►
From YouTube: IETF99-6TISCH-20170717-1330
Description
6TISCH meeting session at IETF99
2017/07/17 1330
https://datatracker.ietf.org/meeting/99/proceedings/
A
A
And
this
is
the
new
not
well,
so
it's
kind
of
the
same
as
usual.
So
you
need
to
remember
that
if
something
is
being
discussed
and
you're
worried
about
IP
are
you
need
to
please
let
us
know
after
the
chance
after
the
meeting
and
then
please
go
through
this
slide.
I
won't
spend
the
time,
but
all
the
contributions
to
the
IDF,
including
this,
what
is
being
said
in
this
meeting
and
on
the
mailing
list,
must
follow
the
rules
in
RFC,
53,
78
and
RFC
81
79,
please.
A
If
you're,
not
too
sure,
go
and
read
those
heiresses,
this
is
a
official
meeting.
Minutes
will
be
taken.
We
have
two
minutes
takers,
thank
you
guys
and,
and
two
helpers
on
the
chibok
as
well.
We
are
taking
the
minutes
on
the
fly
on
etherpad,
so
there
is
a
on
the
data
tracker
agenda.
There
are
number
of
buttons
associated
to
the
meeting
and
one
of
those
buttons
give
you
direct
access
to
these
are
pad.
Otherwise
the
link
is
written
on
the
bottom
of
the
slides
here.
A
So
if
you
can,
please
go
to
each
our
pad
and
if
you
said
something,
for
instance,
you
may
vary
date
that
the
minutes
are
taken
correctly
and
reflect
what
you
in
turn
need
to
say.
Okay
or
if
you
want
to
participate
to
the
mean
taking,
that's
only
appreciated
so
we
go
through
the
agenda.
We've
got
quite
a
full
agenda
today,
so
we
will
be
giving
a
summary
of
the
FF
and
drop
test
which
happened,
Friday
and
Saturday,
and
the
Agathon
activity
as
well.
A
A
Manish
are
talking
about
the
minimum
security,
which
is
mostly
done
piece
of
work
and
then
Michael
will
come
and
present
to
us
the
status
of
design
team
and
if
you
can
address
a
little
bit,
the
street
Rostova,
Michael
and
then
time
committing
so
we'll
use
the
rest
of
our
time.
For
those
those
drafts
we've
got
a
new
scheduling,
function
that
Simon
prisoners.
Then
Jonathan
will
talk
about
the
exam
of
the
60s
example.
A
A
B
A
Perfect,
why
can
I
hurt
you?
Okay,
so
there
we
go,
we
already
start
and
we
start
with
two
announcements.
So
we
have
to
arrive.
Sees
one
of
them
comes
out
of
this
group.
I,
don't
know,
is
very,
very
related,
so
the
first
one
is
thanks
to
Terrell.
Thanks
Tara
for
doing
this
work.
Now
we've
got
a
ITF
information
element
that
is
officially
ours
from
from
the
Acholi
identified.
A
L
and
last
but
not
least,
congratulation
to
the
others
in
particular
shave.
You
did
the
bulk
of
the
addition.
We
now
add
the
RFC
number
for
the
minimum
support
for
60.
So
that's
reader,
the
core
result
of
this
working
group
so
far,
yeah,
okay
and
now,
if
we
look
at
our
milestones
and
it's
good-
that
Suresh
is
with
us-
there
is
an
interesting
thing
is
we
did
not
update
the
my
stunt
with
like
the
security
work,
so
we
really
need
to
work
on
that.
A
Then
we
are
little
bit
late,
but
not
that
late.
So
we
will
discuss
about
the
sixta
protocol,
whether
we
we
are
ready
for
ASCO.
It
appears
that
we
are
from
my
initial
reading
of
the
test
that
we
were
last
week.
I
said
zero
will
discuss
as
well,
knowing
that
we
are
running
for
experimental
for
that
one,
and
then
we
will
be
talking
about
what
we
do
for
the
security.
A
E
On
this
monitor,
it
is
black
and
I
was
saying
it
has
been
organized
by
adhere
list,
and
it
has
been
founded
by
develop
Ian's
Commission
through
the
open
Cola
of
NH
2020
project
called
the
F
Interop.
As
usual.
We
add
the
support
of
two
company
open
most
for
the
ad
and
open
for
the
film.
With
this
time
we
had
the
fighter
and
I
number
of
participated
company
16,
and
we
had
also
several
implementation.
In
particular,
we
had
the
say,
5
different
implementation
for
60
SH
and
6
for
Oscar.
E
So
forgiving
you
an
idea
of
the
people
that
are
organized
and
amid
the
the
event
here
is
a
picture
of
God.
So
all
the
people
that
came-
and
we
thank
them
for
coming
with
their
own
implementation.
And
of
course
we
were
not
only
people
there,
but,
as
you
can
see,
there
were
many
shank
that
we
try
to
connect
and
to
let
them
work
together,
just
a
brief
overview
of
how
the
event
was
organized.
E
E
We
focus
mainly
synchronization
minimal,
because
now
it
is
a
receive
we
wanted
to
check
if
it
work
and
then
sick,
sick
stuff
also
and
finally,
some
security
expected
both
at
links,
layer
and
other
related
more
to
the
joining
function.
So,
overall,
as
you
can
see
here,
we
had
the
16
no
test
for
6p.
We
checked
several
comment
like
a
discount
and
for
minimal.
We
checked
the
format
of
the
image
be
corner.
There
are
setup
of
the
channel
or
finger
the
rotary.
E
My
size
and
okie
formula,
as
I
said,
did
event
run
for
one
day
in
we
have
a
initial
setup
phase
as
usual
and
then
three
main
texturization
and
on
a
Friday
on
Saturday.
After
everything
was
concluded,
we
had
the
finalization
did
it
for
giving
you
an
idea
of
the
fun
that
we
had.
So,
as
you
can
see
that
was,
we
were
happy
to
test
the
different
implementation
with
many
device
connected
again.
C
E
Also
showing
because-
and
we
were
Akito
to
test
and
validate
the
interoperability
once
we
finished
the
test,
all
the
company
involved
were
able
to
report
the
results
of
the
test
using
a
TST
tool
provided
by
Etsy,
and
this
is
a
very
efficient
it
allowed
to
the
and
then
compute
easily.
The
statistic
here
are
the
results.
As
you
can
see,
it
was
a
great
success.
We
had
almost
86
percent
of
the
tested
and
there
were
few
tests
that
were
not
applicable,
mainly
those
related
to
to.
E
C
So
Thomas,
can
you
go
back
into
a
previous
title
to
be
to
be
very,
very
clear:
we
try
to
not
cheat
because
when
so,
let
me
explain
what
did
I
say?
No,
what
we
do
is
last
time
we
didn't
kind
of
enforce
this
rule
and
if
you
do
a
test
that
it
passes,
you're
very
happy
to
click
rate
or
green.
If
it
doesn't
work
or
you
you,
you
kind
of
don't
click
red.
So
last
time
we
had
a
hundred
percent
passing
tests.
What
we
did
now
is
we
enforce
the
rule.
Okay,
look!
C
If
you,
if
you
try
a
test,
then
you
have
to
say
pass
or
fail.
If
it's
fail,
it
doesn't
matter
and
if
you
come
with
an
implementation
which
did
not
implement
that
feature
and
that
we
had,
you
know
eleven
different
limitations.
So
it's
normal
that
not
every
implementation
gets
that
feature.
Please
press
not
applicable.
So
the
gray
part
here
is
that
kind
of
gives
you
an
idea
of
the
percentage
of
what
was
implemented
from
the
things
and
so
I
think
this
is
a
very,
very
positive
number.
C
E
So,
as
I
said,
the
practice
was
organized
in
founded
in
the
context
of
the
offender
of
a
European
project
and
for
those
that
are
not
familiar.
It
is
an
era
piant
project
that
they
aim
to
develop
the
sama
online
testing
tools
that
are
available
on
a
shared
platform
and
the
aim
is
to
be
able
to
run
interoperability
a
test
while
staying
in
without
the
need
of
being
all
in
the
same
place.
So
the
idea
is
that
you
can
be
in
your
own
species.
E
More
often
without
the
need
of
travel
and
the
cutting
costs,
so
during
the
practice
that
we
show
with
a
demo,
how
is
a
platform
can
be
used
for
running,
for
example,
ripple
test,
and
the
idea
is
that
in
the
next
event,
we
will
actually
use
some.
These
are
online
testfor
for
our
sixties,
plug
days,
so
just
forgiving
the
feedback
of
the
event,
as
I
said
already
before
it
was
a
great
success.
We
had
the
very
high
percentage
of
successful
test,
in
particular
for
the
mini
mala
now
Eric's
three,
eighty
one.
H
C
E
E
E
The
same
for
the
generation
mismatch:
we
are
right
now
suggesting
to
send
a
clear,
but
will
be
better
to
add
their
own
technology,
so
these
are
all
feedback
that,
after
what
a
child
will
discuss.
So
what
we
can
say
again
it
was
a
good
success
and
we
could
actually
see
that
there
was
interoperability,
that
the
drafts
are
in
a
good
state
of
maturity
and,
as
usual,
the
event
at
the
inner
progressing
and
a
feeling
of
our
the
technology
is
now
well
implemented
and
it
works.
E
So
our
next
event,
as
I
said,
the
aim
is
to
use
the
online
testing
tool
and
to
ever
Aeromotive
event.
We
will
fix
a
date,
we
didn't
decide
yet
it
will
be
at
the
best,
as
in
other
advanced
of
course,
feedback.
Atre
we'd
come
by
that
time.
For
sure
the
tool
will
be
available.
The
event
will
be
still
of
annals
in
the
context
of
the
filter
of
project,
and
we
will
make
sure
that
there
is
enough
information
to
know
how
to
use
the
testing
tool
right
now.
