►
From YouTube: IETF105-6TISCH-20190725-1740
Description
6TISCH meeting session at IETF105
2019/07/25 1740
https://datatracker.ietf.org/meeting/105/proceedings/
A
B
B
C
B
I
supposed
to
just
say
like
we
usually
do
here,
is
what
changes:
everybody's
okay,
let's
move
on
right,
so
I
I
really
wanted
to
have
a
technical
discussion
on
security
and
close
up
the
security
section
of
many
more
and
we
have
Tejo
and
may
sign
the
room.
So
I
am
quite
confident
that
we
will
achieve
that.
So
this
is
my
HF
meeting.
The
usual
notwell
applies.
B
If
you
are
aware
of
any
IPR
which
applies
something
which
is
discussed
today,
either
you
refrain
from
participating
or
you
announce
the
APR
from
you
for
your
sponsor
form
or
ever
you're,
aware
of,
if
you
don't
want
to
do
that,
are
in
the
meeting.
Please
come
to
the
chairs
at
the
end
of
the
meeting.
It's
not
the
only
best
practice
that
applies
to
this
meeting.
Now
a
number
of
best
practices
there
are
listed,
serious
of
apply
to
harassment
and
the
way
the
working
groups
operate.
All
of
those
best
practices
apply
to
this
meeting.
B
We
are
being
recorded
by
Michiko.
Actually,
they
will
publish
the
recording
on
YouTube
and,
what's
even
better,
you
can
click
on
some
sessions
on
some
part
of
the
blue
YouTube
thing
and
you
can
get
the
transcript.
It's
not
perfect,
but
it's
kind
of
impressive
that
we
are
there
there.
So
it's
being
recorded,
it's
being
transcripted
you'll,
be
on
video.
B
If
you
want,
if
you
come
to
talk,
you
better
sit
on
on
this
red
cross,
that's
behind
my
feet,
so
so
you'll
be
so
nice
on
video,
we'll
be
passing
the
blue
sheets
in
like
10
minutes,
because
we
are
used
to
the
crowd
which
comes
a
bit
late.
So
we
better
pass
the
blue
sheets
with
a
little
bit
of
delay
as
well
and
captures
better
the
people
in
the
room
usual
links,
so
they
are
mostly
right
and
unless
they
are
wrong,
and
so
the
etherpad
should
not
have
tools,
that's
that's.
B
Okay,
so
remove
tools
and
the
P
should
not
be
down
there.
Oh,
it
should
be
I,
don't
know,
I
mean,
since
it
changed
I'm,
not
sure
so
the
usual
intro
status
really
we'll
go
through
the
agenda
by
seeing,
if
you
hand
up
you
win
anything
in
this
edge
on
that,
please
let
us
know
now
so
we'll
give
you
a
status
of
what
has
been
around
in
the
working
groups
that
six
station
tough
years
with
will
give
you
a
status.
B
The
architecture
that's
going
to
be
me,
then
I'll
pass
the
ball
to
Manisha
and,
as
I
said,
militia
will
go
through
the
operation
of
many
more
and
in
particular,
the
resolution
of
the
the
issues
that
we
found
discussing
the
sixties,
architectural
security
section
and
then
Michael
is
here.
Oh
is
the
new
ok,
so
Michael
will
discuss
two
drafts
and
I
have
slides
only
for
one.
So
I
expect
that,
for
the
other
you
can
live
without
slides.
B
Unless
you
send
them
right
now
and
I
can
push
them,
but
I
don't
have
a
level
and
Micucci
Luka
asked
us
to
be
pushed
at
the
very
end
of
the
meeting.
So
we
did
this
this
swap
in
the
in
the
agenda.
Michaels
draft
and
the
dynamic
scheduling
things
were
swapped.
It
doesn't
mean
that
Michael
can
spend
all
the
time
in
the
world
and
and
consume
MSF
time.
It's
not
a
matter
of
priority.
So
Michael,
please
stay
within
your
schedule
and
we'll
be
all
right.
B
The
fragment
drawers
I've
also
been
fast.
Let's
go
at
six
low.
There
is
a
little
bit
question
about
whether
the
first
of
those
two
should
be
informational,
Astana
truck,
because
the
second
one
uses
it
and
it's
down
the
track.
So
there
is
no
match
if
done
right.
If
the
first
one
is
is
just
informational.
So
it's
surest
to
give
us
guidance.
B
There
reported
away
leave,
which
is
also
a
nice
thing
for
us
is
that
up
to
the
troll,
a
score
on
which
we
depend
from
Minnie,
more
is
RFC,
and
so
the
last
thing
that
was
reacting
us,
which
is
the
the
dislike
discussion,
is
now
happening,
and
there
was
above
at
this
HF
and
that's
where
haddock
could
see
it.
If
it's
adapted,
we
have
seen
a
number
of
things
which
happening
that
relates
to
us
in
different
working
groups.
B
At
this
point
of
time,
the
most
interesting
is
probably
the
animal
constraint
vulture.
So
I
expect
that
when
you
talk
about
euro
touch,
Michael
will
give
us
the
status
on
the
constraint
vulture
as
well,
and
here
is
the
kind
of
a
snapshot
of
the
result
activity
on
the
milestone.
So
that's
where
Suresh
you've
heard
about
us
recently.
So
another
good
number
of
our
milestones
on
our
green
and
we
are
added
to
my
stands
for
the
MSF
and
for
the
announce
beacon,
because
we
adopted
a
non-speaking
and
Michael
will
talk
to
us
about
it.
B
It's
a
very
simple
draft,
very
useful,
so
we
expect
to
progresses
rapidly.
Now
and
then
will
be
well
we'll
discuss
what
happens
to
zero
touch,
but
pretty
much
everything
will
be
complete
so
between
now
and
Singapore,
we'll
be
in
a
situation
where
the
working
group
has
completed
its
job
so
we'll
be
facing
two
possibilities.
We
close
the
working
group,
which
is
perfectly
fine.
B
We
are
done,
we
happy,
we
call
it
victory
or
we
look
at
the
one
thing
that
we
started
with
and
we
never
completed,
which
is
scheduling,
time
slots
for
the
gmpls
operation
that
we've
been
discussing
at
the
very
beginning.
Initially,
I
thought
this
could
go
to
the
row,
a
fault,
but
the
raw
effort
will
end
up
being
a
rotting
area
effort
dealing
with
layered
what
happens
above
TCH,
but
not
within
TCH,
meaning
that
it's
a
really
orthogonal
work.
B
So
row
will
not
be
place
for
scheduling
time
slots,
40s,
eh,
meaning
that
we
need
to
find
a
home
for
it,
and
that
could
be
a
reason
to
reach
out
or
six
that's
where
I
am
so
sorry
that
that's
the
best
I
have
at
this
moment.
