►
From YouTube: IETF-LPWAN-20211207-1500
Description
LPWAN meeting session at IETF
2021/12/07 1500
https://datatracker.ietf.org/meeting//proceedings/
B
B
So
it's
pretty
much
the
five
after
the
hour,
not
there
yet,
but
I
guess
we
can
start
because
we
have
a
lot
on
our
plate
today.
B
Let
me
start
doing
the
the
usual
introduction,
and
hopefully
the
people
who
are
missing
will
join
if
they
plan
to
join
at
all
so
hello.
All
this
is
an
official
ietf
interim
of
the
lp1
working
group.
So
welcome
all
usual,
not
well.
As
you
know,
we
we
have
always
played
this,
not
well
with
the
focus
on
the
patterns.
B
So,
if
you're
aware
of
any
ipr
it's
being
discussed
today-
and
that
was
not
announced-
please
announce
it
or
let
the
chairs
know,
but
the
the
iitf
is
also
very
concerned
with
other
misconducts
such
as
harassments
within
the
working
group,
so
be
very
aware
that
there
is
a
best
practice
and
there
is
a
special
group
called
ombudsteam
with
with
the
link
given
in
this
page.
B
If
you
are,
if,
if
you
have
some
primes
related
to
harassment,
you're
being
harassed
or
you
see
people
being
harassed,
please
let
the
embassy
team
know
and
the
atf
will
will
act
on
it
classical
reminders.
The
minutes
are
being
taken.
B
If
you
can
visit
the
link,
that's
in
blue
in
this
page,
it's
actually
13.
Sorry,
there's
a
typo.
This
is
the
code
emd.
I
already
shared
it
in
the
chat.
If
you
need
it,
I
can
share
it
again
on
the
chat
and
the
presents
is
logged
automatically
by
mitico.
So
you
you
don't.
We
don't
have
to
maintain
the
list
of
of
the
people
anymore,
which
is
a
great
service
by
mitika.
B
C
B
Yes,
we
we
asked
anna
already,
but
I
was
giving
the
kodi
to
make
sure
that
anybody
who
wishes
to
help
will
effectively-
and
I
will
take
notes
as
well-
to
be
aesthetic.
B
So
the
so
10
minutes
on
the
lp1
architecture.
We
already
got
some
questions
by
olivier,
so
if
only
joins
that,
then
we
can
discuss
his
points.
Otherwise
I'll
try
to
to
do
my
best
by
memory.
B
B
So
we
had
basically
two
action
items
alex
and
I
were
in
charge
of
updating
the
architecture
which
we
did
and
lauren
the
there
was
some
text
in
the
data
model
that
was
related
to
the
the
sick,
fox
improvements
of
check,
and
we
decided
that
they
should
be
in
a
separate
document.
So
you
were
supposed
to
take
that
off.
A
Yes,
it
has
been
done
so
the
new,
the
new
version
of
the
draft
has
nothing
to
do
with
oam
and
componek,
so
this
has
been
done
and
the
module
for
compondak
is
on
the
repository
and
one
carlos
has
to
integrate
it
into
the
document,
and
the
part
also
for
om
has
been
extracted.
Weapons
have
to
be
also,
we
can
work
on
it.
B
Very
well,
thank
you
very
much.
Lauren
well,
group
status,
so
here
are
dates
and
you
can
see
that
shake
over
sick
fox
is
now
a
little
bit
overdue.
We
are
supposed
to
publish
lp
one
couple
hack
first
and
we
don't
really
have
a
milestone
for
it.
It's
kind
of
you
know
the
split
of
the
sick
fox.
So
the
question
here
is
eric.
Should
I
enter
a
milestone?
Is
it
really
necessary,
or
do
we
consider
that
the
two
documents
are
kind
of
one.
B
It's
just
for
us
to
remember
mostly
yeah
okay,
so
we
don't
change
the
milestone
for
now
the
sick,
fox
document.
We
expect
it
very
soon
anyway,
I
don't
see
if
there
is
value
to
change
the
milestone.
B
And
well
in
the
iot,
we
expect
one
at
some
point,
but
maybe
after
the
discussion
we
have
today
whether
we
need
to
integrate
a
dtn
or
not
that's
pretty
much
it
nothing
new.
We
are.
We
are
missing,
updates
on
the
oem,
so
dominique.
Do
you
have
any
plans
right?
Oh
laura,
laura!
You
wanted
to
say
something
go
ahead.
It's.
E
A
B
That
that
one
is
for
eric,
I
guess.
C
I
I
think
this
is
lauren.
I
think
this
is
the
one
who
requested
that
actually
basically
or
alex
requested
a
month
ago,
so
nothing
is
new.
