►
From YouTube: IETF105-6LO-20190722-1330
Description
6LO meeting session at IETF105
2019/07/22 1330
https://datatracker.ietf.org/meeting/105/proceedings/
A
A
A
This
is
the
note.
Well,
please
make
sure
that
you
have
read
it.
It
contains
a
number
of
pointers
to
policies,
ITF
policies
on
important
topics,
so
please
make
sure
that
you
have
familiar
with
it
and
now,
let's
go
through
the
agenda
for
today.
So
the
first
slot
is
the
usual
chairs
introduction.
That's
currently
in
progress,
then
the
next
slot
is
the
presentation
of
the
ipv6
over
NFC
document.
Basically
the
revisions
after
the
ISDN
evaluation.
A
Next,
initially,
there
was
scheduled
the
presentation
by
Charlie
about
the
deadline.
Time
draft
also
the
revisions
after
is
devaluation.
However,
he
will
present
this
remotely
and
he
will
do
it
at
the
end
of
the
session
because
of
some
unexpected
issues
that
have
a
beard.
So
unless
there's
any
objection,
the
idea,
the
proposal
would
be
to
move
Charlie's
slot
to
the
end
of
the
session.
A
A
Then
there
will
be
a
set
of
two
presentations
that
are
related
with
neighbor
discovery
that
will
be
given
by
Pascal,
starting
actually
by
the
second
one.
So
it's
not
the
same
order.
That's
shown
here.
If
there's
no
objection,
so
the
first
will
be
the
maintenance
proportional
tenant
scheme
for
RFC,
eighty
five
zero
five
and
next
there
will
be
the
any
and
the
unicast
lookup
and
by
the
way,
since
we
have
some
available
time
unless
there's
any
objection,
the
slot
for
the
unicast
lookup
would
also
be
of
15
minutes.
A
Finally,
there
will
be
the
presentation
of
new
work,
which
is
a
new
method
for
header
compression
given
by
Brian,
and
the
related
document
is
entitled
a
symmetric
ipv6
for
IOT
networks
and,
as
I
said
before,
actually
the
last
slot
will
be
the
one
by
charlie
unless
there's
any
problem.
So
is
there
any
comment
on
the
agenda
considering
these
proposed
changes.
A
Okay,
thank
you.
So
now,
let's
go
through
a
status
report
of
the
working
group
documents.
We
have
two
documents
that
have
been
evaluated
by
the
iesg.
These
are
the
ipv6
over
NFC
and
the
deadline
time
documents
there
was
a
number
of
outstanding
discuss
positions
on
both
and
both
documents
were
updated
before
the
cutoff
date,
and
the
next
step
will
be
seeing
whether
the
updates
are
sufficient
to
satisfactorily
address
the
comments
that
were
made.
A
You
can
see
on
the
slide
a
summary
of
the
iesg
ballots
as
they
are
today,
and
there
was
a
another
document
that
was
submitted
to
the
ASG.
However,
it
hasn't
been
yet
evaluated
by
the
SD,
which
is
the
address
protected,
maybe
discovery
draft
then
there's
a
set
of
four
other
documents
that
passed
working
group
last
call
the
first
two
are
the
documents
that
are
related
with
fragmentation,
and
these
are
the
six
open
fragment
forwarding
and
the
fragment
recovery
documents.
For
this,
the
Sheppard
review
has
been
done
and
both
documents
have
been
updated.
A
A
So
today
there
will
be
some
time
for
discussion
of
the
current
status
and
next
steps
for
this
document.
And
finally,
the
last
working
group
document
is
ipv6
over
PLC
which,
as
you
may
recall,
was
adopted
as
a
working
group
document
before
brac
and,
however,
has
not
been
updated.
Since
then,
so
perhaps
is
there
any
author
of
this
document
that
would
like
to
express
anything
on
the
mic
about
current
status
and
next
steps.
B
A
C
Okay,
this
is
the
every
time
I
show
you
the
kick
into
Shibata
NFC.
This
one
is
the
history
and
has
status
up
to
my
trust.
Actually,
there
was
the
I
GS.