So
here
is
what
I
challenge.
If
you
have
any
question
on
this,
we've
deleted
terminology
because
it
was
merged
into
architecture
and
they're.
Both
in
is
your
review.
We
passed
most
of
the
review.
I
guess
we
are
all
green.
B
B
Okay,
so
the
architecture
I
said
first
most
of
the
reviews
for
the
ASG,
so
I
expect
that
will
be
in
earth
48
very
soon
now.
Well,
in
the
we
are
in
RFC
edit
of
queue
already
so
rush,
maybe
not
no,
no
we're
not,
but
I
expect
to
be
there
soon.
So
the
queue
there
are
people
in
the
queue,
so
it
will
take
the
time
it
takes.
But
as
far
as
I
know,
the
current
version
of
the
architecture
is
very,
very
close
in
terms
of
content
to
the
RFC.
B
B
We
had
like
three
revs
for
and
addressed.
You
know
for
cease
curry.
So
thank
you
all.
Another
big
change
was
the
security
section.
We've
seen
all
the
discussions
and
many
lists
about
the
security
section
and
under
review
there.
So
the
we
hope
you
in
the
text
right
now
the
gory
details
go
actually
in
the
minimal
draft,
and
so
that's
what
Manish
on
until
we'll
be
discussing
today
and
then
there
are
many
small.
It's
also
I
won't
list
them
just
look
at
the
D
5000
interested
and
that's
pretty
much
it
and
Manisha.
E
F
E
Mailing
list,
so
the
update
in
10
that
was
discussed
in
Prague.
We
found
the
resolution
that
includes
the
expansion
of
section
8,
3
3,
which
concerns
essentially
the
scenario
of
the
drawing
registrar
and
coordinator,
failing
partially
failing
essentially
failing
to
preserve
mutable
context
parameters
in
or
score,
but
able
to
preserve
the
pre-shared
keys
from
some
database,
for
instance,
and
the
the
pledged
identifiers.
So
we
had
to
recover
from
that,
and
the
recovery
process
works
as
follows.
First
in
case
the
failure
occurs.
E
Then,
once
the
network
is
reinitialized,
this
will
essentially
force
all
the
old,
no
it's
in
the
network
to
rejoin
to
send
the
drawing
requests
as
per
draft
minimum
as
per
a
minimal
security
and
a
normal
join
process.
But
on
the
jrc
side
we
are.
We
know
that
the
failure
happens,
so
we
can
reach
that
we
can
reject
the.
E
We
can
reject
the
first
drawing
request
in
order
to
in
order
to
prevent
the
nonce
reuse
on
the
or
score
level
in
the
in
the
corresponding
draw
in
response,
and
instead
we
trigger
the
we
trigger
the
this
procedure
that
is
defined
in
oscar
in
RFC
86
13.
That
involves
the
derivation
of
a
new
context.
I,
then
a
context
identifier,
so
the
procedure
works
as
follows,
essentially
for
in
the
first
drawing
request,
this
is
Azrael
normal.
E
This
message
here
is
protected
with
the
concatenation
of
eui-64
and
the
random
nonce
and
which
triggers
essentially
the
pledge
to
derive
a
new
to
select
an
around
a
new
random
number
f3
to
concatenate
it
with
our
2
and
to
protect
the
following
join,
join
request
that
it
makes
to
the
network
and
finally,
as
the
output
of
this
procedure
from
kit
context,
that
is
eui-64
of
the
pledge.
We
end
up
with
a
new
kit
context
and
fresh
parameters
that
Allah
that
pray
allow
us
to
prevent
the
nonce
reuse
in
the
drawing
response.
E
So
the
second
update
in
10
relates
to
the
error
handling
on
the
level
of
constraint,
Rhoyne
protocol.
So
in
the
previous
version,
what
we
this
update,
came
after
Christians
mail
or
discussion
on
the
mailing
list,
and
the
issue
in
the
previous
version
was
that
the
error
signaling
was
not
done
in
a
restful
way
such
that
the
request
was
referencing,
a
previous
request,
and
it
would,
it
would
prevent
essentially
the
server
from
handling
the
drawing
requests
in
a
stateful
manner.
E
So
essentially,
what
we
did
is
that
we
redefined
the
cyber
structures
that
carry
the
parameters
and
they
now
contain
all
the
information
for
the
error
handling
to
occur
in
a
stateless
manner.
So
essentially,
we
it
works
as
before
if
the
join
join,
a
request
fail.
If
the
processing
of
the
join
response
fails,
the
the
the
further
the
next
subsequent
the
subsequent
to
join
requests
will
contain
the
Seaboard
unsupported
configuration
parameter.
That
will
include
essentially
what
went
wrong
in
a
stateless
manner,
and
now
we
have
this
structure.
G
E
G
E
It's
the
driver
to
start
coordinator
that
fails
completely,
so
it's
the
failure
event,
it's
the
failure
event.
So
it's
here,
it's
the
failure
event
of
the
drawing
register
coordinator,
where
the
Droid
register
coordinator
loses
all
their
arm,
State,
for
example,
but
it
is
able
to
retrieve
pre-shared
keys
of
all
the
pledges
that
it
manages
in
different
networks
out
there.
E
E
Implication
of
the
attack
is
that
the
join
response
here
would
would
have
to
time
pad
on
the
nonce
essentially
used
for
the
drawing
response
would
be
used.
Weiss
such
that
there
would
be
a
two
time
pad
on
the
join
response.
So
cryptographically.
The
security
properties
of
the
drawn
response
was
no
longer
failure
and
will
no
longer
hold
I.
G
H
So
this
is
yes,
Michael
Richardson
here
so
so
this
is
a
catastrophic
kind
of
event.
Where
you
know
the
the
plant
is
still
alive,
but
the
fire
just
the
fire
in
the
office
destroyed
all
the
equipment.
Okay,
that
operates
the
plant
and
afterwards
you
buy
new
new
servers
and
you
restore
from
backup,
but
the
backups
don't
have
all
the
aren't.
You
know
we're
a
month
old,
and
so
you
don't
have
all
the
latest
information
at
this
point.
H
You
know
you
have
no
front
office
back
office
you're,
going
around
you're
going
to
be
pushing
buttons
on
every
single
thing
is,
as
we
initialize
the
whole
Factory.
Okay.
So
yes
you're
absolutely
right,
but
the
point
is
that
that
it
would
be
terrible
if
you
also
has
to
have
to
unmount
all
the
vice
and
put
new
PS
kaizen,
though
okay,
so
that's
even
worse.