As
far
as
I
know,
that's
not
really
the
old
one.
Okay,
rucker
doesn't
try
so
much
this.
B
So
that
creates
this
entry.
I
was
curious
to
see
it
as
well.
So
that's
interesting!
Okay,
thank
you
eric
thanks!
So
much
and
actually
we
did
it.
You
know
thanks
to
you,
because
it's
we
are
not
used
to
that
procedure.
So
many
thanks
for
helping
us
through
that
and
that's
pretty
much
it
for
the
the
the
bachelor
bashing
etc.
So,
let's
move
to
the
architecture
so,
like
I
said,
zero
one
was
published
there.
It's
kind
of
an
interesting
change.
I
mean
there's
a
not
huge
but
important
change
to
to
the
document.
B
We
re
reshuffled
it
a
little
bit.
So
the
order
is
a
little
bit
changed
and,
as
you
can
see,
the
main
points
are
applicability.
Then
we
discussed
the
architecture
itself
with
the
endpoints
and
the
layers.
Then
we
have
some
words
on
the
data
model
and
we've
got
this
new
section
on
the
life
cycle.
B
That's
the
one
that
was
discussed
on
the
mailing
list
with
olivier
today
and
so
the
we
thought
that
we
needed
to
to
wrap
the
the
data
model
with
some
text
that
explains
where
it
comes
from
and
how
it's
being
used,
how
it's
being
published,
etc,
and
that's
why
we
have
we
have
this
section
seven.
So
it's
not
really
architecture
which
is
five,
but
it's
it's.
Basically
all
the
life
cycle
olivier,
was
concerned
that
it
seemed
to
to
add
requirements
to
the
development
cycle
and-
and
that's
not
really,
the
idea.
B
The
idea
is,
if
you
want
to
follow
this
life
cycle,
which
you
don't
have
to
right.
It's
an
informational
document,
it's
a
lowercase
shirt,
but
if
you
want
to
basically
to
have
the
the
next
steps
like
the
publication
of
the
rules
and
then
the
availability
and
signature
and
all
the
securitization
of
the
rules
etc,
which
are
described
in
this
lifecycle,
if
you
want
that,
then
at
the
time
you
did
you
develop
the
device
and
and
decide
which
messages
it's
gonna
exchange.
B
Certainly
that's
a
good
time
for
publish
writing
the
rules.
Basically,
so
that's
why
we
we
effectively
have
a
7.1
about
the
device
development.
It's
not
saying
how
you
could
develop.
It's
saying
you
probably
want
to
produce
the
rules
at
that
time,
so
they
can
be
published
in
7.2
and
and
available
with
the
device
develop
their
deployments
etc.
B
So
this
is
very
early
text,
but
I
would
really
appreciate
that
people
go
through
that
to
see.
If
that's
the
kind
of
way,
they
thought
that
the
rules
could
be
deployed
for
a
classical
device,
so
low
power
device
shake
is
not
only
for
the
power
device.
So
that's
really
the
case.
The
use
case
of
low
power
devices,
if
you
have
big
error
and
switches,
doing
shake
which
they
really
could.
B
Hopefully
they
they
could
have
other
methods,
but
for
devices
it
makes
a
lot
of
sense
to
to
basically
prepare
the
rules
during
the
development
and
make
that
available
across
the
web
or
on
extra
servers,
etc
and
and
provision
that
into
the
network
at
the
time
the
first
device
is
actually
instantiated,
so
that
that's
why
we
have
this
section.
What
it
describes
again,
it's
it's
not
there
to
say
here
is
how
it
is.
It's
there
to
steer
the
discussion.
C
B
C
C
B
I
think
it's
a
very
short
section
that
describes
what
it
is,
because
we'll
need
that
in
section
seven,
but
now
section
six
could
be
effectively
merged
into
section
seven.
I
don't
know
it's
just
it
describes
that
we
have
data
model,
but
that's
where
the
rules
that's
how
the
rules
are
expressed,
and
once
we
have
that,
I
mean
it's
not
a
big
text
anyway,
but
yes,
I
guess
we
could
remove
it.
A
No
yes,
in
a
previous
mail
I
asked.
Maybe
we
need
a
place
where
we
can
have
a
list
of
data
models
that
are
related
to
chic.
So
now
we
have
the
the
main
one
and
we
have
augmentation
for
copenhagen
soon
or
a.m.
So
maybe
it
will
be
good
to
have
a
place
where
we
list
all
the
all
the
extension
to
the
basic
data
model.
B
That's
that's
an
excellent
point
and
let
me
tell
you
what
we
are
doing
at
raw.
Initially,
we
had
a
document
which
was
both
architecture
and
framework
and
trying
to
mimic
a
little
bit,
but
it's
never
perfect.