He
reviews
on
March
I
have
a
lot
of
critical
discussion
and
then
I
produced
the
14
and
15
draft
version,
and
for
this
meeting,
I
will
give
you
some
kick.
Okay
insertion
to
preview
the
five
slides
the
first
one
is
marketing
is
pressure.
Actually,
there
was
some
something
marketing
stuff
might
route.
C
C
Next
one
is
the
section
for
thought,
one
about
the
figures
racially
this
one.
This
figure
is
actually
something
some
part
make
show
it.
The
physical
and
ink
layer
is
redundant
to
the
ticket
figure
number
two,
so
the
viewer
gave
me.
The
third
suggestion
make
this
one
another
with
new
figure.
I
just
simplified
this
one,
like
the
new
figure.
I,
think
it's
better,
and
this
one
is
the
about
the
nd
section.
C
The
4.5
there
was
some
typos
and
some
major
changes
and
rise
to
sentence
a
sure
it
is
not
really
refer
to
under
since
fight,
so
I
just
decided
to
leave
it.
This
is
okay,
thank
you,
and
next
one
is
the
last
one.
Actually,
it
is
of
a
security
section.
The
review
gave
me
t
the
last
second,
the
sentence.
C
This
one,
actually,
he
given
me
the
suggestion.
This
sentence
should
be
go
to
the
front
in
this
D
section,
so
I
just
accepted
one
and
the
next.
The
second,
the
power
of
actually
is
not
really
clear,
because
this
is
final
version,
so
I
just
remove
it
and
the
the
last
section
it
tells
to
leave
you
give
me
some
division
divides
the
text.
Just
I
accept
that
the
they
weren't
just
I
put
his
suggestion.
Thank
you.
This
is
the
end.
The
end
case
on
their
comment.
Thank
you.
A
C
D
So
we
have
done
some
work
of
all
that
ten
years
on
fragments
at
six
slow.
Initially,
it
started
with
a
very
simple
fragmentation
mechanism
whereby
you
just
cut
message
in
pieces
and
throw
the
pieces,
but
we
would
not
discuss
if
there
is
any
primer
flow
control
with
those
pieces.
If
there
is
any
crime
of
interaction
between
the
previous
pieces
as
they
get
forwarded
about
the
network
like
hidden
terminal
issues
between
various
fragments
things
like
that,
and
we
did
not
even
discuss
how
fragments
could
be
forwarded
without
react.
D
Reassembly
at
every
hub,
so
the
assumption
was
fragments
would
be
various
about
every
hub.
So
we
looked
at
it
in
this
working
group
and
found
that
you
could
actually
get
a
better
result
if
you
would
streamline
the
forwarding
of
fragments,
but
there
are
rules
in
order
to
do
that
unless
otherwise
you
get
interactions
between
the
fragments
and
you
lose
a
lot
of
them.
D
So
we
we
ended
up
with
one
document
in
this
group
and
Gaston
wrote
another
document
that
week
when
we
explained
how,
without
changing
the
formats
in
49:44,
you
can
actually
avoid
reassembling
the
packet
at
every
hop
but
forward
the
fragment
over
Helios
free
path
over
multi
hops.
So
the
technique
is
effectively
documented
in
this
minimal
fragment
raft.
So
mini-mall
doesn't
change
the
formats
of
RC
49:34.
D
D
Now
we
are
still
in
a
world
where
the
fragments
are
sent
once
and
for
all,
and
you
expect
that
all
the
fragments
would
be
received
on
the
other
side
if
they
are
not.
There
is
no
way
for
the
sender
to
know
about
this.
Apart
from
timing
out
at
the
upper
layer,
which
means
that
if
a
fragment
is
lost,
the
receiver
will
wait
till
death
with
this
buffer
locked
with
a
big
message
in
it.
D
So
what
we
did
for
that
is.
We
looked
at
the
primes
of
retransmitting,
individual
fragments,
and
so
for
that.
That's
the
third
draft
that
the
fragments
recovery
draft-
and
we
also
looked
at
questions
like
how
what
kind
of
interval
you
need
to
place
between
packets,
to
make
sure
which,
between
fragments,
to
make
sure
that
the
next
fragment
does
not
interfere
with
the
previous
fragment.