At
least
all
you
have
to
do
is
somehow
power
cycle
them
and
get
them
to
keep
going.
You.
B
H
Said
that
that
they
get
out
of
sync
anyway,
because
while
while
the
firefighters
were
there,
they
got
it
they
sorry,
while
the
firefighters
were
there
putting
out
the
fire,
the
the
network
fell
apart
because
they
got
out
of
sync,
and
so
they
all
go
back
in
to
join
state
again.
So
you
don't
even
have
to
do
what
I
just
said,
I
think
it's
a
pair
of
saying
you
don't
have
to
push
a
button
everywhere.
They
all
just
go.
H
B
H
H
H
E
A
ssin
you
are
not
entirely
correct
on
that,
because
the
problem
is
the
the
parameter,
update
response
that
comes
from
the
GRC,
so
if
jrc
loses
the
state,
yes,
if
the
jrc
loses
the
state,
it
needs
to
right
now
the
text
in
the
draft
says:
okay,
you
have
to
force
out
of
and
all
the
6lb
ours
to
reboot.
Why?
Because
you
know
on
the
parameter,
update
request,
for
example,
for
eking
or
something
of.
H
Course,
as
we
get
to
a
Ricky
we're
screwed
and
we
have
to
rebuild
the
network,
because
we've
lost
all
that
state.
I
totally
agree
with
you,
yes,
you're,
absolutely
right.
What
I'm
trying
to
say
is
that,
during
the
period
that
the
jrc
was
being
rebuilt,
being
reloaded
okay,
unless
the
network
had
to
do
something
involving
the
jrc,
it
could
keep
operating.
H
F
They're,
giving
it,
of
course,
one
of
the
things
we'll
probably
also
happens
that,
because
the
time
book,
the
master
time
coordinating
neighbor
goes
away
because
it
usually
probably
the
first,
you
know
connected
to
very
title
the
coordinator
and
when
that
time
connection
connection
goes
away.
At
some
point,
you
probably
you
see,
but
you'll
have.
F
E
E
Okay,
yes,
so
then
we
also
had
the
update
11
after
Pascal's
review,
so
it
was
mainly
an
editorial
update
without
update
of
references
and
a
couple
of
roots
in
Section
six
one
two
regarding
the
setting
of
the
diffserv
code
points
in
couple
of
needs
fixed
here
and
there
so
now,
I
would
like
to
skip
to
the
latest
discussion
on
the
mailing
list.
So
we
have
essentially
two
issues
open,
but
before
I
go
into
those
details,
I
will
give
some
background
on
I
Triple
E
15
for
that
influences
these
discussions.
E
Essentially,
the
first
three
octaves
are
the
company
identifier,
a
reserved,
zero,
zero,
pan
ID
and
the
short
address,
which
is
of
interest
for
us
to
bite
locally,
unique
to
byte
address
that
is
assigned
by
the
jrc
to
the
to
the
to
the
to
the
pledged
at
the
drawing
time.
So
now
this
this
is
decide
is
important
in
order
to
understand.
When
can
the
nonce
three
use
occur?
Usually
the
DUI
64
is
a
globally
unique.
E
So
when,
when
the
node
is
using
that
in
the
layer,
two
security
procedures,
there
is
no
risk
of
this
part
of
the
nonce
being
reused,
but
then,
as
we
can
see,
lighted
and
in
the
attack
that
Tara
brought
up
the
a
the
notion
of
a
SN
is
critical,
as
if
the
ASN
is
the
old
erase
and
is
used
essentially,
the
the
nonce
reuse
can
occur
so
to
understand
the
better
how
how
the
ASN
is
distributed.
I
gave
here
this
slide
with
essentially
the
pledge
in
scanning
mode.
E
Try
listening
for
beacons
from
from
the
network
and
to
advertising
nodes
that
send
beacons
at
different
instants,
so
the
pledge,
according
to
RFC,
eighty
minimal
8480,
no
eighty
one,
eighty
eighty
one,
eighty
pledge
waits
for
maxi
be
delay
time
to
find
num
numbers
to
wait.
Neighbours
essentially,
two
parameters
that
specify
how
long
should
be
the
maximal
scanning
time
be
if
and
how
many
neighbors
should
at
least
the
pledge
listen
to
for
and
upon
the
expiration
of
this
time.
E
Essentially,
the
pledge
the
pledge
selects
one
of
these
advertising
nodes
as
its
joint
proxy,
with
the
beacon
containing
the
ASN
number,
with
with
beacon
containing
the
the
information
element
carrying
the
current
ASN
in
the
network
and
now
beacon
is
protected.
Cryptographically,
it's
authenticated,
but
at
this
time
here
the
pledge
does
not
have
a
means
of
verifying
it
and
it
does
a
leap
of
faith
and
sense.
E
So
now
to
explain
how
the
attack
works.
We
assume
the
attacker
is
in
bit
place
a
man
in
the
middle
between
the
draw
and
proxy
and
end
the
pledge
and
it
replays
the
replay
it
stores
it
listens
for
some
time
and
stores
the
enhanced
beacons
that
the
legitimate
enhance
beacons
that
were
exchanged
in
the
network
and
replace
them
at
a
later
time
to
the
pledge
now
pledged
without
being
able
to
validate
here
the
content
in
the
enhanced
beacon.
E
It
takes
this
notion
of
a
SN
and
since
the
drawing
requested,
the
corresponding
slots
and
the
attacker
forwards.
This
onwards,
through
the
legitimate
network
to
the
Dre
RC
and
the
Dre
RC
sees.
This
is
a
fresh
drawing
request.
So
it's
a
legitimate
pledge
that
has
the
keys
involve.
So
everything
is
fine.
It
will
respond
with
the
drawing
response,
with
the
join
response,
currently
not
having
any
secure
any
layer
to
protection
on
on
this
lag
between
the
join
proxy
and
the
pledge.
E
E
So
this
attack
still
works
as
if,
because
the
the
enhance
beacon
will
validate
as
it's
a
regular
enhanced
beacon
from
the
network,
but
it
will
be
a
replay,
so
the
ASN
will
essentially
be
lower
than
the
one
that
is
currently
going
on
in
the
legitimate
part
of
the
network
over
on
this
side.
So
the
problem
arises
now,
after
the
drawing
after
the
drawing
process,
once
the
drawing
response
is
received
in
the
replayed,
enhanced
beacon,
validates.
E
The
pledges
notion
of
time
will
be
this
number
here,
23
23,
plus
the
time
it
that
elapsed,
to
get
to
this
point
here.
So
let's
say
that's
55.