What
that
net?
Does
that
net
published
the
architecture
very
early
in
the
process
and
it
the
architecture
says
we
don't
know
exactly
how,
but
here
is
what
it's
going
to
do
and
there
is
what's
going
to
be.
B
Then
we
do
a
framework
which
basically
says
yo
here
is
how
all
those
rfcs
and
all
those
documents
work
together
to
actually
implement
the
architecture.
B
B
A
No,
yes,
I
think
that,
for
the
architecture
it's
we
are
learning.
D
A
We
are
doing
so,
I
think
now
we
don't
have
the
same
vision
of
chic
as
we
had
at
the
first
meeting,
and
the
architecture
is
more
and
more
stable,
and-
and
so
I
don't
know
if
we
can
write
a
document
before
and
it
will
be
true
at
the
end.
B
In
this
case,
the
the
thing
that
you're
asking
in
this
case
the
thing
that
you're
asking
a
repo
a
place
to
explain
what
the
data
models
are.
It
could
be
this
section
six,
in
which
case
it
would
remain.
You
know
to
eric's
point:
it's
not
part
of
the
architecture,
but
it's
part
of
the
framework,
and
so,
if
we,
if
we
turn
section
six
into
a
framework,
then
there
would
be
a
data
model
and
all
the
optional
pieces
that
will
be
the
components
that
will
be
added.
B
Okay,
so
so
I'm
good
with
with
what
lauren
says
for
now,
we
will
not
split.
I
will
turn
section
six
into
framework
and
there
will
be
one
subsection
of
framework
which
will
be
the
data
models
which
will
have
to
describe
the
various
the
domain
and
then
the
various
optional
components
that
fit
into
the
data
bottle.
I
guess
okay
and
that's
not
totally
yet
so.
What's
missing,
we
are
missing
already
some
sections
that
are
there,
but
they're
not
filled.
B
We
want
to
have
some
examples
and
there
was
the
example
of
goose
which
came
to
mind.
Goose
is
a
smart
grid
protocol
to
control
the
smart
grid,
I'd
like
to
see
reviews
that
discussion,
especially
on
the
life
cycle,
and
that
has
started
with
olivier.
B
Probably
we
want
to
to
explain
more
about
the
detailed
discussion
we
have
today
and
and
the
firmware
update
over
the
year
and
then
in
in
this
life
cycle
section.
There
is
a
little
bit
of
discussion
about
key
management,
and
when
do
you
publish
the
keys
of
this
device
in
the
network
in
the
application?
B
B
So
that's,
what's
missing
would
like
to
see
discussions
on
this
and
that's
pretty
much
it
for
this
piece.
So,
yes,
I
guess
a
bit
out
of
time
any
question
any
more
discussion.
Otherwise
we
will
go
to
the
list.
B
Check
the
agenda
for
a
sec
second
was
not.
B
Added
card
who's
presenting
anna,
is
it
you
or
is
it
a
guard?
I
see
both
of
you
are
present.
B
F
B
B
B
B
B
F
So
yeah
this
is
a
presentation
about
basically
a
problem
that
it
was
identified
for
mainly
narrowband
iot.
F
We
can
think
that
I'm
not
sure
if
we
could
find
exactly
the
same
problem
in
other
lp1
technologies,
but
then
we
thought
that
it
would
be
a
good
idea
to
to
discuss
a
little
bit
if,
if
it's
a
good
idea
to
maybe
have
this
a
separate
work
or
just
include
it
as
part
of
the
narrowband
iot
draft,
although
I
think
anamiy
we
have
been
discussing
with
all
that,
it
would
be,
may
be
beneficial
to
split
it
in
a
different
draft,
because
it
is
quite
a
different
thing,
but
we
can
go
through
this
first,
so
the
main
problem.
F
So
one
of
the
thing
is
to
to
to
to
be
clear,
I
mean
shake.
It
has
been
not
so
easy
to
to
be
introduced
in
3gpp
we
have
been
trying
a
little
bit
to
discuss
offline
and
then
one
of
the
problems
that
we
have
found
is
that
the
the
implementations
of
narrowband,
iot
and
and
and
the
the
hardware
implementation,
so
the
modems
are
quite
behind
the
standard,
so
they
they.
F
Let's
say
they
have
taken
the
first
release
and
they
have
taken
only
the
let's
say,
mandatory
features
and
that's
the
only
thing
they
have
implemented
so
trying
to
save
money.
I
mean
there
is
a
lot
of
issues
there.
There
is
patents
and
so
on.
So
then
they
try
to
avoid
the
implementers
to
pay
more
more
than
what
they
they
have
to
so
then
they
actually
reduce
the
number
of
features
that
they
implement.