D
So
there
we
go.
These
are
the
this.
Is
the
history
of
what
we
are
doing
so
49
40
44
is
quite
old
now,
but
those
two
are
just
completing.
Well
group:
let's
go
as
carelessness
just
told
us,
so
the
changes
in
minimum
fragments
we
had
some
commands.
We
had
to
to
great
reviews
by
yeah
and
and
jojos
thank
you
dojo.
So
they
don't
see
ya
in
the
room.
Oh
yes,
thank
you!
Yet
so,
and
then
we
get
your
ratio
commas.
D
D
We
added
a
section
in
the
security
section
about
the
attacks
against
the
VRB
and
and
what
is
the
exposure
and
the
exposure
is
not
that
big,
because
you
don't
keep
much
data
for
each
individual
fragments
right.
So
it's
a
very
small
data
per
fragment,
so
you
can
still
be
attacked
in
the
middle
of
the
network
by
you,
throwing
at
you
many
fragments
of
fake
fragments
which
appear
to
come
from
different
data
ground
so
which
you
need
to
create
a
V
or
B,
but
it
takes
many
of
those
packets.
D
Before
it
starts
representing
a
lot
of
memory,
we
improved
the
cloud,
the
quality
of
the
tags
that
says
how
you
manage
the
vrb
is
like
when
you
create
them
when
you
discard
them
and
yes,
we
we
clarified
things
which
were
not
written
correctly
or
not.
Free
obvious.
You
only
wrote
on
the
first
fragment
the
operation
that
you
do
for
forwarding
fragment
is
create
a
form
of
MPLS
State.
D
You
use
the
Datagram
tag
as
an
MPLS
label,
and
the
first
really
as
it
is
the
first
fragment
is
routed
by
the
IP
header
and
you
lay
out
the
fragment
as
you
go.
The
label
as
you
go
and
then
the
other
fragments
follow
the
label
path
and
you
actually
switch
the
data
gram
tag
as
if
it
was
a
label
how
by
hop
and
that's
how
you
avoid
some
collisions
that
we
described
in
this
in
this
group.
D
So
these
are
the
challenges
for
the
fragment
falling
thing.
The
farmers
recovery
follows
the
same
logic
for
the
initial
path
that
is
created
by
the
first
fragment
based
on
the
IP
address.
What
it
doesn't
up
is
provide
a
bitmap
each
bit
in
the
bitmap
of
acknowledgment.
Bitmap
tells
you
about
one
individual
fragment,
so
you
can
actually
retry
any
particular
fragment
individually.
D
So
you've
got
rules
on
how
you
manage
the
Internet
via
bees,
and
you
find
that,
because
you
have
this
additional
signaling,
the
management
of
the
BMV
RB
can
be
bad.
For
instance,
you
know
when
the
receiver
has
fully
received
all
the
fragments.
That's
when
you
can
destroy
the
VRP.
You
don't
know
that
if
you
don't
have
acknowledgements,
oh
it's
an
addition
to
the
text
that
could
exist
in
the
normal
fragment
fording.
D
So
we
discuss
with
we've
the
chair
here
with
a
shepherd
actually
shepherd
on
whether
that
meant
an
update
of
the
other
spec
or
just
an
extension,
and
we
found
that
it's
not
real
dating
the
other
draft,
because
the
methods
we
are
hiding
come
with
the
fact
that
we
are
hiding
something
in
the
protocol
as
well.
So
so
we
didn't
write
update,
we
wrote
extend
that
we
did
not
indicate
in
the
Arab
theatre
that
we
update
minimal
fragment
with
the
recoverable
one
at
the
definition
clarify
number
of
things.
D
E
Not
a
question
I'm
here,
it's
hard
to
find
me
but
yeah,
so
one
of
the
things
that
you've
identified.
This
is
not
like
just
a
problem
here.