So
when
the
pledge
now
uses
the
keys
that
it
obtained
from
the
drawing
response,
it
will
use
the
ASN
55
in
this
derivation
here
that
might
have
been
replayed
before
from
a
previous
attempt
of
joining
the
network
from
a
previous
network
exchange,
essentially
and
which
might
lead
to
the
dis
frame.
This
to
the
nonce
reuse,
at
layer,
2
of
this
particular
frame
with
a
s
and
55.
E
E
So
so
here
the
proposed
resolution
is
quite
simple,
but
it
involves
some
cross
layer
signaling,
some
some
cross
layer
operations,
essentially
what
we
proposed.
We
had
a
side
meeting
involving
meet
arrow
and
we
also
discussed
with
Pascal
on
on
the
proposed
resolution,
how
we
should
approach
this
problem,
and
essentially
it
involves
authenticating
at
layer
to
the
join
response
on
the
lag
between
join
proxy
and
the
pledge
where
previously
there
was
no
security
involved.
E
So
this
is
a
bit
tricky
because
join
response
join
proxy,
contain
or
already
is
part
of
the
network,
and
it
has
the
keys
that
are
used
in
the
network,
so
it
can
send
this
frame
authenticated,
but
not
encrypted.
For
for
the
pledge.
So
essentially
when
pledge
does
in
the
process
is
as
follows:
the
pledge
sensor
drawing
requests
as
before
without
any
layer
to
security.
The
join
request
goes
to
the
jrc
jrc
responds
with
the
drawing
response.
So
all
of
this
logical
connection
here
goes
over
the
network
encrypted
at
layer
two.
E
So
this
is
all
fine,
but
the
trend
comes
here
so
essentially
when
the
join
response
comes
at
to
the
to
the
join
proxy
instead
of
sending
it
without
any
layer.
Two
security
join
proxy
sends
this
with
one
of
the
first
three
security
levels
in
802
dot
15.4,
which
involve
only
authentication
of
the
frame.
So
now
this
means
that
the
pledge
receives
a
frame
that
has
the
security
level
bit
set,
but
that
it
doesn't
have
the
key
to
verify
the
validity
of
this
frame
because
the
key
is
inside
this
frame.
E
So
the
procedure
is
as
follows:
we
are
essentially
the
pledge
discards,
the
layer,
two
Olympic
verification
it
decrypts
and
verifies
the
end-to-end,
a
security
session
of
Oscar.
From
this
co-op
message
and
the
coding
of
the
Seaboard
object.
It
gets
the
layer
two
keys
and
now
passes
back
the
whole
thing
into
the
layer,
two
machinery
in
order
to
validate
the
frame
to
validate
specifically
D
authenticity
at
layer,
two
of
the
frame
that
that
that
it
that
it
used
to
obtain
the
drawing
response.
E
So
there
are
a
couple
of
carrots
with
with
this
approach.
First
off,
it
is
a
bit
tricky
in
terms
of
the
implement
implemented
wise
to
process
this
to
process
the
frame
that
has
the
security
level
bit
set.
But
the
security
processing
fails,
especially
because
this
frame
is
unicast,
so
there
is
a
special
mode
in
a
to
2015
for
called
promiscuous
mode
that
allows
the
when
set
when
this
attribute
is
set
in
802
15
for
all
the
frames,
regardless
of
the
of
the
state
of
the
security
processing,
are
passed
to
the
upper
layer
there.
Oh
so.
F
You
call
us
what
about
is
you
know
if
everything
has
persisted
Mac
OS
or
the
FCS
is
passed
up,
so
it's
a
checksum
that
matches
it's
passed
that
we
can
as
if
it's
testing
it
to
you
or
not.
There
is
an
other
way
which
is
currently
not
possible,
but
it
would
build
the
in
future.
Hopefully
next
year,
is
that
when
we
receive
a
frame
that
is,
we
don't
have
a
keys.
We
I
Triple
E
in
54,
says
a
big
all
this
MLM
me.
F
Does
it
say
what
Emily
call
that
this
comma
indicates,
and
that
tells
you
that
there
was
something
happening
from
the
network
and
if
we
have
received
the
frame
that
we
don't
have
a
key
currently
that
MLM
recall
doesn't
receive
the
frame
there
is
no
frame.
There
is
no
it
nothing
there.
It's
just
a
stat.
Oh,
we
appreciate
the
frame
from
this
person
and
we
don't
have
a
key
for
it.
It's
not
the
keys,
please
what
we
are
planning
to
do
in
the
renovation.
We
are
doing
15
for
revision.
F
E
B
B
E
B
B
E
E
F
Also,
a
problem
that
we
want
to.
We
want
to
tie
this
out
of
the
case
and
the
layer
to
ASM,
and
the
problem
with
ASM
is
that
when
the
you're
turning
proxy
or
Tony
knows
it's
ended
the
frame,
they
have
no
idea
what
the
SNS
called
to
be
because
I
mean
what
they
put.
The
things
create,
the
frame
they
send
it
out.
The
back
layer
takes
the
frame
waits
for
the
minute
suitable
channel.
You
know
the
time
spot
to
be
and
okay
see.
Oh,
it's
busy
I
go
to
the
next
one.
Oh
that's
empty.
E
So,
even
though
it's
a
single
hop
away,
it's
the
upper
layer,
so
I
concur
with
Tara
yeah
yeah
yeah,
so
this
is
one
so
the
use
of
promiscuous
mode
and
the
and
the
update
for
plan
for
2020
revision
of
15.4
will
make
this
essentially
possible
in
terms
of
security
processing.
Another
carry
it
is,
is
related
to
the
layer
to
AK
of
the
join
response
and
the
join
response.
E
Since
the
drawing
response
frame,
as
you
can
see
here,
it
is
now
sent
with
the
security
level
one
two
three
so
authentication,
but
no
encryption
from
at
the
join
proxy.
So
at
the
pledge
side
it
does
the
special
processing
that
we
do
I
just
presented
here.
But
the
problem
is
that
when
now,
since
this
is
a
unicast
frame,
the
pledge
needs
to
send
an
acknowledgment,
a
layer,
two
acknowledgment
that
normally
in
TS
CH
is
detected
as
well.
E
Using
the
same
level
of
security
that
was
used
for
the
incoming
frame,
and
the
problem
here
is
that
there
is
no
enough
time
to
do
the
whole
to
do
the
whole
upper
layer
processing
and
before
sending
that
layer
to
an
acknowledgment.
So
the
the
enhanced
acknowledgment
that
is
sent
back
here
is
own
can
only
be
sent
in
without
any
layer
to
security,
and
this
still
works
with
the
current
configuration
in
the
draft,
because
the
drawing
proxy
has
this
SEC
exempt
attribute
set
for
this
pledge.