F
Then
that
means
that
the
user
plane,
which
is
the
most
powerful,
let's
say
featuring
in
for
cellular
it-
is
not
fully
implemented.
F
So
instead
of
implementing
the
full
user
plan,
where
you
you
transmit
data
packets
and
normally
like
you
would
do
in
any
technology,
what
is
on
is
that,
instead
of
that,
what
is
implemented?
Is
this
data
or
nas,
which
is
a
is
a
type
of
transmission
where
the
data
packets
are
piggyback
to
the
control
signaling
of
the
of
the
control.
A
F
So
when
you
have
a
control
channel
and
then
you
have
mobility
information
or
you
have
a
paging
or
you
have
a
reconfiguration
of
the
device.
So
then
pick
back
that
you
can
actually
put
some
small
pieces
of
payload
and
this
was
meant
to
be
used
for
very,
very
infrequent
on
little
data.
So
we
are
talking
about
maximum
kilobits
of
data
and
there
are
several
reasons
for
that.
F
This
is
actually
a
problem,
because
if
you
have
many
devices
which
are
in
this
channel,
then
what
happens
is
that
they
might
actually
start
to
affect
the
cell
capacity
because,
as
I
say,
this,
the
the
cell
measurements
and
then
the
the
handover,
metrics
and
and
even
the
handover
commands
goes
over
that
channel.
F
F
But
then
the
reality
is
that,
well
all
the
devices
have
this
kind
of
firmware
updates
and
in
some
cases
there,
even
these
log
files
that
need
to
be
uploaded
or
sometimes
and
download
I
mean
the
firmware
updates-
are,
of
course
downloaded
and
the
large
log
files
are
uploaded
and
and
then
that
is
causing
these
options
in
in
the
sites.
So
basically
people
who
are
doing
phone
calls
cannot
do
the
phone
calls,
because
this
these
channels
are
saturated.
F
The
problem
is,
of
course,
is
that
this
is
also
transmitted
with
other
protocols
like
tcp,
and
then
you
have
retransmissions
and
of
course,
if
the,
if
they
are
congestion,
then
the
retransmissions
make
the
situation
worse
and
the
saturation
of
the
channel
is
even
they
could
a
signal
in
storm.
Basically,
so
then
that
has
caused
that
pretty
much
is
not
possible
today,
but
it's
very
rare
that
you
can
do
firmware
update
on
your
devices
over
narrow
iot
or
even
in
some
cases
ltem.
F
But
it's
mainly
narrowband
iot
is
the
problem
which
is
the
because
they
have
the
mandatory
thing
so
felt
you
would
have
to
implement
use
the
the
user
plan
for
narrowband
iot.
You
don't
have
to
so.
You
can
go
away
with
this.
F
So
then,
one
possible
possible
solution
is
like
well
if
the
modems
are
going
to
continue
having
this
is
why
not
to
transmit
these
packets
very,
very
slowly
over
a
large
period
of
time,
and
then
then
the
the
solution
could
be
to
enable
this
kind
of
delay,
tolerant
transmission,
where,
instead
of
taking
let's
say
10
minutes
to
to
transmit
10
megabytes
or
100
megabytes,
you
can
take
one
week
and
then
you
don't
have
to
deal
with
this.
Also
congestion
problems
due
to
retransmission
or
synchronization
mechanisms
of
protocols
which
are
not
delayed,
tolerant,
friendly.
F
F
If,
as
if
the
the
whole
object
would
be
one
huge
ip
packet,
then
we
would
set
up
the
whole
timers
on
and
window
sizes
and
so
on.
Accordingly,
to
some
kind
of
delay,
tolerance
parameter
configuration.
F
So
then,
we
could
actually
reuse
cheek
to
to
pro
provide
this
kind
of
delay,
tolerance
service
by
using
fragmentation,
fragmentation
services
and
then
a
little
bit
the
reality.
If
we
want
to
use
the,
for
example,
akon
error
which
for
a
mobile
cellular
network,
it
would
be
good
enough.
F
The
idea
is
to
have
something
like
this,
so
you
will
have
configurations
which
are
map
mapping
to
certain
let's
say:
values
of
maximum
maximum
packet
size
or
object
size.
So
then
you
could
have
the
opening
a
configuration
and
then
downloading
a
different
configuration,
so
it
doesn't
necessarily
have
to
be
the
same.
B
Yes,
I
have
a
knee
jerk
reaction.
If
you
don't
mind
which
is
well
that
there
are
several
things
that
you
kind
of
asked,
but
not
really
one
is,
is
it
something
that's
very
specific
to
an
iot
or
would
there
be
interest
in
other
technologies
seems
to
me
that
it
would
be
huge
interest
in
many
technologies.