E
C
So
I
wanna
I
want
to
stress
how
cool
is
this?
Well
right,
we
tested
everything
from
synchronizing,
15,
454
e,
secure,
joining
and
then
scheduling
function.
Sorry
Sixto
protocol
to
schedule
different
nodes.
So
this
is,
you
know,
we're
we're
almost
almost
services
very
cool
to
have
seen
this.
If
you
have
any
suggestions
or
comments
about
the
event
that
we
wouldn't
do
in
the
fall,
please
let
us
know
if
you
have
any
days
that
work
or
don't
work
for
you.
C
H
H
By
updating
the
bootstrap
of
the
test,
accession
of
the
finger
up
platform,
it's
more
like
flashing,
the
hardware
control
rheumatory,
the
testing
device
before
the
users
really
do.
That
has
the
finger
at
the
future
of
test.
Before
so
also,
we
have
for
progress
on
the
joint
security
we
refactoring
the
GRC
code
and
also
the
key
for
the
network.
It
can
be
generated
at
the
right
time
when
the
open
visualizer
is
you
need
initialized.
H
So
with
this
and
the
460
show,
and
of
the
with
the
security
bootstraps
and
now
is
in
the
open,
Dobson
project
and
on
the
open
doors
insider,
we
also
do
the
housekeeping
and
fixing
and
clean
up
some
the
project
ready
to
the
open
mode,
and
it
has
the
platform
and
I
also
complete,
said
that
it's
on
of
a
commission,
which
is
a
OneTouch
security
through
the
Navi
param
param
module,
and
also
we
fix
the
Sun
back
from
the
repo.
C
I
Okay,
hi
everybody
I'm
presenting
an
update
of
the
sixties
protocol.
We
are
in
version
7.
The
last
update
was
last
at
the
7th
of
June.
This
draft
is
pretty
stable,
there's
several
implementations
of
it
completely
independent
implementation,
supported
in
in
vendor
platforms
and
also
enough
in
open
source
platforms.
I
I
I
We
fixed
some
meaner
things
in
some
typos
and
errors
on
with
it,
some
reordering
of
sections
to
make
it
more
clear
and
maybe
remove
some
some
text.
It
was
kind
of
not
important
or
like
repeated
on
what,
but
we
are
now.
This
is
status
that
we
are
now
on
after
the
black
test.
We
have
identified
a
couple
of
things
that
we
want
to
comment
on.
We
want
to
see-
maybe
your
your
opinion
or,
if
you
can,
if
your
your
point
in
either
here
or
in
the
mailing
list.
I
One
point
is:
when
we
do
a
60-odd,
the
6p8
command
always
returns
a
success
and,
depending
on
on
the
outcome
of
the
art
of
the
add
command
it,
it
responds
a
list
of
the
cells
that
had
been
allocated.
For
example,
if
I
issue
an
ad
and
I
want
two
cells
and
both
of
them
are
allocated.
The
response
contains
its
circuits,
a
success
plus
in
the
list
of
cells
that
it
returns.
There
are
these
two
cells.
I
This
is
full
allocation
if
I,
if
I
issue
an
ad,
but
the
receiver
cannot
allocate
one
of
them,
but
only
one.
It
returns
that
cell,
so
I
know
its
success
and
I
allocated
only
one
cell
and
if
it,
if
it
none
of
them,
can
be
allocated,
then
it
returns
success.
This
is
the
current
thing
now,
but
the
list
is
empty.
So
there
are
some
complaints
on
the
implementation.
Saying,
okay,
buddy
I
need
to
look
at
the
list
and
I
know
because
of
the
size
of
the
list
is
zero,
that
I
couldn't
allocate
them.
I
But
this
is
a
kind
of
a
strange
case
because
you
are
saying
success,
but
you
cannot
allocate
any
so.
The
concern
here
is
whether
we
have
to
use
nor
s
response
instead
of
or
onea
or
another
another
code,
as
as
you
indicated,
thankfully,
so
that's
one
one
of
the
possible
things
that
we
can
improve.
I,
don't
know.
If
there's
any
comment
on
this.
H
So
this
is
ten
free
from
area.
So
here
is
the
no
racers
for
me
as
beginning
and
saying,
if
I
preserve
a
cell
like
five,
maybe
five
is
not
occupied,
but
my
slot
buffers
is
already
full.
I
cannot
reserve
you
more
so
that
case
I
return
no
resource.
So,
but
in
this
case
it's
different
from
that,
so
maybe
they
will
get
another
resort.
A
return
code
for.
C
Tomas,
so
if
we
add
an
extra
return
code
for
saying
we
have
no
space
at
all
I'm,
sorry,
no,
no,
a
return
code
for
saying
none
of
those
cells
in
your
in
your
cell
list
satisfies
us.
We
might
as
well
add
another
written
code,
three
for
the
same
partial,
because
I
think
we
try
to
I
mean
the
proposal
of
no
res
is
to
be
able
to
reuse
its
return
code.
So
we
might
as
well
make
some
specific
return
codes.
It's
we
have
one
byte
for
the
return
code
and
we
currently
have
like
five
allocated
right.
I
C
A
I
Okay,
this
is
another
point
hum
when
when
for
some
reason
a
generation
error
is
detected,
you
know
when
we
maybe
I
look
at
cells
and
for
some
reason
we
lose
an
AK
and
then
one
side
of
the
of
the
allocation
fails
and
one
side
allocated
cells,
but
the
other
one.
Not.
We
have
an
inconsistency
in
the
schedule.
So
in
this
case,
in
the
current
version,
we
recommend
that
when
this
inconsistency
is
detected,
we
just
we
recommend
is
not
mandatory,
but
we
recommend
to
clear
the
schedule.
I
I
It's
only
can
happen
because
the
previous
operation,
with
it
messed
up
or
we
lost
an
AK
or
we
lost
a
packet
and
once
one
of
the
sides
did
not
detect
that,
so
what
we
propose
to
cope
with
the
situation
and
enable-
and
it
will
be
policy-
so
we
don't
say
how.
But
we
enable
a
policy
to
correct
that
problem
is
change
a
little
bit
the
text
on
correcting
or
coping
with
generation
errors
and
saying
that
when
we
have
when
we
detect
this
error,
you
cannot
add
or
you
cannot
delete
cells.
I
Well,
you
cannot
add
or
delete
cells,
but
you
can
still
the
least
encount
operations
will
work.
Will
you
will
be
able
to
send
a
list
and
count
to
your
neighbor
and
then
the
return
of
these
of
this
operation
will
tell
you
what's
happening
and
then
you
can
correct
your
own
schedule
with
this
result
and
then
you
can
then
tweet
kind
of
cheat
on
the
generation
number
and
then
stay
in
sync
again.
I
So
that's
our
proposal
and
the
text
we
propose
is
the
one
in
the
middle,
the
two
bullets,
the
first
one
and
the
last
one
are
already
in
the
draft
and
the
one
in
the
middle
is
if
the
code
of
the
6p
request
is
count,
or
least
when
I
detected,
that
generation
error,
the
node
must
execute
the
operation
and
return
the
requested
values
and
this
can
be
used.
Despite
of
we
have
the
generation
error
to
correct
the
inconsistency.
So
that's
what
we
proposed.
We
don't
say
how
that
we
say.
We
support
that.
J
This
is
a
Tanaka
from
Toshio,
actually,
the
the
giant
generation
management.
The
works
very
well
they're
actually
too
well,
so
we
at
that
time.
We
just
only
thing
that
to
issue
the
clear,
so
my
proposal
is
maybe
sound
a
little
bit
too
radical
just
throw
away
the
general
generation
management,
so
both
peers
can
detect
the
the
last
transaction
fares
or
not
so
each
the
time
alt
should
occur
when,
for
example,
when
the
the
pier
cannot
receive
the
the
macaque
or
something
it
can
be
detector
and
also
the
count
or
list
the
command.
J
I
Thing
is,
if
you,
if
you
drop
this
mechanism,
you
are
forced
somehow
periodically,
maybe
to
issue
lists
or
counts
to
in
order
to
see
if
everything
is
is
correct,
which
I
think
this
is
safer
in
terms
of
consistency,
because
at
the
first
action
you
do
you
detecting
consistency,
and
you
can
react
quickly.
If
not,
you
can
like
the
version
yeah.
My
point
is
the
timeout
could.
C
A
list
so
so
Thomas
I
have
a
I
have
a
counter
example,
and
this
is
the
reason
why
we
put
the
the
the
counter
in
there
generation
the
problem.
You
know
yes
with
Max
and
all
that
you
can
detect
that
something
went
wrong
on
the
communication.
The
only
case
you
cannot
detect
is
when
an
AK
is
lost,
so
you
try.
You
send
me
a
response.
C
I,
send
you
an
act,
but
you
don't
get
the
AK,
and
this
is
your
last
retry
McCleary.
Try
so
I
think
everything's
fine
and
you
say
that
you
think
the
whole
transaction
has
failed.
So
now
what
happens?
You
generation
has
bumped,
your
Jerusha
has
not
bummed,
and
so
that's
the
consistency.
I
think
we
wanted
attack
with
that
with
that
mechanism.
Is
that
you
say
yes,
so
I
assume
this
cases.
You
fully
have
full
of
this,
but
you
still
don't
agree
and.
C
B
I
So
it's
my
last
light
on
the
next
steps
arm.
There's
a
little
thing.
I
want
to
do
on
its
to
clarify
the
use
of
the
ITF
ie
in
the
draft.
Now
it's
just
a
very,
very
simple
sentence:
I
went
I
want
to
clarify
how
this
II's
use
well,
how
its
encapsulated
will
be
just
one
more
line
explaining
how
to
use
it
and
then
resolve
all
these
to
black
test
outcome
issues
or
problems
that
we
have
identified
and
then
I'm
asking
for
a
word
group
last
call.