This
updates
thing
is
an
issue
for
everybody
in
the
idea
of
pretty
much
like
we
used
the
updates
tag
to
mean
different
things
at
different
times,
so
a
few
of
us
on
there
years.
E
You
have
taken
upon
ourselves
to
write
something
to
clarify
this
and
just
to
give
you
like
a
brief
intro
right
like
it's
gonna,
be
an
individual
drafts
coming
out
right
now,
it's
Raleigh
today
or
tomorrow,
it'll
be
out.
You
want
to
replace
the
updates
tag
with
something
else.
Okay,
so
right
now
the
goals
are
like.
One
of
them
is
like
you
know
it
changes
things
or
adds
things
like.
So
we
want
to
separate
these
things
out
so
like
what
the
update
means
is
gonna
become
clearer.
That's
really
the
goal
I
think
like
by
this
time.
E
This
like
hits
the
iesg
you
might
have.
This
have
like
a
initial
idea
of
like
whether
this
is
gonna
fly
or
not,
but
it's
not
probably
going
to
be
final,
but
I
think
at
least
like
in
the
text.
You
could
explain
how
something
abase,
that's
like,
probably
going
to
be
the
stopgap
arrangement
for
going
forward
with
this
right.
So
you
just
say:
okay,
this
is
what
we
do.
Cuz
like,
like
you
said
right
like
it's
an
it's
an
extension
point,
so
you
decide
what
to
say,
but
so
one
of
the
things
is
like.
E
So
if
I
started
digging
back
a
bit
right,
why
do
we
have
an
update
stag
in
an
RFC
right?
So
the
thing
is
like
the
RFC
that
got
updated
right.
The
only
way
it
can
get
pretty
immutable
to
point
to
something
that
has
more
information
is
the
update
stack,
so
I
would
rather
be
like
liberal
and
assigning
the
tag.
If
you
think
is
relevant
for
somebody
reading
the
updated
RFC
to
read
this.
That's.
E
It's
not
that's
what
I'm
saying
so
like
I'll,
rather
be
liberal
and
doing
that
right
now
until
we
go
through
this
proper
cleanup
procedure.
So
just
because,
like
somebody's
prolly
gonna
ask
like
somewhere
down
the
line
right
like
and
I,
would
rather
be
liberal
here
and
then
remove
it.
If
somebody
says
like,
why
does
it
do
it
as
long
as
you
explain
it
I
think
they're
gonna
be
okay.
Yes,.
D
And
so
weird
and
I
guess
this
case
is
one
you
could
have
in
your
thought
process
as
you
define
these
new
tags,
because
in
this
case
the
minimum
is
still
self
sufficient.
The
addition,
don't
need
to
be
known
by
somebody
wins,
does
mini-mall
for
RC
49:44.
This
is
a
closed
world
that
it's
enough
and
there
is
nothing
to
add
to
it.
Yes,.
D
D
D
Forwarding
fragments,
as
we
go
to
the
next
slide,
where
is
more
complex,
that
meets
the
eye,
because
if
you,
if
you
just
push
the
fragments,
bang,
bang,
bang
bang
and
you
operate
your
mesh
on
a
single
frequency.
Clearly,
the
first
fragment
as
it
progresses
from
the
first
to
the
second
half
well
for
the
second
to
the
third
hub
will
collision
on
the
second
hub
with
the
second
fragment,
which
comes
from
the
first
up
to
the
second
hub.
It's
like,
because
you
think
you
forward
in
this
way,
actually
because
the
radio
is
broadcast.
D
It's
also
for
us,
the
other
way
and
the
to
interfere.
So
you
get
immediately
hidden
terminal
and
the
second
hop
loses
the
second
packets.
Basically,
so
it's
something
that
that
forces
you
to
pace
down
your
fragment,
it's
something
you
need
to
know
now.
If
you
operate
on
different
channels
like
you
use
the
ACH,
for
instance,
the
the
you
still
need
to
pace
a
little
bit
down,
because
when
this,
the
second
fragment
is
sailed.
The
receiver
on
this
on
the
second
heart
must
be
open
for
receiving
meaning
that
it
must
already
have
sent
the
first
fragment.