I
don't
see
why
it
should
be
at
least
iot.
B
Does
it
fit
in
this
working
group?
I
think
it
really
does
if
we
need
to
reach
out
there.
Maybe
we
do,
maybe
we
don't.
I
would
be
pushing
for
that
to
happen
so
so
very
interested
now.
Can
we
use
shake
as
he
is?
B
I
probably
I
don't
think
so,
because
you're
creating
a
long
lasting
session,
and
I
believe
that
this
should
be
formalized
in
particular
when
the
session
breaks
in
the
middle
of
one
megabyte.
A
B
Don't
know
it
might
not
be
one
megabyte,
but
you
want
to
be
able
to
restart
from
there.
So
so
there
are
functions
like
this.
That
will
need
to
be
added
to
check
to
make
it
what
you
want
it
to
be.
So
I
I
I
really
believe
in
this
work,
but
I
don't.
I
don't
really
believe
that
you
can
say
hey.
Let's
pick
chic
asis,
I
guess
that
would
be
a
study
of
what
needs
to
be
done
to
check
and
with
that
I
will
give
the
ball
to
eric.
C
Thank
you,
pascale
and
thank
you
edgar.
I
think
that's
indeed
a
sensible
problem
to
solve
and
the
real
one
it
seems
now.
I
got
two
commands
one
which
is
minor.
There
is
a
delay,
tolerant
network.
We
can
group
at
the
ietf,
so
we
may
want
to
change
your
vocabulary
here.
To
avoid
confusion,
that's
easy
to
do
now.
I'm
afraid
it's
again
interesting
work
right,
but
it
will
require
rechartering
of
lp1
to
doing
it
and
I'm
a
little
bit
concerned
that
we
are
going
really
into
the
transport
layer.
C
I
mean
implementing
fine
state
machine
and
so
on.
So
I'm
not
sure
whether
this
kind
of
work
should
be
done
in
lp1
completely
and
not
in
another
place
into
the
transport
area.
So
it's
a
little
bit
too
early
to
to
think
about
it.
So
we
can
continue
to
work
on
this,
of
course,
but
transport
problems
are
really
really
specific.
C
B
But
we
eric,
we
already
have
retransmissions,
you
know
there
are
always
this
sort
of
thing
on
on
a
short
distance
at
the
lower
layer,
like
wi-fi,
we
transmit
up
to
64
times
and
we
do
retransmit,
some
lost,
fragments
and
and
the
transport
is
the
center
and
thing
right
and
they
won't
want
to
work
below
layer
3.,
and
here
we
are,
we
we're
kind
of
what
you're
saying
makes
sense,
but
I
really
want
would
like
to
to
to
debate
that.
I
mean
there
are
pros
and
cons
and.
B
C
F
Then
then,
then
we
we
are.
We
can
have
a
philosophical
discussion,
how
big
an
ap
packet
can
be,
but
then
of
course
I
I
agree
that
they
are
transport
aspects
to
this,
and
one
possibility
could
be
that
in
the
after
very
long
time.
Let's
say
what
happened
if
the
ip
change
could
be
a
possibility,
I'm
not
sure
if,
if
it's
feasible,
but
then
it
might
be
that
you
need
to
do
some
kind
of
you
know
some
some
kind
of
the
ancient
ap.
B
I
don't
think
I
don't
think
you
should
be
saying
it's
one
big
ip
packet,
because
it
doesn't
have
to
be
so
don't
make
any
assumption.
You
say
what
what
you
said.
I
have
this
big
file.
Transfer
upload
or
I
have
this
big
firmware-
update,
update,
download
I've
heard
that
in
pretty
much
every
iot
use
case,
I've
been
facing
over
15
years,
so
these
are
really
real
cases.
B
B
F
No,
no,
that
was
just
half
joke,
so
don't
take
it
fully
fully
seriously,
but
then
I
think
I
mean
there
is
a
matter
of
discussion
how
how
we
want
to
model
it
and
so
on,
because
then
you
you
enter
in
the
in
the
issues
like
so
are
what
what
are
we
having
a
chic
over?
What
is
it
ip
you
know
so
then
I
I
think
that
that's
the
thing.
Maybe
we
need
to
discover
that
later
offline,
most
probably
in
the
mailing
list.
A
A
Okay,
so
maybe
that's
a
solution
for
when
you
close
the
session
and
start
again
it's
you
know
you
can
have
a
bitmap
of
what
you
have
received
or
not.
So
maybe
I
think
there
is
solution
and
the
adaptation
is
not
that
big
to
what
pascal
said
before
our
speed
can
be
used
also
to
to
do
some
focus
for
control.