C
Sorry,
it
wasn't
paying
attention
so
so
I
think
the
changes
are
that
we
have
identified.
We
have
to
have
this
discussion
for
sure.
First,
the
the
the
draft
has
been
implemented
in
many
times.
It
works
well,
I
think
we're
very,
very
close
to
having
a
final
version.
I
think
we
need
to
resolve
the
process.
Outcomes
may
be
bumped
of
revision,
republish
this
and
then
I
propose
that
we
have
a
workgroup
last
call
add
on
the
mailing
list
right
after
this
0:08
is
published.
Does
that
work
yeah.
I
A
Would
like
to
add
that,
as
you
know,
this
draft
uses
information
element
III
happening
at
the
Mac
layer
and
that
the
hie
is
very,
very
well
that
we
are
doing
this
and
they
have
created
a
working
test
group
15:12,
which
is
the
equivalent
of
our
working
groups,
which
deals
with
the
link
layer,
communications,
it's
an
LLC
and
they
are
integrating
the
concept
of
6p
as
part
of
their
design
already.
So
they
know
it
exists.
They
they
included
it
in
their
design
and
I
hope
that
charlie
is
here
at
the
end
of
the
room.
Charlie
Charlie.
A
Can
you
help
me
so
so
I
hope,
maybe
Charlie.
You
can
tell
us
a
little
bit
more,
not
now,
but
at
the
end
of
the
meeting
during
the
I/o,
be
if
you
can
just
tell
us
a
little
bit
about
the
1512,
and
maybe
Bob
will
tell
us
more
about.
What's
happened
happening
globally
in
54,
but
I'm
just
saying
that
in
our
C's
day
and
it's
wrapping
this
this
operation
nicely
so.
K
L
M
M
Now
a
lot
of
revision
from
from
the
Commission.
Thank
you.
I
would
like
to
thank
javi
and
child
and
I
thank
Jean
who
have
made
a
very
detailed
revision
of
the
draft.
So
I
had
some
time
to
address
every
other
comments.
So
there's
a
lot
of
tickets.
We
have
the
address
so
the
next
question:
let's
see
what
happens:
okay,
I!
Thank
you.
M
M
Well,
have
a
collection
of
typos,
so
I
have
some
type
of
okay
and
every
other
problems
we
have
so
some
expression
is
that
in
an
extra
text
will
be
there.
So
those
are
all
the
details:
I
don't
tickets!
If
you
want
to
see
them
exactly
so,
I
just
focused
on
the
specific
tickets
were
to
do
the
concepts
and
mechanisms
that
were
defined
on
the
assets,
Europe
and
and
some
movement,
a
big
movements
off
of
text
between
them,
so
the
first
ticket
was
67,
which
does
for
the
some
a
couple
of
sections.
M
B
M
Second
and
68th,
but
the
difference
between
allocated
and
use
cells,
meaning
that
located
on
schedule,
cells.
That
means
we're
trying
to
measure
how
many
cells
we
are
using
from
the
ones
which
are
already
scheduled,
and
this
is
kind
of
aware
how
to
measure
the
ibaka
usage
of
that-
and
this
is
the
only
point
we
can
sense
and
that
there's
a
change
and
that
we
need
to
add
react
to
that
change.
So
that's
why
I
added
a
definition,
pacific
definition
page
and
then
the
different
is
taken
from
one
slot
frame
to
the
next
one.
M
So
we
react
in
the
universe.
Not
same
time.
See
so,
and
we
have
have
tried
to
clarify
that
we
only
work
on
transmission
cells,
the
XS
from
one
neighbor
to
the
other.
This
is
a
neighbor
to
neighbor
algorithm.
Only
then
we
are
going
to
the
over-provisioning.
Okay,
there's
a
define
and
not
like
that's
a
clear
definition
whatsoever.
M
Okay,
so
we're
going
to
attacked
it
anyway,
but
we
want
to
take
more
cells,
are
required.
Okay,
this
is
was
the
problem.
We
wanted
others
on
this
time,
then
about
the
relocation.
Okay,
there's
another
one
of
the
mechanism
does
that
a
zero
are
also
also
addresses.
Is
there
a
location
of
cells,
so
location
is
clearly
defined
on
the
for
III
section
of
six
P
draft,
so,
which
is
the
mechanization,
would
I
should
first
over
there,
and
it
only
decides
when
okay
I
said
zero
only.
M
It
says
when
their
cells
should
be
relocated
according
to
the
packet
delivery
rate
value.
So
we
make
sure
the
hope.
The
the
average
buckle
give
a
rate
of
the
whole
scale
that
we
have
control
on
okay
and
there
could
be
any
many
as
functions
working,
so
they
want
the
part
which
is
belongs
to
this
scheduling
function.
We
calculate
the
average
and
then
we
compare
each
of
the
cells.
We
are
using.
Okay,
if
there's
an
increase
or
decrease
after
a
specific
threshold
and.
M
Well
and
of
course,
we
have
selected
randomly,
we
are
going
to
win
to
keep
this
very
simple.
So
that's
why
there
are
main
things
right:
get
random
within
the
the
dissection
is
just
a
random
here:
there's
no
specific
criteria:
how
to
define
which
cells
are
going
to
relocate
use
and
there
are
no
transmissions
again.
Okay,
we
react
only
if
the
transaction
was
unsuccessful.
We
only
retry
afterwards
because
a
requirement
still
exists
so
there
in
order
to
keep
it
simple.
Okay,
again,
we
don't
use
the
transmissions
a.
M
Shall
we
check
here
the
number
of
cells
which
are
scheduled
to
each
of
the
neighbors
and
the
number
of
you
cells
against
each
of
the
neighbors
so
in
case
there's
a
change,
because
once
that
frame
and
the
next
one
we
detected
and
react
accordingly,
we
were
asked
also
to
add
a
diagram
how
it
works.
So
there's
a
trigger
event
over
there,
so
we
collect
number
of
sides
to
understand
which
are
the
number
of
cells,
which
is
the
number
of
sides,
in
fact
that
we
are
willing
to
use.
M
We
add
the
over-provision
value,
know
that
two
attacks
in
the
future.
If
you
need
more
of
cells
or
not,
and
and
then
did
we
pass
this
value
to
the
allocation
policy
as
well,
they
repair
cells.
But
so
what
happens?
Our
question
was:
what
happens
when
the
over
position
is
zero?
For
example,
say
if
the
provision
is
zero,
we
cannot
detect.
Is
there
any
more
use
cells
if
you
are
required,
if
there
and
possibly
that
we
generate
a
packet
loss
in
the
queue?
M
So
that's
one
of
the
problems
way
of
using
another
provision
of
zero,
so
we
just
had
a
fixed
value
for
our
provision,
which
is
implementation-specific
and
then
the
the.
So
there
are
two
thresholds
here.
The
two
values
fact
not
result
the
over
provision
and
SFC
ofso.
Then
there's
no
relationship
whatsoever
between
them.
We
try
to
keep
the
both,
thus
estimation
algorithm
and
the
and
the
allocation
policy
independent
between
them.
M
So,
even
though
the
within
the
whole
mechanism,
then
maybe
our
established
our
relationship
between
them
at
the
beginning
or
we
don't
define
any
any
relationship
between
them.
So
there's
our
own
provision
and
there's
to
be
a
threshold
at
the
design
when
we
are
going
to
change
over
to
add
more
cells.
M
This
location
policy,
they
won't
define
how
many
and
the
other
way
size
decide
when
to
make
a
change
so
about
the
cell
list.
Error
handling,
so
f
0
does
not
handle
those
errors,
meaning
that
we
only
if
you,
if
there's
the,
only
handle
what
we
are
doing.
Okay,
if
there's
the
unsuccessful,
we
try
again
because
the
requirement
is
still
active,
and
this
is
of
course
I
rather
be
chosen.
M
We
have
the
we
have
defined
a
per
transaction
timeout
value
because
you
are
required
to
assess
scheduled
function.
However,
ok,
there
were
still
under
discussion
apparently
that
if
the
time
of
value
should
belong
to
one
site
go
to
the
app
and
the
packet
delivery
rate
definition,
there
was
no
definitions,
beginning
of
how
to
measure
the
packet
linear
rate,
so
it's
calculated
per
cell
okay
as
a
percentage
of
the
number
of
our
knowledge.
M
M
There
are
the
initial
value
of
scheduled
cells,
implementation
specific
then
at
least
a
sub
0
threshold
number
of
cells
must
be
allocated
to
each
of
the
neighbours,
meaning
that
this
is
the
initial
value
that
we
didn't
define
me
in
the
meaning
and
the
gear
so
at
the
beginning,
and
we
should
be
defining
in
order
to
start
this
and
to
bootstrap
the
process.
So
they
we
need
at
least
to
have
this
number
of
cells
to
start
the
process
and
oops.
And
finally,
if
you
have
any
questions
about
this,
thank
you.
B
C
M
H
C
C
Think
they
go
per
our
discussion.
You
know
yeah
last
IETF
as
well.
Yes,
I
mean
these
edit.
What
you
presented
our
editorial
changes,
making
things
clear
and
we
can
keep
doing
this
for
ITF
after
idea,
verse,
right,
yeah
right.
What
is
needed
is,
is
you
know,
I'm
I'm,
a
company
and
I.
Look
at
this
and
I
say
you
know:
I
have
a
smart
dis,
smart.
That
application
is
sm0
going
to
be
good
enough.
I
mean
how
much
you
know
if
I
switch
on
my
notes.
C
How
long
will
they
take
to
join
if
I
send
it
back
it?
What's
the
average
max
min
max
Lane
see
how
much
data
we
need
that
kind
of
results
you
know,
especially
because
this
is
zero.