D
So
the
center
of
the
second
fragment
must
wait
till
the
receiver
of
the
first
fragment
has
done
transmitted
transmitting
right.
So
you
still
need
to
insert
some
time
here.
Even
if
you
don't
have
hidden
terminal
problems,
because
there
you
need
to
be
able
to
turn
the
receiver
horn
on
this
on
the
receiver,
so
it
must
have
sailed
the
first
fragment.
So
it
are
a
lot
of
non-obvious
things
like
this.
G
Okay
hi,
my
name
is
Jung
Hwan,
home
I
will
present
the
sixth
row
applicability
and
Yusuke
document.
So
actually,
artists
raft
was
not
update
after
last
primary
about
we
have
our
email
discussion
or
between
the
authors,
and
we
have
some.
The
block
consists
of
a
lot
conclusion
how
to
improve
and
help
to
update
his
trip.
So
this
time
I
will
shortly
it
is
described
all
result
of
the
email
discussion
shown.
G
The
first
one
is
the
also
always
to
adjust
the
scope,
because
the
comment
come
from
the
last
meeting,
so
we
will
under
or
down,
and
the
pro
course
on
and
we
found
there.
Some
section
was
not
considered
between
each
section,
so
we
make
confidence
between
each
section
and
we
debating
total
determine
the
six
row
technology.
The
first
one
is
LT,
empty
C
is
removed
already,
but
we
are
considering
how
about
handling
the
I
Triple
E
area
to
that
15,
the
poor
technology,
because
Arab
see
six
five.
G
Six
eight
already
provide
6lowpan
use
cases,
but
in
this
prep
thought
are
typical
on
deployment
of
the
six
row,
and
it
is
the
Jupiter
mesh
and
whites
on
the
two
technology
is
our
typical
six
royalty
premiere
scenario
and
it
is
based
on
a
decrypting.
The
poor
technology
and
another
technology
is
a
triple
aerial
to
the
attenti
echo
technology
or
is
I.
Remember
that
it
come
from
the
Samhita
comments.
M,
it
has
suggested
a
triple-a
area
to
that
pivotal.
G
D
Just
a
mention
that,
for
he
has
been
wrapped
into
fifteen
fourths,
which
is
part
of
15
for
now
the
one
you
mentioned,
why
centripetal
measure
in
the
four
chief
Remini
world,
the
Phi
is
actually
subdeacon.
You
go
wrong
distance,
whereas
for
he
was
in
the
Y
could
be
used
for
4G,
but
it's
Annabel,
Mac
technology,
like
TCH
density
general
being,
is
part
of
the
for
he
amendment.
No,
it's
all
wrapped
up
into
54,
so
it
there's
no
reason
to
mention
for
he
anymore.
Okay,.
G
G
It
includes
design
space
based
on
the
city
open
technology,
but
are
we
identifier,
another
design,
space
dimension,
po6
wrote
deployment
because
sixth
row
technology
include
a
wider
technology,
for
example
the
net
world
PFC,
so
it
there
will
be
a
different
or
design
space
or
design
space
dimension,
but
we
found
that
each
region,
space
dimension,
is
not
laced
to
each
six
rhotic
analogy
so
to
narrow
down
this
rep.
We
would
like
to
remove
the
section
five
point
one.
G
A
Think
the
last
point
is
the
clearest
action
that
we
can
do.
It
doesn't
hurt.
If
we
remove
subsection
5.1
all
together,
then
the
one
in
the
middle
is
a
bit
more
tricky.
There's
the
question
of
currently
in
the
description
of
technologies.
We
have
subsection
4,
15.4
II.
The
question
might
be:
ok,
should
we
have
a
section
subsection
dedicated
to
15.4
in
general?
Perhaps
yes,
also
Pascal
just
mentioned
that
now
for
e4g
are
integrated
anyway
in
1504,
although
that
might
open
a
bit
the
scope
as
well.
So
that's
a
little
bit
tricky.
G
H
Question
to
the
authors:
how
do
you
want
to
progress?
It
was
in
workgroup
last
call's.
We
didn't,
we
didn't
submit
it
back.