If
we,
if
we
need.
F
F
Transmitting
is
a
large
charge
and
then
the
the
network
would
then
do
this
kind
of
transmission
pacing
it
accordingly
to
the
network,
scheduling
that
that
can
handle
for
that
specific
site
where
the
device
is
so
that
you
don't
get
the
congestion,
then,
when,
if
there
is
any
packet
miss,
so
it
can
be
retransmitted
with
the
icon
error,
as
we
mentioned
so
and
then.
Finally,
there
could
be
a
notification,
but
this
is
our
details
already
from
the
network
so
that
the
link
to
the
application,
okay,
this
object
has
been
delivered.
F
So
in
that
sense,
it's
not
fully
end
to
end,
at
least
not
in
this.
In
this
example,
because
you
would
need
to
to
communicate
to
the
network,
use
it
as
a
service
to
transmit
this
packet,
and
then
you
will
get
indication
saying
yeah.
This
was
done
for
you,
of
course,
it
could
be
done
through
ip
proxies
and
maybe
then
in
that
way
could
could
work
something
similar
and
the
the
doubling
case
would
be
a
little
bit
more
complicated
because
there,
the
device
would
need
to
kind
of
choose
between
the
profiles.
F
So
it
would
have
several
raw
files
with
different
configuration
and
then
it
have
to
choose
the
right
one.
According
to
the
let's
say,
the
object
size
that
it
needs
to
transmit
and
then
based
on
that
it
would,
it
would
initiate
this
transmission
and
of
course,
sometimes
the
network
might
overwrite
those
configurations
and
say
like
no
before
now.
My
my
congestion
levels
are
different.
Please
use
this
other
configuration
because.
F
We
we
need
to
have
even
even
more,
let's
say,
parameters
which
are
more
aggressive,
but
then
the
rest
of
the
fees
are
the
same,
pretty
much
the
same.
Yep.
B
Yes,
I
mean
that
when
you
write
the
prime
st,
I
hope
you
write
a
small
prime
statement
which
represents
what
you
said,
but
when
you
write
it
insist
on
that
I
mean
the
the
kind
of
if,
if
there
is
any
kind
of
congestion,
because
eric
raised
that
topic,
that's
not
a
end-to-end
congestion
that
has
to
be
dealt
by
the
source.
It's
just
over
this
hop
it's
just
slowing
down.
It's
just.
You
know
feedback
from
this
link
and
we
on
many
radios.
I
mean
we
look
at
dilip
everything
delay
up
in
money.
B
I
mean
we
have
this
kind
of
feedback
about
the
radio
where
the
transmitter
has
to
slow
down
just
over
that
hop
and
does
it
is
really
way
below
transport
it's
below
ip,
and
here
we
are
in
the
same
situation
as
dilip
having
to
slow
down
below
ip.
Just
because
the
link
changes
and
that's
fine-
I
mean
that's
our
job,
that's,
but
that's
really.
B
Where
I'm
saying
it's
not
just
check
you,
you
have
a
session
and
and
you
you
have
to
maintain
it
over
multiple
connectivity
loss
and
and
come
back,
and
so
that's
that's
more
of
the
and,
yes,
you
have
to
to
basically
adapt
the
rules
that
are
used
for
that
session.
B
F
B
Like
dilip
well,
the
lip
is
is
basically
a.
I
have
this
router
and
I
have
an
external
modem
for
my
radio.
Okay,
it's
not
the
router,
but
I
have
a
I
don't
know:
100
megabit
ethernet
between
the
router
and
the
radio.
If
the
router
pushes
everything
it
can
over
ethernet,
it
will
push
100
max,
but
actually
the
radio
can
do
one
mag.
B
F
So
you
you
can
then
do
the
more
or
more
devices,
let's
say
at
the
same
time
or
more
more
customers
at
the
same
time,
depending
because,
of
course,
you
are
mix
it
with
all
the
others.
The
problem
of
the
narrow,
iot
networks
is
that
they
coexist
with
lte,
so
they
take
also
part
of
their
they.
They
take
part
of
their
bandwidth.
F
So
then
yeah
the
problem
is
basically
these
data
channels
and
so
on.
So
then,
if
you,
if
you
have
too
many
of
those
users
at
the
same
time,
so
it
might
be
that
the
network
say
like
well
wait
a
minute.
I
would
need
you
to
because
I
I'm
I'm
feeling
that
I'm
kind
of
reaching
some
level
of
too
many
users
at
the
same
time
of
concurrency
so
then
use
instead,
these
new
parameters,
so
that
when
you
are
going
to
send.