So
this
is
the
this
is
the
default
one,
and
so
we
want
to
be
it
to
be
good
enough.
So
I,
you
know
again
we
need
evaluation.
So
so
it's
good
it's
good
to
their
implementations,
but
there
needs
to
be
effort
and,
as
editor
I
mean
I
understand
that
you
might
not
be
I
mean
or
you
might.
C
B
A
B
B
A
A
There,
how
do
you
said
that
the
last
time
we
had
the
discussion
about
the
opportunity
to
make
this
an
experimental,
RFC
and
I
like
this
hedger,
because
now
we
have
we
have
written
to
the
world
a
there
is
a
base
on
which
we'd
like
you
to
be
out.
There
is
what
we
would
like
you
to
experiment
with
and
be
on
about.
If
we
don't
publish
it,
then
people
will
say
ok,
there
is
this
thing,
but
I
don't
know
where
it
is
I,
don't
know
what
state
it
is.
It's
a
choice.
A
It's
also
there's
a
time
when
we
are
happy
enough
to
kind
of
cast
in
stone
that
okay,
it's
not
a
solid
track
document,
but
at
least
it's
the
base
on
which
we
want
to
work
and
I
hadn't
got
the
impression
that
last
time
that's
pretty
much
where
we
well
the
idea
of
finishing
this
as
experimental.
Now,
if
you
care
about
the
name
sf0
looking
lightly,
that's.
A
Of
that
could
be
to
rename
it
but
having
publishing
this
this
base,
let's,
let's
use
that
as
the
base
of
the
work
about
you
know
how
we
make
those
decision
of
allocation,
de-allocation
I
think
we
need
it.
I
think
it's
good
to
throw
that
at
you
know,
Academy
on,
say
a
all.
Those
points
are
fantastic
research.
Please
help
so.
C
Tomas
says
you
know,
this
is
a
discussion.
We
need
to
have
a
scale.
Of
course.
I
have
a
I,
have
a
problem
having
the
0
SF
being
experimental,
I
mean
what
message
does
this
send
out?
I
yeah
this
is
a
you
know.
All
the
work
we
do
here
is
is,
you
know,
is
for
industrial
critical,
that
net
kind
of
things
and
having
the
default
won't
be
experimental.
You
know,
doesn't
stand
out
yeah,
so
it's
where
she
had
some
so
input.
Jvic.
I
You
have
it.
No
ok,
so
just
another
point
what
it
would
be
good
to
evaluate
the
interaction
of
SF
with
six
be-because
6p
has.
B
B
I
D
Suresh
krisshnan,
so
if
not
this,
then
what
right?
So?
What
becomes
as
a
zero
so
like
we
have
do
we
have
something
in
play?
I,
don't
know
right,
like
so
think
of
that
first
thing.
Second
thing
is
like:
if
you
want
to
make
this
experimental,
that's
okay,
but
I
would
like
to
see
what
the
experiment
is
right.
Sure
like
what
are
you
actually
trying
to
collect,
like
the
trap?
D
Doesn't
have
anything
like
that
so
like
just
putting
something
in
experimental
like
previously
like
a
couple
of
years
ago,
used
to
be
a
way
of
lowering
the
bar
of
review,
saying
like
oh,
this
is
experimental.
Like
you
know,
people
take
a
lighter,
read
of
the
draft.
I
didn't
I,
use
you
and
everything,
but
I
think
it's
more
valuable
to
actually
document.
What
is
the
experiment?
What
are
you
gonna,
collect
and.
G
D
What
are
you
gonna
do
with
it
right
and
if
you
decide
to
go
the
experimental
path
and
I
will
support
it.
Just
make
sure
that,
like
you
know,
what
do
you
want
to
figure
out
like
you
know
like
you,
what
are
the
things
you
want
to
collect
and
how
you
want
to
bring
it
back
to
Sanders
track
in
the
future
yeah.
C
O
I
O
So
we
published
0:03
version
on
June,
15
and
I
was
mostly
busy
with
the
implementation
efforts
in
open
wsn,
and
the
good
news
is
that
we
have
no.
As
thank
very
sad.
We
had.
We
have
an
implementation
complete
in
open
WS,
an
environment
and
an
ongoing
implementation
in
Kentucky
base
and
pre-shared
keys.
I
will
go
through
the
summary
of
updates
that
we
did
in
all
three
since
the
last
IDF,
and
then
I
will
show
some
of
the
updates
that
I
did
that
I
intend
to
do
based
on
the
implementation
experience.
O
So
first
there
is
a
couple
of.
There
was
a
couple
of
exchanges
on
the
mailing
list
regarding
the
communication,
how
communication
proceeds
during
the
drawing
process,
so
I
wanted
to
clarify
this
and
I
will
make
an
attempt
to
clarify
it
in
the
draft.
Basically,
we
have
a
pledge
which
is
or
in
all
terminology
the
joining
node,
which
is
pre-configured
with
locally
relevant
credentials,
such
as
a
pre
shared
key
or
a
row
locally
relevant
certificate.
O
We
have
a
1:1,
hop
radio,
neighbor
of
the
pledge
or
the
three
terminus
joint
proxy
and
the
join
register,
and
the
coordinator,
which
is
the
central
entity
that
manages
the
drawing
process,
the
pledge
in
the
drawing
proxy
communicate
over
a
link.
Local
addresses-
and
this
will
be
important
for
the
later
slides-
is
to
note
that
pledge
only
needs
to
know
the
address
of
the
joint
proxy.
O
The
drawing
proxy,
however,
needs
to
know
the
address
of
the
jrc,
and
there
was
one
issue
that
I
would
discuss
later,
how
DRC
finds
out
the
address
of
how
drawing
rocks
he
finds
out
the
address
of
the
jrc.
The
path
from
the
pledge
to
the
drawing
proxy
Islay
is
insecure
in
what
concerns
layer,
2
security
and
it's
passes
over
secure
and
if
I
and
the
path
from
drawing
proxy
to
the
Dre
IRC
is
the
mesh
part
that
is
secured
a
layer
to
secure
a
15-4.
C
O
O
O
The
ad
hoc
D
etat
can
shake
where
jrc
can
respond
with
an
optional
sac
which
is
defined
in
coop
as
meaning
that
okay,
hang
on
I
will
respond
to
you
later
on
in
the
process,
and
this
can
be
useful
in
in
the
use
cases
where
are?
There
are
many
concurrent
requests
by
multiple
pledges
on
the
GRC
and
the
GRC
wants
to
alleviate
the
alleviate
the
load,
that
is
on
the
network
by
signaling
those
to
the
pledge
to
hold
to
hang
on
and
wait.
O
This
is,
of
course,
optional
with
pre-shared
keys
and
mandatory
with
a
symmetric
keys.
The
Hinda
handshake
is
so
right
now
to
come
back
to
the
initial
point
of
how
pledge
or
the
drawing
proxy
learns.
The
JR
sees
address
as
ice
drawing
proxy
essentially
plays
the
role
of
a
co-op,
proxy
and
stateless
lis
forwards.
The
requests
and
realized
the
responses
back
to
the
pledge
for
this.
O
We
use
an
option
that
is
defined
in
my
co-op
option
that
is
defined
in
minimal
security,
that
we
term
stateless
proxy
option,
and
but
the
issue
is
how
the
drawing
proxy
knows
the
ipv6
address
of
the
GRC
and
thanks
to
Michael.
Oh,
if
I
remember
right,
we
settle
down
for
this
approach
where
basically,
the
the
the
address
of
the
GRC
is
returned
in
the
join
response.
O
So
basically,
when
a
pledge
drawings
in
order
to
fully
complete
the
drawing
process,
it
needs
to
receive
at
the
I/o
with
the
dog
ID,
and
then
it
configures
the
the
address
of
the
GRC
and
one
assumption
there
is
that
the
dagger
root
is
pre-configured.
With
the
address
for
the
whole
process
to
kick
off,
we
had
the
discussion
in
the
security
design
team,
about
MTI
algorithms
to
support
and
Inter.
We
mostly
discussed
the
symmetric
part
and
we
we
quite
easily
settle
down
for
the
CCM
star
equivalent
in
cozy.
So
this
is
a
SCCM
16,
64
128.
O
It
has
eight
byte
authentication
tag
and
in
terms
of
non
non
non
flanked
it
corresponds
to
the
15
for
CCM
so
that
we
can
reuse
the
same
hardware
for
symmetric
encryption
on
the
application
layer,
as
we
do
for
link
layer
security
in
terms
of
hash.
We
said
we
in
the
draft.
Currently
we
mandate,
sha-256
and
in
terms
of
asymmetric
keys.
We
mandate
the
P
256
elliptic
curve
and
ECDSA
signature
algorithm,
but
this
is
prone
to
change.
I
would
like
to
discuss
this
further,
maybe
during
the
meeting.
O
So
let
me
come
back
to
the
implementation
status
in
the
open,
WS
n
ecosystem.
We
implemented
the
minimum
security
with
pre-shared
keys
and
we
finished
the
implementations
of
Oh
boss,
co-op
in
Python
and
minimal
security.
The
jrc
part
in
Python,
as
well
as
the
embedded
part
or
squat
in
C
and
minimal
security
Oh
tree
in
C,
in
what
concerns
pledge
and
the
drawn
proxy
and
the
draft
is
fully
implemented
apart
from
the
asymmetric
variety
in
contour,
is
also
an
independent
of
implementation
ongoing
in
Kentucky
that
we
try
to
test
during
the
practice.
O
So,
in
terms
of
the
implementation
experience,
I
found
a
couple
of
issues
when
trying
to
implement
this.
The
first
issue
was
regarding
the
packet
size
and
the
most
critical
was
the
drawing
response
that
comes
down
from
the
from
the
dag
route
to
the
first
hop
into
the
into
the
network
because
of
the
source,
routing
header
and
basically,
with
this,
the
due
to
the
join
response.