E
B
B
F
Because,
even
if
you
know,
even
if
you
don't
be
able
to
do
this
kind
of
crypto
methods,
we
can
be
retained
or
we
can
actually
sustain
that.
You
know
the
responder
type
bridge.
The
state
spraying
ignores
that
that
it
failed
the
risk
security
takes
the
OS
core
there
and
put
the
keys
in
and
waits
for,
the
trust
and
many
hints
of
retransmission.
Then
he
knows
that
okay,
he
can
verify
that
the
retransmission
conduct
is
saying
that
previous
that
he
is
check
out
that
it
same
because
otherwise
he
doesn't
know
its
a
hardcore.
You
know.
F
E
So
what
I'm
discussing
here
is
a
generic
802
15
4
situation,
not
specific
to
6
dish,
but
since
we
are
using
the
TS
chmod,
obviously
it
relates
to
us-
or
we
can
put
some
text
in
the
minimum
security
draft
so
to
come
back
so
essentially
CCM
star
security
proofs
do
not
apply
to
the
TSE
h
because
they
assumed
that
they
required
that
the
norms
contains
the
security
level,
which
is
the
case
with
regular
15
4,
but
not
1540
SCH.
So
with
this
we
revert
back
essentially
to
the
you.
I
mean
we.
E
We
can
still
rely
on
the
proofs
of
CCM
CCM,
but
the
limitation
there.
It
is
that
a
single
key
can
only
be
used
with
fixed
line
length,
authentication
tags,
meaning
that
we
need
to
prohibit
the
mixing
of
different
security
levels
in
the
network
where
different
pairs
of
nodes
use
different
length
of
the
authentication
tag
or
message
integrity
chromic,
as
in
in
with
the
same
key,
obviously
with
the
same
key.
So
the
I
mean
this.
E
So
that's
pretty
much
it
I,
don't
know
if
I
think
I'm
good
on
time.
So
the
conclusion
is
so.
These
were
the
latest
discussion
that
were
triggered
essentially
after
the
review
of
the
architecture
draft
that
that
that
inspired
this
discussion.
So
it's
good.
The
document
is
shipped
already
to
the
ad,
so
I
think
I'm,
not
sure.
Now
whether
we
should
wait
for
the
reviews
or
do
these
changes
immediately.
F
F
E
B
B
H
Don't
think
I
sent
slides
because
I
didn't
think
there
was
news,
so
0
touch.
Yes,
I
have
posted
a
new
revision
of
the
document,
it
is
significantly
shorter,
it
had
expired,
I
believe
sometime
last
summer
and
it
is
significantly
shorter
because
the
goal
is
always
to
have
the
work
done
elsewhere
and
that
work
is
getting
done.
So
specifically,
you
may
be
aware
that
the
base
zero
touch
brewski
document
is
at
the
is
G
and
a
lot
of
discusses,
hopefully
they'll
get
cleared
in
the
next
week
and
that
will
go
forward.
H
The
0dt
security
zero
touch
join
document
depends
upon
Animas,
constrained,
voucher
and
constrained
brueski
document,
which
is
a
remodeling
of
brewski
to
ed,
Hawk,
plus
Cosi
signed
Seaboard
vouchers
and,
what's
the
other
part,
let's
go
up.
Yes,
OS
core
and
co-op
okay,
and
so
that's
what
this
document
says
and
so
I
essentially
reduced
it.
There
I've
actually
removed
DTLS
from
this
document.
If
you
don't
like
that,
we
could
discuss
that
point
and
it
I've
removed
many
pieces
that
have
since
moved
into
minimal
security.
H
So
that's
why
it
got
much
smaller
and
that
was
always
the
goal.
The
document
that
was
always
supposed
to
say
do
ABCD
they're
over
here
and
not
document
themselves.
So
that's
what
that!
Well,
that's
where
that
went,
but
there's
no
substantive
dot
changes
to
the
protocol
or
the
concepts.
Just
the
text
is
no
longer
redundant.
That
makes
sense
to
you.
Pascal
does.
H
We
had
AB
off
on
Monday
called
lake,
and
if
that
goes
forward,
which
I
believe
it
will,
then
we
will
have
IDI
hawk,
which
allows
us
to
do
zero
touch
without
DTLS,
otherwise
we're.
How
do
we
fit
monster
DTLS
packets
over
and
it
doesn't
involve?
It
doesn't
interoperate
very
nicely
with
minimal
security
or
OS
core,
and
if
that's
the
only
option,
then
the
answer
is,
we
probably
don't
have
anything
to
advanced,
but
if
we
have
IDI
hawk
or
something
equivalent
in
the
lake
working
group
or
Boff,
then
that's
good.
H
B
As
we
said
only
in
the
in
the
group
right,
the
group
has
two
options
is
to
which
other
just
for
the
scheduling
or
to
just
disband
and
call
victory.
If
we
do
the
latter,
then
this
will
probably
be
the
only
outstanding
draft
that
the
would
well
group
would
have
when
we
quote-unquote
shut
down.
So
we
would
have
to
see
how
we
we
keep
some
lights
on,
so
you
can
finish
the
work.
That's.
H
A
good
point,
thank
you
actually,
and
some
of
us
I
think
you
included,
are
going
to
have
a
chat
with
Ignace,
who
is
an
84
anima
as
to
how
do
we
do?
How
do
we
regroup
some
of
these
documents
together
into
the
right
place
and-
and
maybe
Suresh
you'd
like
to
join
us?
It's
where
we're
gonna
meet
right
here
right
after
this
meeting?
If
you
have
ten
minutes
afterwards
that
point
so.
H
That
would
that
would
please
me
just
fine.
The
contents
of
this
document
are
not
particularly
six
tissue-specific
anymore,
and
this
essentially
all
of
the
things
that
are,
except
for
the
constraints
of
the
solution.
All
of
the
things
that
worse
but
six
dishes
specific
in
this
document
sometime
ago
have
moved
to
the
minimal
security
document
and
that's
exactly
what
the
goal
was
was
to
get
get
it
out
of
something
and
do
that
so
any.
H
H
They
would
like
to
know
if
they
have
found
the
correct
network,
or
rather
if
they
have
find
the
diversity
of
networks
that
they
wish
to
find
and
they
get
beacons
from
different
places,
and
they
don't
really
know
where
they
belong
in
in
the
the
co
chip
joint
process,
that
six
dish,
minimal,
describes
and
DT
security
depends
upon
as
well.
You
are
supposed
to
go
through
all
the
joint
proxies
for
given
network
and
try
them
all
out.
H
So
it's
important
to
know
which
one
proxies
belong
to
the
green
networks
and
which
runs
brawlin,
so
the
blue
network,
so
that
you
can
either
try.
