
the
co-op
document
where
actually
it
was
one
ist
review
which
pointed
out
that
having
those
two
different
stages
of
check
was
a
major
architectural
point,
and
since
we
didn't
have
an
architecture,
it
didn't
show
anywhere.
D
A
A
Okay,
so
yes,
let's
call
it
if
you
don't
mind-
and
you
know
for
now,
it
doesn't
deserve
the
name,
but
let's
create
architecture
and
framework.
So
we
know
what
to
expect
to
do
in
there.
A
We
don't
want
to
have
two
documents
either
it's
very
hard
at
some
point
to
say:
oh,
this
is
purely
architecture
and
that's
purely
framework,
that's
very
hard,
and
we
don't
want
to
throw
two
documents
at
the
asg.
On
the
other
hand,
we
want
to
have
this
one
early
enough,
because
it's
already
too
late,
I
mean
we
already
suffered
in
some
reviews
because
we
did
not
have
it.
A
C
Yes,
yes,
and
and
and
thanks
also
edgar-
for
pointing
out
tls.
C
So
if
you
have
you
know,
we
are
really
welcome
to
having
other
use
cases
or
or
or
points
you
know
to
include,
and
the
point
would
be
and
also
have
like
the
in
this
document
has
at
least
pointers
or
to
to
give
the
big
picture
as
much
as
as
we
can,
of
course,
with
to
the
details,
but
at
least
the
big
picture
of
okay.
How
are
all
the
questions
that
we
that
we'll
have
for
for
some
time?
You
know
how
are
they
handled
and
when
a
device
goes
into
a
network?
C
Well,
you
know
there
are
certain
things
to
be
done
and
all
that
you
know
we
need
to.
We
need
to
to
position
right
now,
so
that
we
have
a
shared
vision
on
it.
A
I'm
sorry
since
we're
at
it
and
we're
talking
to
edgar
edgar.
Can
I
have
a
quick
question
to
you
because
I'm
happy
to
see
you,
but
we
we
kind
of
had
to
do
from
last
time
to
find
some
help
to
quote
and
quote
replace
you,
because
we
we
thought
that
you
were
no
more
with
us.
So
I'm
very
happy
to
see
you
and
I
wonder
if
you
you
plan
to
have
time
to
work
with
anna
on
the
5g
document
or
or
if
it
was
just
by
chance
that
we
had
you
today.
H
Well,
the
plan
is
that
I
I
will
try
to
help
bring
up
the
latest
things
missing.
Hopefully
I
I
can
continue.
I
it's
not
yet
fully
agree
with
my
manager,
but
then
I'm
also
looking
for
some
support
from
the
product
people
to
see
if
we
can
actually
raise
their
profile
of
shake
in
general,
so
maybe
try
to
see
if
we
can
bring
it
to
3gpp
and
so
on,
because
they
haven't
been
so
far
much
interest,
but
then
now
that
there
are
some
concrete
use
cases
for
it,
then
maybe
the
story
will
be
different.
G
G
A
A
Pretty
much
so
first,
I
mean
basically
spring
2022,
like
february.
G
A
Okay,
so
thank
you,
okay
and
so
I'm
placing
on
old,
actually.
D
Well,
I
think,
last
time
we
have,
we
have
mentioned
that
we
were
planning
to
or
we
were
targeting
to
do
last
call
spring
and
sorry
to
get
it.
You
get
it
finalized
by,
let's
say
more
releases,
yeah.
E
No,
in
fact
I
I
made
some
updates,
but
I
think
that
I
can
push
a
version
very
similar
to
this
one,
because
it's
quite
good-
and
I
need
comments
and
now
that
young
is
more,
is
popular
and
I
cannot
feedback
on
on
it
and.
A
E
Oh,
but
I
will
publish
this
week
or
next
week,
the
new
year,
the
version
four,
I
think,
of
the
document.
Okay,
so
I'm
moving
the
to-do
from
last.
A
Juan
carlos
is
done
for
2022,
okay,
the
last
to
do
was
to
to
request
the
meeting
for
atf
110
and
the
placement
as
conflict,
which
I
did
okay.
So
we
are,
we
are
done
with
the
to-do,
which
is
pretty
good.
A
C
No,
no,
it
was
it
was
that
so
we'll
be
working
on
the
architecture
document,
and
we
consider
this
to
be
to
be
our
our
job
as
a
as
chair
from
the
start
with
pascal-
and
the
point
would
be,
you
know
that
it's
it
takes
all
the
considerations
for
for
the
working
group.
So
please
don't
hesitate.
You
know
to
give
your
feedback
and
and
to
actively
propose
or
criticize
the
the
text
as
it
is,
and
and
yes
so
that
there
was
maybe
a
final
final
thing.
C
I
announced
that
on
the
mailing
list.
I
just
wanted
to
to
make
that
clear
out
here
as
well.
So
chair
hat
off,
of
course,
there
might
be
some
ipr
factio
in
in
part
of
the
section
of
that
document
and
of
course
you
could
get
further
on
it
will
be
under
the
standard,
acceptable,
itf
licensing
release.
A
G
G
I
don't
know
if
they
will
are
valid
for
what
we
are
doing,
because
we
are
not
doing
data
compression
as
deflate
or
sip
or
whatever.
We
are
doing
header
compression
with
the
logic
and
the
analysis
of
how
it's
going
on
the
data
in
the
header.
So.
G
I
It's
not
so
much
about
header
or
data
by
the
way.
It's
simply,
if
you're,
using
something
like
gzip
or
anything
which
is
this
compression,
then
you
are
open
to
something
because
you
exchange
dynamic
dictionaries,
kind
of
which
is
not
the
case
with
so
this
attack
does
not
fly
yeah,
so
it's
basically
conforting
the
message
that
yeah
this
attack.
That's
not
a
public.
G
Okay,
so
isn't
that
chic
is
not
as
just
sip
or
this
light.
It's
really
a
header
compression
looking
for
the
logic
of
what
we
are
sending
and
trying
to
reduce
redundancy,
but
never
never
trying
to
compress
data
for
the
packet
or
the
packet.
As
that.
H
A
A
But
but
fact
is
it's
much
better
that
you
have
published,
because
you
can
basically
send
him
in
your
response.
Gary
is
the
deaf
and
you
give
him
the
url
of
the
diff
or
something
you
say,
publish
17.,
and
that
was
just
for
you.
So
all
the
gifts
that
are
there
are
for
you,
something
and
and
then,
when
he
processes
that
that
would
be
a
lot
easier
and
faster
for
him,
because
he
has
to
reload
the
full
context
of
his
previous
review.
G
E
E
Okay,
just
just
one
point:
we
dominique,
I
don't
know
if
you
you
got
him
and
you
talk
about
that,
but
we
we
are
working
and
we
fold
the
cover
on
the
aom
draft.
A
A
Nice,
okay,
actually
dominic
will
listen
to
the
I
mean
the
minus,
the
the
recording
that
I
made
this
time
so.
A
A
I
will
update
the
target
dates
for
the
the
two
chicago
food
documents
and
I
guess
we
meet
in
two
weeks
now.