Payload.
If
we
assume
no
fragmentation
in
the
network,
we
are
hitting
the
maximum
limit
of
the
network
in
terms
of
the
depth
that
we
can
reach.
O
So
this
is
so
it's
extremely
important
that
we
optimize
this
in
order
to
build
deep
networks,
and
my
my
understanding
is
that
wireless
heart
for
the
moment
goes
up
to
eight
hops.
Dibs
I
was
able
to
test
with
six
hop
with
five
hops
in
the
with
six
hops,
but
from
the
notes
way
from
the
same
manufacturer,
which
you
know
which
compresses
the
source
routing
header
quite
efficiently.
So
sorry,
some.
C
C
The
packets,
okay,
so
assuming
there's
no
fragmentation
in
my
packet
I
have
in
yeah
Paquette
I
have
a
source
route
which,
which
lists
the
different
hops
to
go
through,
followed
by
a
bunch
of
headers,
followed
by
the
encrypted
Oscorp
payload
right
and
yes,
okay
and
so
is
is.
Is
there
anything
we
can
do
to
that
or.
O
O
About
that,
yes,
I
have
done
the
next
slide.
I
have
the
dissection
and
the
packet
dissection,
basically
with
the
join
response
in
the
case
with
five
five
hops
downwards
and
as
you
can
see,
we
have
the
source
routing
header
at
the
beginning,
and
these
are
some
modifications
that
I
had
to
do
in
order
to
make
it
work.
So
I
use
the
token
length
set
to
zero
in
coop,
which
is
which
is
allowed.
B
O
Co-Op
spec,
which
is
allowed
in
coops
work
in
certain
circumstances
that
fits
very
well
with
the
drawing
process.
I
removed
the
content
format
option
from
the
draft
in
my
local
copy
I
still.
This
is
still
not
present
in
all
three,
so
the
join
response
does
not
contain
the
content
format
option,
but
this
is
implied
I,
think
from
the
in
terms
of
pledge,
because
we
are
defining
the
protocol
on
top.
Yes,
it
does
contain
a
URI.
The
join
response
does
not.
It
contains
the
message
ID
the
message:
ID,
the.
B
F
A
Is
pascal-
and
there
are
two
venues
that
we
can
look
at
now-
to
make
this
a
bit
deeper
without
fragmentation.
First,
one
is
is
the
way
you
allocate
your
addresses
and
the
way
you
build
them
and
the
way
they
are
because
I
see
there
that
you
seem
to
be
using
a
six
relish
with.
Is
that
four
octet
that
you're
using
so.
A
Yeah
it
is,
but
the
question
is,
you
know
there
are
multiple
ways
to
compress
the
address.
What
makes
them
more
compressible
is
if
they
share
more
bytes.
You
know
if
everybody
builds
on
the
same
thing
and
the
best
way
to
have
everybody
build
on
the
same
thing
is
really
linked
to
what
you're
doing
it's,
because
if
you,
if
you
harm
the
short
address,
yeah.
L
A
You
can
derive.
You
can
make
sure
that
our
unique
along
a
path,
because
this
is
the
route
to
win
both
does
everything
in
its
own
right.
So
you
can
derive
shot
address
every
hobbies
to
work.
That's
but
I,
don't
think
I'm,
seeing
hops
off
to
act,
I
see
I,
don't
know
how
to
read
what
you're
right
but
I
think
that
you're
using
frog
that's
per
hop
and
if
you
read
your
eye
stuff
from
the
shorter
dress
and
you
sign
the
shop
to
try.
So
you
know
you
think.
A
A
There
is
a
draft
in
the
making,
a
troll
which
is
basically
saying
okay
you're
doing
on
storing,
because
you
know
we
don't.
We
know
you
don't
have
so
much
room,
but
actually
actually
I
can
shorten
the
sauce
rot
headers
by
having
just
a
little
bit
of
state
that
the
route
decides
to
install
into
particular
devices.
A
So
now
it's
not
like
an
explosion
of
state
because
the
route
can
maintain
how
much
state
it
installs,
but
that
you
can
really
use
that
to
have
a
little
bit
of
storing,
which
again
will
shorten
your
right
path
and
in
part.
Well,
you
can
decide
where
you
place
this
debt,
etc.
That
would
be
a
limit
Michael,
because
these
things
finds
us
a
tree
and
at
some
point
you
know
you
don't
have
as
enough
state.
So
you
can't
you
can't
shut
down
every
path
to
everybody,
but
that
that
can
be
a
tool.
A
If
your
network
is
not
well
balanced
and
you
have
a
direction
where
a
lot
of
things
go,
then
you
can
shorten.
You
know
this
and
I'm
thinking
particular
in
am
ia
more
use
cases
where
you've
got
those
long
lines
of
matters
along
a
straight
and
typically
for
that.
The
rod,
projection
thing
will
save
a
lot,
and
then
you
can
you
without
the
fragment
connection.
You
can
authenticate
all
the
metals
along
the
street
without
taking
a
problem
between
so.
O
The
good
news
is,
this
is
orthogonal
to
the
draft
I
mean
this
is
showing
the
full
dissection,
but
it's
orthogonal.
What
is
not
implemented
in
on
the
implemented
in
our
implementation
currently
is
the
assignment
of
short
analysis
to
the
stack.
So
this
will
further
save
but
I
think
it's
already
good
news
that
we
can
go
down
five
hops
without
without
fragmentation.
Yes,.
B
C
Just
just
a
clarifying
question:
I
think
we
have
to
stop
after
that.
So
the
number
is
the
five
hops
that
is
assuming
eui-64
s
are
used
in
source,
a
header.
I
understand
you,
then
allied
them
if
the
beginning
is
the
same,
but
this
is
assuming
you
are
60
or
64-bit
addresses
right
to
inquire.
Yes,
link
layer,
yeah,
yes,
okay,
so
this
will
go
down.
C
O
This
is
another
thing
that
we
can
improve.
Basically,
this
is
dissection
sibour
dissection
of
the
join
response,
sabor
decoding
of
the
drawing
response,
and
we
are
currently
using
26
bytes
to
encode
16,
byte
key
and
one
byte
key
ID.
So
it's
quite
an
overhead
and
one
proposal
would
be
to
use
the
binary
encoding
to
compress
the
drawing
response
in
something
like
what
they
did
in
or
scope
with
the
compressed
cosy.
So
that
would
be
one
proposal
that
I
would
like
to
discuss.
Also
during
the
meeting
and
the
sec.
O
This
shoe
that
I
find
quite
important
is
the
drawing
proxy
policy
and,
although
I
understand
that
we
will
not
stand
up
as
a
policy
in
ITF,
there
is
a
problem
of
how
join
proxy
accepts
those
insecure
layer.
Two
frames
at
the
during
the
joint
process,
because
in
the
implementation
currently
I
had
to
allow
it
from
any
node
that
I
haven't
seen
before,
and
this
can
be
easily
spoofed
such
that
an
arbitrary
node
can
inject
packets
into
the
network.
O
O
A
Pascal
again,
I
think
it's
a
very
classical
to
have
a
few
word
in
a
draft
like
that
with
which
size
throttle
like
never
accept
any
quantity
in
your
amount
have,
and
it's
not
necessarily
a
value,
but
that
there
is
something
which
says
Oh
any
sense.
Sensible
implementation
will
will
have
a
mechanism
to
short
out
the
request
you
just
take
as
many
per
unit
of
time
or
so
oh
it's
one
sentence.
You
want
it
yeah.
C
O
So
that
would
be
it
so.
Basically,
the
PSK
variant
is
quite
stable
and
implementation
ready
and
we
have
the
implementation
and
open
WS
and
complete.
We
settle
down
for
ad
hoc
roles
in
the
automatic,
but
it
we
have
yet
to
implement
this
and
I
will
publish
all
four
draft
with
implementation
experience
before
what
I
hope
to
be.
The
working
group
last
call
and
the
reviews
are
of
course
very
welcome.
A
A
quick
note
that
at
the
moment
there
are
two
drugs
that
are
that
we
depend
upon
to
have
this
work
and
maybe
Michael
can
come
out
a
bit.
But
there
are
debates
at
this
moment
and
there
will
be
a
lunch
on
Wednesday
about
whether
a
dark
in
particular,
and
then
there
is
a
draft
I
Peter
as
well,
that
we
need
whether
they
should
have
a
home
at
the
ATF
and
without
the
F
will
endorse
that.
This
work
continue
continues
and
obviously
will
be
pushing
for
this
thing
to
happen.
Right,
yeah,
okay,.
F
Hi,
so
that's
not
the
right
side
here,
interesting
they'll,
never.
F
In
the
end,
so
I
have
to
look
here
all
right,
so
I
get
my
car
in
front
of
me.
So
the
goal
is
zero
touch
joint
protocol
and
it's
supposed
to
look
very
much
like
the
enema
brewski
would
work
but
optimized
for
six
Tish
and
minimized
and
you're
using
methods
and
protocols
that
we
already
have
code
space
for
so
in
news.
The
Enver
voucher
document,
which
is
the
document
that
allows
in
zero,
touch
the
pledge
to
understand
that
this
is
in
fact
the
network
that
it
belongs
to.
F
These
are,
in
fact
the
droids
you're
looking
for
is
is
actually
I
said
almost
working
group
last
call,
but
actually
it
it
written
group
last
call
was
over
and
it
was
actually
supposed
to
be
done
in
parallel
with
this
working
group,
and
that
was
actually
a
mistake.
So
that's
very
good.
All
right,
I
think
it
may
take
it's
been
through.
I
think
it
may
take
a
little
while
to
get
through
the
iesg,
but
I
anticipate
by
the
end
of
the
summer.