Also,
you
can
try
the
green
network
and
a
blue
network
and
then
a
green
network,
and
then
a
blue
network,
and
one
of
them
might
like
you
in
one
of
them:
I.
Not
if
you
waste
all
your
time
on
the
network
that
doesn't
that
isn't
going
to
accept
you,
then
that's
a
bad
thing.
So
that's
those
that
there's
a
couple
of
different
things.
H
The
second
thing
that
people
asked
is
well
I
would
like
to
have
some
information
about
the
networks
that
are
there
so
that
I
can
join
the
network,
which
either
is
less
loaded
or
the
parts
of
the
network
which
are
less
loaded.
So
you
may
have
several
choices
in
the
blue
network.
As
you
see
as
two
different
places
you
could
join
and
that
would
determine
what's
going
on.
H
So
the
question
is:
can
you
get
this
information
before
you
have
done
all
the
stuff
in
particularly
this
not
just
applies
to
the
join
step,
the
enrollment
step,
but
it
also
applies
to
Sleepy
nodes.
That
may
have
slept
too
long
too
and
gotten
out
of
sync
on
their
ASNs
and
therefore
need
to
join,
have
the
keys,
but
they
need
to
then
figure
out
which
part
of
the
network
they
need
to
join
and
do
the
right
thing.
So,
six
dish
part
we've
created
an
information
element
which
goes
in
the
enhanced
beacon
that
uses
the
IETF
allocation.
H
A
multicast
Neera
discovery
to
find
it,
assuming
that
the
lower
bits
are
not
the
same
as
the
in
the
l2
address
of
the
beacon
that
you
heard,
and
then
you
get
some
other
numbers
ranked
priority
priority
priority,
and
these
are
containers
for
things.
So
the
rank
priority
I
remember
how
right
comes
from
ripple
and
there's
a
document,
a
ripple
that
tells
you
how
you
might
like
to
fill
that
in
not
this
working
groups
bother
just
it's
a
container
to
put
something
in
the
our
flag.
H
Tells
you
whether
or
not
the
router
would
accept,
is
a
factory
router
and
would
accept
unicast,
router
solicitations,
which
you
might
like
to
do,
because
that
again
saves
another
broadcast.
So
that's
really
it.
There
hasn't
been
a
lot
of
movement.
We've
changed
the
format
of
this
added
and
removed
some
things
over
the
last
year
and
a
half
or
two
years
since
it's
been
alive,
and
that's
really
yet.
We
think
it's
ready
for
working
group
last
call
there's
a
lot
else
to
do.
B
H
B
H
H
B
B
Message
so
thank
you.
Thank
you
very
much
so
so,
basically
we
had
the
choice
of
taking
of
inviting
one
of
the
main
authors,
but
from
Terry.
So
well
you
know
music,
oh,
but
Josh
was
with
us.
He
knows
the
spec
very,
very,
very
well,
and
so
we
asked
him
to
do
the
presentation
of
behalf
of
the
remotes
editors,
so
thank
you.
Yeah
Chuck
actually
accepting
to
give
us
the
status
the
slides
were
done
by
thankfully,
yes,
yeah
hi.
K
So
the
in
the
previous
that
bested,
the
Virgin
tree
we
had
old
up
cells
and
down
cells
and
upsell
is
basically
used
for
the
upward
traffic
to
chose
to
do
the
parent
and
all
down
cells
is
most
of
used
for
downward
traffic
to
your
children,
so
that
specs
has
been
changed,
change
that
actually
the
removed,
and
we
have
now
the
two
new
Ottomans
tire.
The
cell
types
one
is
auto:
TX
cells
and
the
other
is
auto
or
Excel.
K
So,
basically
them
node
the
schedule
old
or
Excel,
the
its
slot
offset
in
the
China
offset
or
calculate
based
on
its.
You
are
the
MAC
address
and
when
you
have
the
frame
to
your
neighbor
and
you
don't
have
any
delicate
cell
to
the
neighbor,
you
schedule
the
auto
txl
on
the
map
to
send
out
the
frame.
K
So
the
sort
offset
an
eternal
offset
for
the
auto
t
XL
is
again
the
calculate
based
on
the
debt
neighbors
the
MAC
address
once
you,
for
example,
you
don't
have
any
frame
to
transmit
to
the
the
neighbor
in
your
text.
Hxq
you
can.
The
remove
autotext
cell,
so
autotext
cell
is
expected
expected
to
be
removed
automatically
so
that
that's
why
we
on
demand
on
the
schedule.
So
these
figures
show
the
difference
between
auto
down
cell
upsell
and
auto
texas
and
auto
or
sell
their
ideas.
K
Another
major
changes
change
is
dedicate
ourselves
to
parent.
So
before
the
MSF
defines
the
note,
just
after
joining
the
network,
the
node
schedule,
the
data
cos
T
Excel
to
the
parent
that
in
the
end
of
states
of
the
journey
process,
but
now
the
the
node
schedule
dedicate
arcs
as
well
with
a
parent
for
for
incoming
traffic.
K
So
this
is
another
the
major
thing:
it's
it's
it's
a
relatively
new
idea,
so
I
think
the
authors
expect
the
more
feeder
box
on
this
feature
because
it's
new
another
thing
is
it
just
the
sorry
the
the
comments
which
are
already
resolved
so
and
also
you
can
find
the
details
in
marrying
these
archives.
So
there
are
two
things
remain
so
to
be
discussed.
One
is
a
the
cell
this
about
service.
This
is
passed
all
the
suggestions
to
how
to
how
to
select
the
cell
to
propose
in
a
other
requests.
K
B
B
What
we
have
today
is
a
reactive
validation
of
the
cell,
meaning
that
if
we
observe
that
one
cell
is
not
as
efficient
as
the
others,
meaning
there
is
more
collision
that
cell
than
others,
then
it's
probably
getting
some
collisions
now
it
really
depends
on
the
kind
of
traffic
you
may
you
may
get
away
cell.
You
use
it
a
little
bit
more
a
little
bit
less
so
I
argue
that
the
efficiency
of
the
reactive
discovery
is
very
difficult
to
chew
properly.
B
It
could
take
time
before
you
actually
discover
it's,
not
okay
and
so
I
think
that
listening
little
bit
to
the
cells
before
you
actually
allocate
them.
Maybe
it's
enough.
You
turn
it
on
or
off,
but
argue
that
it's
unusual,
your
useful
thing
to
do
it's
being
polite
right,
like
it's
a
little
talk,
but
it's
the
TCH
way
of
doing
listen
before
talk.
F
Tarek
unit,
it's
just
trying
to
think
about
how
to
actually
do
that.