F
Not
a
packet
an
object,
so
then
you,
you
are
even
having
more
aggressive
parameters
for
that,
but
that's
it
so,
but
then
you
don't
do
changes
on
the
middle
of
most.
Probably
you
don't
need
to
do
changes
on
the
middle
of
a
session.
So
once
you
have
started
the
session,
I
think
the
network
can
handle
that
and
it's
taken
in
account
the
problem
is
beforehand.
F
So
then
that's
what
the
network
might
might
preemptively
change
the
parameters
and
and
then
do
this
kind
of
dynamic
configuration
and
that's
another
thing.
That
is
a
bit
different
with
the
chic
nature,
because
chic
is
supposed
to
be
quite
static
and
then
here
we
are
having
these
configurations,
which
are
they
might
be
static,
so
they
might
be
already
set
beforehand,
but
then
you
might
need
to
choose
between
them
and
so
on.
So
then,
in
that
sense
it
make
it
a
bit
more
dynamic
that
it
you
meant
to
be
in
the
beginning,
when
chick
was.
B
Oh
very
good,
so
let's
continue
on
the
mailing
list.
I
mean
I
can
see
eric's
objections,
but
I
think
we
can
work
around
them.
It's
just
that
it's
it
might
be
bigger
than
then
help
you.
I
don't
know
because
it's
not
just
you,
don't
really
use
shake
as
a
compression
mechanism
and
it's
shake
would
become
the
thing
for
everything
in
the
iot.
That
needs
big
big
data.
That's
that's
huge,
okay
anyway!
So
anna!
Yes,
please.
B
E
B
E
B
It's
like,
like
eric,
said,
we
are
not
even
chartered
for
it
at
this
time,
so
good,
okay
and
lauren.
So
can
you
please
stop
sharing
and
lauren?
Can
you
please
do
the
same?
Okay,.
A
So
it's
just
because
we
we
had
some
discussion
with
juan
carlos
this
week
and
how
to
do
with
companak,
and
so
the
goal
is
maybe
not
to
document
but
to
to
discuss
about
that,
but
because,
as
I'm
not
a
full
specialist
of
young,
so
maybe
I
very
think
that
I
don't
know
but
yeah.
I
will
tell
you
what
I
have
learned.
A
So
the
next
slide
is
just
to
to
show
you
how
we
did
for
companak.
And
so
do
you
see
the
mouse
when
I'm
showing
things
or.
B
A
Okay,
so
how
do
I
do
because
I
I
some
stop
slideshare?
I
have.
C
B
B
A
With
the
mouse
okay,
so
when
you
create
a
new
model,
for
example,
for
chic
components
for
components,
so
what
I
we
did
is
to
put
the
chic
at
the
beginning
to
remember
that
it's
something
that
belongs
to
the
chic
model
and
we
give
a
prefix
that
can
be
used
after
in
the
model.
That
is
a
cheek
component.
A
A
So
we
do
it
as
usual,
and
the
last
thing
you
have
augment
on
in
augment
what
we
specify
is
the
identifier
of
the
element
and
we
put
chic
at
the
beginning,
which
refer
to
the
fact
that
it's
in
the
ietf
chic
data
model
and
in
this
example
we
add
a
leaf,
and
this
lip
is
a
bit
format.
And
then
we
we
have
tests.
A
A
H
Okay,
but
then,
and
is
there
a
is
there
a
common
way
of
doing
it
like
does
it
have
to
be
in
the
in
the
body
of
the
draft
or
is
it
an
annex
or
or
more
cookers?
You
know.
A
H
Okay,
all
right
so
so
I
guess,
then
the
question
would
be
to
the
to
the
chairs,
because
we
have
the
chic
compound
draft
after
last
call,
and
we
got
a
couple
of
comments-
the
ones
on
on
the
beat
map
and
stuff
from
from
dominique.
I
think
we,
we
know
how
to
address
them,
and
we
have.
We
have
a
proposal
and
what
the
question
that
we
still
had
was
how
to
how
to
move
forward
with
this
compound
axo.
H
So
my
proposal
would
be
to
include
this
text
as
lauren
is
saying
showing
it
in
the
in
the
document
and
then
publish
it
as
a
response
to
the
to
the
last
call
that
we
had
and-
and
I
don't
know
if
the
wish
would
be
to
to
do
another
last
call
or
to
request
comments
or
send
it
for
review
with
some
directorate
or
or
what
would
be
the
next
step.
B
I
don't
know,
I
don't
think
we
need
another
last
call
for
the
young
data
model
right,
but
maybe
we
we
said
we
could
trigger
like
we
did
last
time
review
by
the
young
doctors,
or
I
mean
what
you
did
please
say,
or
is
it
this
disturbed
pyeong
that
you
asked
about?