I
think
it
probably
will
be
out.
So
that's
good
for
us.
F
The
anima
burski
document
was
rewritten
in
april/may
and
you
can
come
tomorrow
and
learn
about
that
if
you
like,
but
the
key
thing
is
that
the
the
the
structure
of
it
is
now
stable.
It's
about
15,
page
short
pages,
shorter,
that's
always
good,
and
our
document
which
parallels
it
now
has
to
be
rewritten
in
that
same
style.
And
so
that's
on
my
to-do
list.
F
F
F
So
the
next
steps
are
essentially,
as
I
said,
we're
rewriting
brueski
with
with
with
six
Tish
in
mind,
there's
already
a
document
that
we
have
already,
but
you
know
it
it.
Essentially
it's
been,
it's
been
cut
up
into
pieces
and
and
pieces
have
gone
elsewhere.
I've
gone
into
six
Tish
minimal,
I've
gone
into
other
places,
I've
gone
into
Bruschi,
and
things
like
this,
and
so
now
it's
time
to
rewrite
a
document
that
basically,
we
actually
can
pass
as
opposed
to
just
a
collection
of
texts
and
ideas,
and
essentially
that's
it.
So
you
see.
F
So
when
you
read
the
documents
you
should
see,
essentially
you
should
be
able
to
put
the
two
documents
next
to
each
other
and
say:
oh
yes,
so
in
brewski
for
big
equipment,
you
do
this
and
for
small
equipment.
You
do
this
and
there
would
be
a
kind
of
a
one-to-one.
Then
that
should
be
I'll
just
make
everyone
happy
so
that
they
say.
Okay
is
essentially
the
same
protocol
conceptually,
but
on
the
wire.
If
the
bits
are
are
much
smaller
and
faster
or
slower,
yeah.
So
small.
F
The
bits
are
the
saint
that
the
same
number
of
bits
but
they're
smaller.
Yes,
that's
right
and,
and
I'd
like
to
change
I'd
like
to
change
the
document
to
to
include
the
word
zero-touch
in
the
title
and
actually
I'd
like
to
change
the
the
the
internet
draft
name
to
include
that
as
well.
I
think
that
will
be
simpler,
but
that
probably
that
may
require
the
blessing
of
the
chairs
or
maybe
they'll
describe
decide.
F
The
whole
working
group
has
to
decide,
but
so
what
I
would
like
to
do
is
change
the
name
of
the
document
so
related
document.
So
with
some
some
things
that,
as
we
were
working
on
both
six
dish,
minimal
and
zero
touch,
we
realized
there
are
some
things
that
were
not
specifically
in
related
to
the
security,
but
that
the
security
needed.
So
the
two
of
them
are.
F
We
have
one
is
this
enhanced
beacon
document,
and
so
this
is
an
additional
information
element
which
we
can
now
allocate
care
of
that
RFC
that
you
announced
the
beginning
of
the
meeting,
and
so
we
can
put
some
additional
information
in
the
beach
and
so
the
specific
things
that
we
needed
to.
We
wanted
to
put
in
was
essentially
there's
a
bit
that
says:
I
am
a
joint
proxy
okay
and
there's
another
bit
that
says
I'm
a
router
so,
which
means
please
don't
send
a
if
you're
a
leaf
node,
not
a
routing
node.
F
Please
don't
send
a
router
solicitation
eating
up
a
multicast
saw
a
slot.
You
can
unicast
one
to
me
and
I'll
answer
you,
okay,
and
that
saves
a
lot
of
you
know.
Multicast
broadcast
lots
are,
are
pretty
pretty
precious
and
we
don't
really
want
to
use
them
up
for
stuff.
We
have
to
them
for
da
O's
and
so
the
the
specific
enhancement
that
the
join
part
was,
in
fact
a
preference.
F
You
know
how
how
willing
am
I
to
to
allow
pledges
to
join,
and
so
the
suggestion
is
essentially
is
that
if
there's
a
maximum
number
255
that
it
means
I'm
not
willing
to
have
proxies
joining
and
as
a
result,
if
you
propagate
this
through
the
beacon
from
the
DOE
Dagon,
if
the
DOE
tag
happens
to
say
255,
then
at
the
root
says
that
then
everyone
else
is
gonna
have
to
say.
Well,
obviously,
the
DOE
tag
doesn't
want
anyone
joining.
F
So
none
of
the
rest
of
us
should
let
anyone
join,
and
so,
in
effect,
the
doubt
the
the
ripple
root
can
turn
off
join
processing
through
the
thing
it's
a
good
idea
or
not.
It
seems
like
a
good
work
for
me.
Such
information
could
also
be
in
the
rippled
eio,
as
another
possibility
crippled
the
iOS,
don't
necessarily
propagate
us
as
often
sort
of
a
Michael
sort
of
a
mix.
Yes,.
F
A
A
A
Got
this
compression,
which
is
more
efficient
if
the
route
is
a,
you
need
to
know
that
somebody
has
to
signal
it
and
it
can
only
be
the
route
some
point
we
secure
those
things
and
the
route
will
securely
say:
I
am
a
jersey
of
Netaji
us.
It
belongs
to
what
the
route
exposes
and
what
to
the
device
is
for
now.
If
you
want
to
be
able
to
announce
part
of
that
to
the
joining
node.
Yes,
you
take
it
from
the
DAO
as
proxy.
A
F
F
So
so
the
the
the
scenario
is
that
that
I
had
envisioned
is
where
you
had.
Some
diagram
would
help
here,
where
you
had
some
dense
intermediate,
a
bunch
of
mesh
okay,
where
there
was
quite
a
number,
let's
say
two
to
two
ranks
down:
there's
quite
a
number
of
devices
and
the
device
says:
look:
I
can
here
six
other
beacons
willing
to
do
proxy.
My
battery's
low
I'm
not
going
to
announce
okay,
because
I
can
hear
that
there's
other
other
will
other
willing
participants
who
have
obviously
higher
more
battery.
F
C
The
drunk
Rd
increase
increases
higher
join
parody,
actually
means
less
priority
right,
so
higher
number
is
less
priority.
I
I
think
you
know
the
joint
proxy
is
a,
and
this
is
my
personal
opinion.
That's
why
I'm
here.
So
the
joint
proxy
feature
is
a
15-4
ET
SCH
feature
so
III,
don't
see
why
we
should
start
mingling
this
with
ripple
and
what
not
I
mean
I
can
run
a
54,
Network
54
HT
ACH
network
without
ripple,
and-
and
you
know
why
are
we
doing
this
like
across
everything
so
so
propagate
this
on
off.
C
B
A
It's
mostly
I
mean
basically
the
view
of
it
and
then
the
delay
of
to
view
of
it.
So,
yes,
we
keep
them
independent,
but
to
find
sense
like
just
the
ability
to
turn
it.
On/Off
kenseth
make
sense
that
the
global
to
your
dark
level
right,
because
the
GOC
wants
to
be
a
G
or
not,
or
it's
busy
D
at
this
time
and
and
it
wants
to
come
back
and
and
find
since
there
are
just
too
many
things
coming
in
from
that
particular
network
announced
rattle.
A
It
I
don't
know,
maybe
you
your
link,
for
instance,
the
the
preference
with
with
kind
of
the
throttling
level.
So
by
pushing
your
di
OS
and
the
lower
preferences
you
will
ask
the
distributed
process
is
to
throttle
more
and
there's
a
lot
of
thing.
You
can
do
if
you
signals
for
repo
how
to
do
it,
but
that's
the
repo
level,
that's
a
layer,
3
global
level
and
then
at
the
local
level.
The
node
can
take
this.
You
know
doing
even
less,
but
it
cannot
take
this.
This
is
not
doing
more.
F
O
Is
malicious,
so
I
would
just
like
to
say
that
I
agree
with
the
layer
to
view
because
the
the
problem
with
the
the
implementation
was
only
how
to
configure
the
exempt
flag
in
802
the
15.4,
so
basically
when
to
accept
insecure
packets
at
layer,
2
and
everything
else
is
taken
care
of
by
the
ipv6
forwarding
and
the
the
join
protocol.
So
I
would
rather
do
it
very
simply
with
the
enhance
beaker
with
the
flag,
depending
only
locally
on
the
join
proxy.
A
A
Well,
maybe
the
level
is
depends
on
network
wide
metrics
like
how
many
people
are
trying
to
join
at
this
time,
so
each
GN
should
slow
down
more
because
too
many
people
are
coming
here
and
that's
not
something
that
an
individual
layer
2
can
know,
and
the
one
thing
I
really
opposed
is
that
you
learn
something
from
your
parent
at
layer
2,
because
you
don't
know
in
which
direction
propagates.
You
need
to
learn
it
from
your
layer,
3
parent,
just
like
we
do
for
that.
Yeah.
F
That
may
be
a
really
good
point.
You
need
to
learn
it
from
your
layer,
3
parent,
so
from
a
process
point
of
view,
I
believe,
following
the
ionic
considerations
in
6550,
we
can
allocate
new
ripple
d
io
options,
that's
not
a
problem.
We
don't
have
to
go
to
role
to
do
that.
The
question
is:
should
we
do
it
in
this
document,
which
is
defining
a
layer
2
object?
Should
we
define
a
layer?
3
object
at
the
same
time,
because
we're
still
the
extensions
to
these
hands
beacon
for
the
pledge
to
listen
to
right.
A
D
F
Yeah
yeah
I'm
not
worried
about
that
process,
I'm
worried
about
what
the
you
know
about
handing
off
or
whatever
so
the
other
point
is.
This
document
has
not
been
adopted,
so
we
good
to
adopt
it
as
it
is,
and
then,
if
you
want,
we
want
to
have
a
tell
I
will
write
as
another
document
that
talks
about
the
layer:
three
option:
okay
and
then,
if,
if
you
want
to
make
it
one
document,
yeah.