If
you
think
about
the
15,
for
you
know,
MLM
e
things,
because
I
don't
think,
there's
a
way
to
do
that,
you
cut
the
other
link
yeah,
you
sure
you
can
say
that
up,
I'm
called
risking
this,
but
then
you
only
listen
the
frames
of
our
listeners
to
you
and
I.
Don't
know
if
I
remember
is
there
anything
that
allows
you
to
this
kind
of
promiscuous
listening
on
one
specific
link
on
the
channel?
F
It's
I
don't
know
if
there
other
option,
of
course,
is
to
do
you
know
to
send
one
broadcast
frame
to
that
link,
because
one
of
the
things
is
that
if
there's
anybody
using
that
link,
there's
somebody
listening
that
way
and
he's
listening
it
all
the
time
and
if
you
send
any
kind
of
broadcast
frame
on
that
link,
he
will
act
it.
He
will
send
an
acknowledgement
back
to
you
if
you
receive
an
ACK.
F
F
The
problem
really
is
that
I
don't
know
what
do
you
want?
Th-Th
things
in
MLM
a
are
quite
vague.
They
are,
you
know
very
high
level
where
it
is
kind
of
wavy.
They
don't
really
describe
that
much
of
the
processing.
So
that's
why
it's?
Actually
it's
a
little
bit
hard
to
say
if
you
can
actually
use
this
on
the.
F
F
F
I
F
J
F
It,
which
is
a
little
bit
more
difficult
I,
was
thinking
about
more
or
less.
It
would
be
something
that
you
could
actually
make
a
link
and
say
that
I
want
to
make
this
link
as
a
promise
cross
lake,
meaning
I,
won't
receive
me
whatever
somebody's
sending
on
that
link,
even
if
not
tested
it
to
me.
I
just
want
to
know
that
if
there's
something
on
that,
that
would
probably
want
way
to
doing
that.
So
we've
read
one
more
bit:
they're
saying
that
we
have
an
error
X,
that
we
have
T
X,
that
we
have
shared.
F
B
L
I'll
risk
the
hand,
but
actually
what
I
wanna
say
it's
kind
of
a
water
Tara
wants
a
it's
about.
The
CCA
actually
didn't
detect
the
frames.
If
we
really
want
to
like
listening
to
talk,
we
have
to
make
the
cell
like
your
Excel
first
and
then
on
the
TX
l
like
it's
a
reserve
@
TX
out,
but
we
can
even
listening
at
our
Excel
first
for
a
while
and
we
it
detects
anything
and
then
we
uses
txl.
Is
it
just
sounds
little
I,
don't
know
whether
this
is
stand
compliant
word
or
any?
L
What
I'm
saying
is
if
we
wanted
to
do
something
like
listening
this
poor
talk,
we
have
to
like
treat
this
DXL
as
a
or
excel
first
and
listen
for
a
while
to
check
whether
there's
a
frame
in
the
air
and
then,
if
there's
like
after
a
while,
there's
no
frame
in
the
air
and
then
use
the
cell
as
TX.
Later,
yes,.
L
K
K
That
idea
itself
is
a
new
so
and
the
many
comments
are
expected
it
on
a
merriest,
so
yeah
and
also
there's
no
information,
but
the
MSF
is
implemented
in
an
open
dementia,
of
course,
and
the
six
is
scimitar,
and
now
we
have
the
MSM
implementation
for
the
kinetic
energy,
so
tech,
fair
and
I
are
planning
to
have
the
local
interrupt
the
test
based
on
the
latest
rough,
to
make
sure
the
spec
is
ready.
Thank
you
know.
B
B
Guys?
If
you
could
raise
this
question
to
the
mailing
list
and
explain
exactly
what
they
are
and
make
sure
we
discuss
them
so
because
the
goal
is
to
work
with.
Let's
call
this
this
year
right
so
somewhere
info
to
get
there.
We
need
to
get
those
two
Ayesha
resolved,
and
so
please
bring
them
to
the
mailing
list
and
then
we'll
do
well
go
presscon.
Okay,
thanks!
I
totally
agree!
Nice!
B
B
M
Hi,
this
is
Marco
from
rice
and
an
update,
an
updated.
This
room
is
kind
of
being
deaf
now
version
2
quick
recap:
the
document
describes
both
an
attack
and
the
solution.
The
attack
is
possible
for
an
external
adversaries,
even
in
the
presence
of
security
and
the
league
failure
can
be
performed
very
efficiently
in
a
stealthy
way,
low
power
consumption.
M
So
even
if
you
are
on
battery
it's
about
essentially
inferring
100
percent,
precision,
the
the
full
schedule
or
not
and
then
selecting
which
I'm
with
a
hundred
percent
accuracy
and
effectiveness,
every
single
transmission
and
reception
of
that
note,
the
document
proposal
is
also
a
solution
that
doesn't
require
any
additional
communication
among
the
network
nodes.
It's
all
about
local
efficient
computation
for
shuffling
cells
at
each
slot
frame
in
a
synchronized
way
among
network
nodes.
M
So,
as
a
result,
you
have
schedule
distill,
collision-free,
a
consistent
for
all
network
nodes
plus
is
now
also
unpredictable
for
an
external
adversary.
So
essentially,
the
attack
is
neutralized
what
happened
after
Prague?
We
got
reviews
on
the
list
and
promised
Pascal
thenthe
Michael
thanks
very
much
for
that,
and
then
we
got
digital
comments
on
a
particular
point.
Arrays
later
from
Michael
Emma
on
the
list
again
as
a
summary
of
updates,
we
mostly
as
they
refused
I,
made
a
number
of
improvements
in
the
document.
A
quick
overview
about
this
thing.
M
We
also
fixed
the
presentation
of
the
actual
shuffling
mechanism
used
to
produce
the
robust
scheduling
to
be
fully
aligned
with
the
later
with
latest
fisheries
version.
So
to
shuffle
an
element's
you
don't
really
need
and
swaps,
you
need
n
minus
one,
so
it's
even
a
little
bit
more
efficient
and
it
makes
the
related
counter
growth.
Even
thank
you.
M
Militia
also
recommended
to
relax
a
little
bit
a
recommendation.
We
were
having
in
the
previous
version
when
we
were
saying
well,
it's
just
recommended
to
shuffle
not
only
channel
offsets
but
also
time
offsets
since
in
some
use
case
or
networks
can
make
sense
to
just
not
do
that
for
the
sake
of
a
network
performance,
especially
in
the
multi-hop
context.
We
remove
the
recommendation,
so
we
are
just
now
describing
the
two
possible
shuffling
and
describe
it
the
pros
and
cons
of
not
doing
or
doing
time
shuffling.