Do
you
discuss
because
I
understand
you
can
run
a
tool
on
on
the
draft
itself
and
it
will
say
if
the
embedded
young
is
correct
right
is
that
is
that.
A
B
C
Yeah,
so
juan
carlos
one
easy
way
is
to
use
the
young
validator.com
website
where
you
can
upload,
either
the
module
itself.
So
the
few
lines
we
see
here
on
the
complete
rsc
as
a
dot
txt
and
extract
the
module
from
the
txt
and
then
you
can
check
the
validities
also
using
piang
and
other
validators.
If
you
prefer-
and
I
concur
with
pascal-
it's
nothing
is
changing
really
on
the
protocol
itself
here.
So
I
don't
believe
in
group
last
call
again
for
20
lines
of
data
model
is
required.
C
H
Okay,
okay,
so
I
will
coordinate
with
logan,
because
I
guess
he's
more
familiar
with
the
with
the
young
models
and
the
tools
and
and
once
we
have
the
the
new
version,
we
will
publish
it
and
announce
it
on
the
on
the
list.
G
B
When
you
asked
the
young
doctors,
they
reviewed
this
piece
as
well
right
because
it
was,
it
was
together
with
the
rest
at
the
time,
so
it
has
been
reviewed
kind
of.
Is
it
correct.
A
B
Is
that
something
which
will
naturally
happen
during
the
asg
review?
I
guess
so.
C
There
will
be,
I
mean
during
the
ie
review.
There
will
be
a
young
review
by
rob
wilton,
which
is
the
ops
id
in
charge
of
yang,
and
most
probably,
if
I
remember,
I
would
request
at
the
itf
last
call
for
the
next
step,
a
young
doctor
review
on
this
one
and
suggestion
basically
to
reuse.
I
don't
remember
who
did
it
for
the
first
early
review
or
asking
for
okay.
B
C
B
That
tells
us
that
we
can
really
ask
for
publication.
I
mean
once
juan
carlos
has
addressed
all
the
the
issues
and
and
added
the
my
main
issue.
Was
the
data
model
itself?
So
once
all
this
is
addressed,
we
can
ask
for
publication,
I
mean
as
soon
as
juan
carlos
publishes.
I
will
press
the
button
right.
C
The
better
right
in
ideal
work
to
get
the
normal
data
model
from
lauren,
anna
and
juan
carlos
extension
requested
for
publication
roughly
at
the
same
time,
so
I
can
build
them
together.
C
It's
not
delaying
something
for
weeks
right
and
anyway,
as
this
compound
now
depends
on
the
data
model
right,
there
will
be
a
cluster
after
iag
evaluation,
so
they
will
be
published
at
the
same
time.
Yes,.
B
B
A
A
Model
is
the
full
chic
data
model
is,
is
complete,
so
we
can
go
to.
We.
B
Can
publish
yeah
so
that
so
dominique
being
the
the
editor
of
the
chic
document,
did
you
did
you
make
a
pass
on
the
completeness
of
the
data
versus
appendix
d
or
whatever?
Yes,
I
guess
it
was
up
at
xd
right
off
of
the
the
chic
spec.
Is
it
complete
in
the
terms.
B
I
would
like
to
see
that
done
by
us
before
we,
we
publish,
but
after
that
I'm
ready
to
press
the
button
as
well.
D
D
B
B
So
basically
what
I
suggest
is
I'm
waiting
for
the
dominic
to
to
say.
Yes,
I
found
you
know
the
entries
for
what
I
think
is
necessary,
and
then
we
get
the
the
juan
carlos
to
publish
his
new
version
and
then
I
will
press
both
buttons.
At
the
same
time,.
B
B
B
Okay,
so
early
next
week,
dominic
do
you
have
a
chance
to
do
to
to
to
make
a
pass
before
next
week,
then
the
target
that
would
be
a
week
from
now
before
christmas?
I
mean
sometime
before
christmas.
B
That's
nice,
okay,
okay,
so
so
the
objective
is
we
push
the
buttons
for
both
drafts,
some
sometimes
next
next
week,
depending
on
exactly
what
you
could
do
that
that's
great.
So
thank
you.
Thank
you
very
much
that
that's
really
huge
progress,
and
now
we
have
amazed
we'll
have
those
milestones
this
year,
which
is
nice
okay,
and
I
guess
we
are
we
passed
the
hour.
B
So
thank
you
all
very
much
for
all
the
great
progress
that
we've
done
recently
and
this
group
is
still
rocking
and
I
hope
it
will
be.
It
will
keep
rocking
next
year.
So
looking
forward
to
talk
to
you
again
next
year,
bye.