C
F
C
C
C
F
C
F
Then
you
can
see
what
it
says:
okay,
because
I
think
that'll
be
two
pages
to
start.
So
the
other
document
that
came
out
of
or
was
removed
from
the
the
zero
touch
in
one
touch
document
was
rekeying.
Okay,
my
slide.
So
this
is
intended
to
be
a
Comey
based
thing,
leveraging
a
key
that
was
set
up,
ideally
through
through
the
OS
co-op
used
for
or
a
so
sorry
leveraging
the
key
setup,
either
through
ed
hawk
or
the
existing
keys.
That
were
one
touch.
Okay,
so
that
there's
no
additional
key
management
overhead.
F
That's
happening
this
way
how
that
works
as
remains
to
determine
so,
basically
it
is.
You
know
you
have
your
network
wide
key.
You
have
a
key
ID
for
it.
One
okay,
I
believe
we
have
in
minimal.
We
have
eliminated
key
one
key
to
I,
can't
remember
if
we
did.
F
So
so
that
the
there's,
if
we're
gonna
very,
we
have
to
return
two
key
IDs.
So
you
have
key
one
and
key
to
that.
Maybe
sorry
ID
one
and
ID
to
that
responds
to
six
dish,
minimal,
x',
key
one
key
to
k1
k2,
and
then
we
have
to
rekey
them,
which
involves
giving
everyone
a
new
set
of
keys
and
then,
when
the
dough
DeGroot
believes,
the
jrc
believes
that
it
has
communicated
with
enough
nodes,
preferably
all
of
them.
But
maybe
there
are
some
that
just
there,
no
one
exists
anymore.
F
It's
given
up
on
them
at
which
point
starts
using
the
new
keys
and
when
it
starts
using
the
new
keys,
that's
a
signal
to
all
other
nodes
that
have
the
new
keys
to
stop
using
the
old
keys.
So
we
have
a
basically
a
rolling
rekey
there.
So,
first
of
all,
the
first
step
is
that
everyone
gets
a
new
key
and
then
the
next
step
is
that
we
use
it
and
actually
the
third
kind
of
step,
as
we
tell
it
them
to
delete
the
old
keys,
but
actually
put
a
timer
on
that.
F
So
this
is
all
intended
to
be
essentially
Komi.
Based
which
is
simple
management
over
co-op
as
the
based
on
a
yang
model,
it's
relatively
small
and
simple
in
terms
of
bytes
used
and
that's
a
second
document,
so
Peter
and
I
have
done
some
work
on
that.
Nor
work
needs
to
be
done.
I,
don't
think
either
of
us
have
looked
at
the
document
since
last
IETF,
maybe
peter
has
I
don't
remember.
F
I
have
I
have
heard
that
that
the
situation
you
know
this
working
group
tried
to
use
Comey
for
6p,
and
we
weren't
happy
with
it
at
the
time
and
that
it
has
gone
in
the
last
four
years
gone
through
a
significant
change
and
it's
much
much
easier
and
smaller
to
deal
with,
and
so
that's
a
good
thing
and
it's
probably
the
right
answer
from
a
IETF.
Why
you
know
what
are
we
doing
are
with
management
kind
of
thing?
F
Why
would
you
need
to
rekey
well
either
because
you
actually
care
about
crypto
hygiene
and
want
to
recycle
your
keys
now
and
then
or
be
because
there's
some
node
in
your
network,
which
has
gone
bad
and
the
only
way
you
have
of
cutting
them
off
from
your
network
is
to
change
everyone
else
key.
So
that's
the
the
oh
I
need
to
move
move
I
need
to
get
rid
of
of
malar
from
my
network.
F
Okay,
so
that's
pretty
important,
but
rekeying
is
going
to
take
recognize
that
rekeying
means
talking
to
every
single
node
with
at
least
a
couple
packets
and
getting
a
response
from
them,
and
you
can
do
it
slowly.
It
could
happen
over
hours
or
even
days
depending
on
the
bitrate
of
your
network,
but
that
the
slower
you
do
it
the
longer
you
have
a
bad
guy
on
your
network.
Okay,
so
that's
the
like
the
caveat.
There's!
No,
the
document
doesn't
say
how
fast
to
do
it.
That's
up
to
you!
F
That's
gonna,
be
you
know,
quality
of
implantation
or
something
like
that
to
do
it.
So
that's
this
document
so
we're
completely
out
of
time.
Oh
okay,
I'm
sorry,
I
thought
this
was
earlier
in
the
document
and
well
that's
good.
So
that's
a
second
document.
Please
read
it
if
you
think
that
it
belongs
in
this
working
group,
then
I
would
love
to
have
a
mailing
list.
Discussion
about
that
yeah
and
who's.
Shabby
no
Simon
Seymour.
C
So,
thank
you.
Thank
you.
My
phone
is
so
weird,
no
more
questions
for
Michael.
Okay.
So
now
we're
entering
in
the
last
we're
right
on
time,
because
we
jumped
one
ten-minute
thing
so
we're
right
on
time,
we're
now
entering
the.
If
time
permits
and
time
permits
section,
we
have
four
new
work.
That
is
the
first
time.
I
believe
that
they're
being
presented
so
Seymour
has
a
scheduling,
function
called
a
SF
tour
zeroes.
Thank
you,
so
so
yeah
sorry,
one
thing,
that's
important!
Please
keep
your
time
you
have
ten.
So.
R
Switching
back
to
the
scheduling
topic
here,
what
we
are
proposing
is
a
scheduling,
function
called
AES
F,
which
stands
for
autonomous
scheduling
function,
where
you
have
two
main
concepts.
The
first
one
is
that
you
have
autonomous
slot
frames
or
spot
friendlies
cells
in
them,
which
means
that
in
the
SWAT
frame
you
have
cells
that
are
self
maintained
locally
by
the
node
only
by
the
knowledge
of
the
neighborhood.
So
the
basic
idea
there
is
that
the
at
the
node-
you
know
your
neighborhood,
and
you
know
the
MAC.
R
Your
neighbors
and
you're
going
to
use
the
MAC
address
and
to
hash
it
to
actually
find
out
a
slot
and
channel
offset
deterministically
to
communicate
with
that
node.
The
second
concept
in
a
SF
is
that
you
have
one
slot
frame:
Pro
traffic
plain
typically
one
for
the
teach
synchronization
one
for
the
routing
control
one
for
the
application,
but
you
could
have
a
bit
already
as
many
as
you
want
and
each
one
benefit
of
that
is.
That
is
so.
These
lot
frames
are
isolated
completely
by
design,
so
they
don't
interact
with
one
another.
R
In
that
context,
with
ripple
both
storing
an
on
storing
mode
abroad
and
or
not
trafficking,
now
this
comes
obviously
with
major
limitations
which
I
wanted
to
highlight
here.
It's
a
least
suboptimal.
For
few
reasons.
The
three
main
reason
are
on
the
slide
number
one:
the
cells
are
not
cascaded
along
the
past.
You
cannot
do
very
quick
forwarding
a
to
be
followed
by
B,
2,
B,
2,
C,
etc.
The
slots
are
all
shell
we
don't
have,
and
the
schedule
is
always
provided
for
the
worst
case.
So
you
going
to
be
you.
B
R
To
choose
between
energy
efficiency
incapacity,
you
cannot
dynamically
add
more
cells
for
traffic
between
a
and
B.
If,
suddenly,
you
need
more
traffic
between
A&B
I'm
going
to
show
here
a
brief
overview
of
each
of
the
three
different
types
of
such
frames.
We
have
three
types,
the
first
one,
the
simplest
one
is
absolutely
equivalent
to
the
sixties
minimum.
R
In
that
case,
you
have
a
single
slot
which
is
used
for
a
reception
transmission.
It's
a
shared
spot
and
all
know
that's
a
rendezvous
slot.
So
all
the
nodes
are
waking
up
at
the
same
time
same
channel
offset-
and
this
is
used
typically
for
unable
discovery.
It's
just
for
any
broadcast
communication
with
node.
You
do
not
know
yet.
R
So
it's
it's
used
most
often
at
the
defaults,
not
frame
and
forwards
in
control
data.
So
if
you're
on
the
side,
you
see
only
four
channel
offset
so
the
one
thing
that
for
each
of
the
stratum
you
have
in
a
staff,
we
actually
dedicate
a
subset
of
the
16
or
of
the
M
available
slots
for
channel
offsets.
So
here
in
that
example,
we
have
a
that.
That's
what
frame
has
four
channel
up
set
to
play
with
the
second
one?
R
Is
the
receiver
base
last
time
in
which
is
used
for
unicast
communication
in
that
one,
the
nodes
maintain
well,
they
have
one
static
receive
but
which
called
OBC
base.
They
have
one
static
receive
slot
here.
The
green
one
at
the
channel
of
certain
time
of
self
is
on
the
hash
of
the
of
the
MAC
address
of
the
node
and
then
for
each
neighbor,
which
is
unknown
by
the
886
neighbor
cash
for
each
neighborhood,
you
compute
the
hash
of
the
Mac,
and
that
gives
you
coordinate
sin
x
and
y
for
a
transmitted
shared
slot.
R
That
name
it's
a
shot
slot,
because
you
have
multiple
senders
every
one,
also
because
you
might
have
multiple
nodes
listening
on
the
same,
because
the
hash
is
not
clear
area.
The
last
one
is
a
sender
by
slot
frame
the
exact
same
thing,
except
that
now
the
cell
that
is
fixed,
is
for
sending.
So
you
have
always
once
sell
based
on
a
hash
of
own
MAC
address,
and
then
you
maintain
a
receive
cells.