M
Then
the
biggest
update
came
actually
out
of
a
comment
from
Michael.
I
wanted
to
have
some
example:
outputs
from
essentially
proof
of
concept
implementation,
showing
the
produced
out
their
schedules
and
schedules
out
of
the
original
one.
So
I
made
a
proof
of
concept,
implementation
using
the
OpenSSL
primitives,
and
you
have
an
appendix
now
in
the
document,
with
a
full
example,
performed
by
a
network.
Node
four
tiers
load
frames
in
the
row
with
full
details
on
the
configuration
that
node
on
the
network,
the
original
schedule,
and
now
it
becomes.
M
M
M
Yes,
so
after
the
reviews,
the
discussion
continued
on
the
list,
especially
between
militia
and
Michael,
to
cover
the
particular
aspect
of
key
renewal
in
the
network.
So
right
now
using
the
the
minimal
security
framework
in
the
joy
response,
we
are
additionally,
additionally
providing
also
the
the
permutation
keys
needed
to
implement
this
solution,
and
in
case
you
have
network
cracking
essentially
well,
you
are
supposed
to
use
the
same
message
again
and
distribute
new
permutation
keys
as
well.
Of
course,
different
nodes
will
get
the
new
key
sets
at
different
time.
F
M
For
while
you
might
have
misalign,
permutation
keys
and
different
schedules
will
be
produced
for
a
while,
and
this
needs
to
be
minimized,
at
least
as
an
impact.
So
the
big
question
is:
when
are
you
supposed
to
switch
to
the
new
permutation
keys
and
the
best
answer
is
apparently
well
not
exactly
when
you
switch
to
the
new
network?
Is
that
was
followed
by
a
pretty
complicated
discussion
on
the
list
that
I'm
trying
to
summarize
here
in
this
slide
that
looked
like
a
promising
way
forward?
M
M
Then
how
the
switch
can
happen
based
on
that.
Well,
the
exact
permutation
key
you
want
to
switch
to
can
be
included
in
indiana's
beacons
as
a
newly
allocated
information
element.
Value
such
and
this
becomes
protected
with
the
new
network
is
key.
Set
will
be
sent
in
the
network
by
the
trusted
border
router,
and
the
network
knows
that
some
point
will
start
eventually
to
be
able
to
process
and
validate
that
an
end.
M
Speaking
will
for
sure
have,
of
course,
network
is
also
the
new
permutation
keys,
and
at
that
moment
they
will
switch
to
them
and
we'll
start
producing
the
the
new
schedule
summary.
We
address
the
reviews
and
we
have
a
big
open
point
with
the
promising
direction,
especially
to
when
to
switch
to
the
new
permutation
keys
in
the
presence
of
network
cracking
so
all
considering
the
status
of
the
document,
the
outcomes
from
the
reviews.
We
believe
this
is
in
shape
for
working
group
adoption
consideration.
B
So
I
guess
this
is
a
question
for
the
chairs.
Okay,
so
the
one
thing
is
that
3
depends
if
the
workgroup
continues
or
does
that
continue
because
end
up
teaching
something
in
a
while
group
which
will
stop.
You
know
it
doesn't
make
much
sense.
Another
question
is
like
for
Tarot,
because
you're,
the
only
802
the
54
hat
in
this
room
is
like
what
I've
seen
about
you
know
we
change
the
keys
like
we
do
for
the
other
layer
to
kiss.
B
F
F
Link
first
used
to
sent
the
frame
out.
It
doesn't
know
what
time
slot
or
ASN
or
her
channel
is
sent
to
frame
out
it.
Just
it's
just
configuration
to
the
Mac
macro,
then
running
stuff
anyway.
Yes,
you
could
do
you
could
actually
change
that.
You
know
the
channel.
You
know
hop
in
sequence
every
now
and
then
would
actually
give
you
certain.
You
know
these
kind
of
features,
but
but
you
know
but
do
that,
but
but
of
course
the
problem
is
that
a
player
can't
coordinate
that
they
can
say.
F
Ok,
I,
do
it
now
and
Mac
will
do
it.
You
know
10
a
sense
later,
because
that's
when
it
everything
that
the
Stanford
epilator
happens
at
some
point
of
time
in
the
future,
in
the
back
and
there's
no
way
of
you
know
to
coordinate
or
force
the
upper
layer
to
do
a
lot
of
math
to
do
something
you
know
so
then
I
don't
think
it
actually
would
be.
F
Very
yes
in
some
implementations,
you,
probably
if
you
do
everything
a
software,
you
can't
do
it,
but
it
wouldn't
be
part
of
the
you
know:
standard
15,
4
or
let's
put
tape.
If
you
do
15
4
thca
its
implementation,
it
might
not
be
able
to
allow
you
to
do
that.
It
might
be
saying:
oh
I
mean
I'm,
just
you
know,
implementing
things
that
are
required
by
I
Triple
E,
and
they
don't
require
you
to
have
these
kind
of
features
that
would
allow
you
to
mess
up
the
scheduling
it
that
way
so.
B
F
F
How
much
it's
going
to
be
really
used,
because
that's
actually
is
one
of
the
things
that,
because
you
need
to
have
a
you
need
to
be
able
to
have
an
know,
I
Triple
E.
We
have
a
power,
you
have
to
have
a
power
which
tells
you
what
projects
are
going
to
be
doing,
and
then
you
need
to
have
this
error
of
great
criteria.
Saying
that
you
need
to
have
a
void
market.
You
know
use
for
it.
F
It
doesn't
enough
to
be
easy
enough
to
be
all
we
in
academia
will
think
of
it
be
nice
to
have
this
kind
of
feature.
That's
not
enough
for
them.
They
don't
want
to
start
standardizing
T.
If
it's
only,
for
you
know
some
very
limited
use
they
want
to
have
you
know
lots
of
people,
they
would
actually
would
need
to
have
some
real
hardware,
vendors,
implementing
or
say
that
they're
going
to
be
opening
this.
Otherwise
they
did.
They
are
not,
and
actually
you
know
more
than
just
one
so.
B
Contact
for
six
dishes,
our
liaison
or
something
right
so
so
to
do,
for
the
chairs,
I
didn't
start
a
discussion
with
Pat,
Kimmi
and
yourself
and
see
because
I
mean
without
support
from
a
triple-a.
This
isn't
get
out
right.
It
has
to
happen
there
because
they
have
to
control
of
what
you're
suggesting
we
don't
have
without
okay.
Thank
you.
Okay
with
this,
we
are
closing
this
meeting.
So
thank
you
all!
It's
probably
not
the
last
sixties
meeting.
We
hope
we
meet
in
bangkok.