R
For
the
other
note
that
you
want
to
listen
to,
and
that
is
typically
used
for
privileged
neighbors,
such
as
the
teach
time
source
or
a
poor
parent,
because
it
gets
more
costly.
If
you
do
it,
you
cannot,
you
could
also
choose
to
do
it
for
all
neighbors.
It
gets
more
costly
because
then
you
need
to
listen
to
all
neighbors,
it's
a
little
more
cost
in
energy,
so
this
is
all
put
together.
R
A
Yes,
well
there
is
this
habit
to
use
different
slot
frames.
Pascal's.
We
have
four
different
priorities.
I
was
just
wondering
since
you're
making
different
slots.
Fives
is,
if
you
keep
them
or
align
time
wise
or
because
there
is
this
risk
risk
of
a
collision.
So
at
some
point
you
would
say
a
what,
if
I
could
use
some
CSMA
in
there
and
I
was
just
wondering
if,
since
you
have
so
many
of
them,
you
constantly
use
them
as
a
priority.
Now
they
kind
of
most
of
them
are
the
same
priority
like
traffic.
A
Now
some
of
them
may
still
the
aya
priority
like
like
the
time,
and
you
may
be
thinking,
oh
those
which
are
higher
priority.
I
may
synchronize
them
a
bit
ahead
of
time
like
two
milliseconds
before
the
lower
ones.
So,
if
there
is
you,
you
can
still
do
some
form
of
CSU
mail
I'm.
Just
now
you
get
this
possibility,
I'm
just
asking
sure
using.
A
C
C
R
R
But,
on
the
other
hand,
we
also
have
different
channel
of
set
religious
zealots
frame.
So
whenever
you
start
listening,
you
actually
are
on
one
particular
channel,
so
you
will
not
start
hearing
from
from
one.
That
starts
a
bit
earlier
on
you.
You
will,
like
you
know
when
you
start
receiving,
which
one
you're
actually
listening
out
anyway,
so
that.
A
R
C
R
A
L
B
R
R
So
we
have
a
definition
of
configuration
parameters.
One
thing
that's
not
defined
yet
is
how
to
how
to
discover
that
configuration.
There
are
two
options.
The
preferred
one
currently
would
be
to
have
a
new
information
in
amount
in
EB,
so
because
it's
a
join
time
thing.
So,
as
you
join
the
network,
you
should
know
which
slots
friend
we're
gonna
have
and
how
to
maintain
them.
So
here
what
I
mean
in
the
world
that
tell
you
what
are
the
length
of
strata
frame,
which
which
says
our
receiver
base
Center
base,
which
one's
a
rendezvous
etc?
C
R
C
S
Hello
I
like
to
talk
about
these
six
teach
examples,
and
this
draft
was
but
this
document
was
written
like
two
years
ago,
then
the
protocols
have
evolved.
Some
other
protocols
appear,
so
we
needed
to
update
this
document.
The
goal
of
this
document
is
just
to
provide
a
reference
for
new
implementers
to
like
how
cystic
frames
should
look
like.
So
here
we
see
different
cystic
Franks
in
this
document.
S
Okay
and
what,
in
order
to
read
this
document,
we
have
we
use
two
tools:
we
use
open,
WM
project
using
the
open,
aureus
and
framework,
and
also
the
open,
dahlias
and
visualizer
as
a
simulation,
and
also
using
wireshark
the
latest
release,
which
implements
a
situation
dissection
and
six
stop
detection
as
well.
Here
we
see
the
date
apology
of
the
simulation
note,
so
we
use
three
nodes
in
simulation
being
not
one
day
de
bruit,
not
to
win
the
child
of
night
one
and
no
to
being
the
parent.
S
S
S
We
see
note
1
in
the
source
in
this
in
the
source,
and
then
we
see
here
the
destination
is
well,
not
3,
so
here
no
to
past
the
frame,
2,
node,
3
and
then
just
the
reply,
the
replied
going
for
3
to
2
and
then
the
reply
going
for
2
to
1.
Well,
this
is
a
very
simple
drop
of
yeah.
All
the
all
the
Cystic
frames.
S
C
Thank
You
Jonathan
I
want
to
add
a
couple
of
points
here
for
this
draft.
This
is
indeed
a
copy/paste
exercise
of
a
wireless
out.
Wireshark
output
agreed,
but
our
experience
has
been
we've.
We,
you
publish
this
a
couple
of
last
couple
of
eyes
years
ago.
It
was
used
extensively
at
blood
tests.
It
was
then
kind
of
forgotten.
We
didn't
update
it,
it
disappeared
and
at
the
new
block
test
we
had
again
the
same
questions.
How
does
this
header
fit
with
this?
C
What's
this
bit
with
this
byte
and
I'm
I'm,
you
know
I
I
I,
like
the
fact
that
it
shows
on
paper
exactly
how
the
headers
fit
one
after
the
other
and
how
exactly
the
six
large
fit
with.
You
know
that
kind
of
stuff,
so.
C
A
D
As
Christian,
so
so
personally,
right,
like
I,
leave
it
up
to
you
for
a
minute.
Okay,
but
like
I'll,
tell
you
what
Guilin
I'm
using
myself?
Okay,
when
I'm
talking
about
this.
Do
you
think
this
will
be
useful
in
five
years
after
its
published?
Okay?
So
that's
the
kind
of
like
how
I
judge
whether
something
would
be
useful
or
not.
D
So
if
you
want
to
have
a
new
deliverable
for
this
right,
that's
what
I'm
gonna
apply
so
I'm,
saying:
okay,
is
there
archival
value
for
this?
If
not,
no,
don't
do
it
as
a
separate
document
put
it
in
the
appendix
or
something
or
or
wiki
or
whatever
or
a
draft.
That's
maintained!
You
decide!
Okay!
Thank
you.
Okay,.
A
F
Would
really
be
awesome
and
it
would
also
to
address
what
suresh
said
it
would.
It
would
also
document
the
situation
as
of
now
as
of
that
document,
which
is
a
standards
track
document,
it
would
be
useful
in
five
years
when
somebody
goes
to
implement
something
that
needs
to
be
compatible
with
what's
with
what's
deployed,
and
so
even
if
we
do
things
after
that,
they
would
be
saying.
F
I
am
compliant
to
this
particular
use
of
ripple
document
in
my
data
plane
and
here's
some
example
packets
that
you
can
test
with
right,
so
that
actually
would
be
awesome
just
we
need
to
make.
We
need
to
make
sure
that
we
got
the
the
relationship
between
there
and
that,
if
there's,
if
there's
a
bunch
of
sixth
ish
stuff
there
that
it's
clearly,
this
is
layer
two.
So
if
you're,
not
six
dish,
don't
do
this
find
some
other
other
thing.
Thank
you,
yeah
I
think
Euler
and.
Q
So
assuming
that
we
have
our
wireless
topology
as
a
source,
this
destination
and
the
two
relays
here,
we
may
implement,
what's
called
promiscuous
overhearing
to
do
so.
Assuming
considering
that
we
have
the
wireless
reading,
which
is
by
nature
a
broadcast,
then
any
neighbor
may
overhear
this
transmission
right.
Q
This
transmission
will
take
place
in
two
different
unicast.
While
they
may
be
overseen
here,
we
need
to
pay
attention
on
the
ripple
vio
message,
but
potentially
we
need
to
be
extended
to
a
low
so
that
each
60
node
to
have
some
additional
information
about
their
potential
parents
and
a
list
of
parents,
but
we
may
need
to
have
what
we
call
here:
alternative
parent
next
by
employing
packet
replication,
it's
straightforward
that
we
are
having
duplication,
so
the
we
we
have
unnecessary
are
trafficking
the
network.
Q
We
need
to
employ,
what's
called
elimination,
to
reduce
this
unnecessary
traffic
to
do
so.
Each
sixties
node,
once
it
received
the
first
coffee,
the
first
copy
of
the
data
packet.
It
will
discuss
the
following
copies
here:
are
esprit
60s
architecture.
The
sequence
number
is
not
65.
Detector
does
not
take
position
about
how
the
sequence
projects
are
tracked
in
a
packet.
However,
it
comes
with
tagging
packets
for
floated,
if
occasion,
which
will
be
great
to
efficiently
apply
in
this
procedure
of
packet.
Q
So
we
have
things
to
toggle
with
with
multiple
receivers
the
acknowledgment
for
the
from
two
different
guys,
so
we
need
to
avoid
collisions
or
whether
we
need
we
need
to
discuss
whether
we
are
going
to
have
one
acknowledgement
to
acknowledgment
or
not
knowledge
meant,
and
then
another
thing
that
I
want
to
point
out
here
is
how
s
is
going
to
request
from
both
our
listing
it
from
both
destinations.
These
are
this
cell
there's
another
thing
that
we
need
to
be
discussed
if
we
want,
if
this
packet
application
elimination
procedure
is
interesting.
Q
C
C
Thank
you,
so
thank
you,
I'm
afraid
we're
out
of
time.
Thank
you,
I'm
afraid
that
we're
also
out
of
time
for
the
last
presentations
or
usual
I
understand
it
I
mean
you
already
presented
this.
The
interim
meeting
I'm
happy
to
continue
discussing
the
draft
that
a
follow
up,
injury,
meaning
or
on
the
main
in
list,
but
we're
running
out
of
time.
C
E
Just
want
to
let
you
know
that
at
6
p.m.
I
will
be
talking
about
an
ongoing
project
founded
by
Issa,
where
we
are
looking
out
to
integrate
IOT
network
with
satellite
connection.
If
you
are
interested
be
app,
if
you
join,
it
is
in
Paris
room
and
I
will
show
to
you
what
we
have
done
so
far,
and
the
defending
many
results
that
we
have
I
think
the
event
would
announce
it
on
some
many
English
already,
so
I
will
be
a
peaceful
joy.
Thank
you.
Thank.