We
didn't
get
any
comments.
We
didn't
submit
it
to
the
IHC
review.
So
how
do
you
want
to
proceed
it?
You
want
to
devise
the
raft
and
go
for
another
workgroup
last
call,
or
you
want
to
gather
some
thoughts
from
the
workgroup
now
to
see
if
there
is
interest
in
actually
progressing
this
draft
and
see
if
there
is
an
archival
value
for
this
job.
At
this
point,
okay,.
D
So
this
is
an
update
that
I
prepared
not
for
this
group
actually
but
for
six
man,
but
then
talking
with
the
chairs,
we
we
thought
were
Levin
to
to
give
this
group
an
update
on
it,
because
it
reveals
on
what
we
have
produced
in
six
low
and
that
now
may
continue
its
life
at
six
man.
So
for
those
who
did
not
follow
the
story,
now
is
your
chance
to
to
get
to
get
it
kind
of
the
full
full
story
of
wireless
indie.
D
So
in
the
actually
the
the
work
we
started
with
RFC
6775,
which
I
should
have
written
here
so
RFC
6775
in
twenty
five
introduced
the
concept
of
a
registration,
which
is
how
a
host
can
tell
the
router
a
I.
Have
this
IP
address
so
I
formed
this
IP
address?
Can
I
use
it?
Is
it
unique
in
the
network
etc
and
the
router
using
a
central
registrar?
We
call
the
six
LVR
is
capable
of
saying.
Yes,
this
address
is
unique.
D
You
may
use
it
basically
and
now
the
router
is,
and
the
6lb
are,
can
protect
the
ownership
of
this
advice
to
this
host.
So
that
was
the
beginning
of
the
story
of
this
interface.
We
mostly
can
do
duplicate
a
trace
detection,
but
there
is
no
security
mechanism
to
ensure
that
a
bad
guy
will
not
try
to
register
the
same
address
with
the
same
MAC
address
and
still
the
traffic
and
be
able
to
sauce
traffic
on
behalf
of
the
other
guys.
So
for
good
willing
guys.
D
This
is
dad
equivalent
what
we
have
on
wire,
but
for
attacks.
It's
just
as
what
we
see
on
wire,
it's
possible
to
attack
this
and
claim
the
traffic
for
somebody
else.
So
that's
sixty
seven,
twenty
five,
eight
five,
four
five
updated
this
changing
little
bit.
The
way
we
use
link
local
drives,
making
it
simpler
and
also
changing,
which
field
advertise
that
so
it's
mostly
conceptually
the
same
thing
in
terms
of
exact
form,
mati
changes
a
bit
and
then
we
added
something
which
is
important
in
some
environments,
which
is
a
sequence
counter.
D
For
instance,
we
refer
to
this
Draft
at
five,
four
five
in
datacenter
applications
where
VM
can
be
a
mess,
can
move
and
the
fact
that
you
have
the
sequence
counter
and
each
time
the
VM
moves.
It
declares
itself
to
the
router,
then
the
routing
infrastructure
alcohol
can
always
know
which
is
the
most
recent
location
for
this
virtual
machine.
It's
a
prime!
D
D
D
So
if
you
think
for
a
minute
of
what
802,
that
11
is
providing
you
@ly,
au,
@ly
or
2
in
the
infrastructure
mode
infrastructure
mode
provides
you
with
an
emulation
of
a
broadcast
domain.
So
you
can
operate.
Your
ipv6
stack
exactly
quote-unquote
as
if
it
was
Ethernet
and
to
emulate
the
broadcast
domain
with
multiple
wireless
nodes,
which
is
each
of
a
different
broadcast
domain.
D
They
all
agree
that
the
broadcast
domain
of
reference
is
that
of
the
access
point
right,
meaning
that
if
you
want
to
send
a
broadcast
from
a
station,
what
you
actually
do
is
say
to
the
AP
AP
and
the
AP
would
broadcast
it.
So
everybody
is
who
can
receive.
This
broadcast
must
be
in
the
broadcast
domain
of
the
AP
or
the
APS.
If
you
have
an
extended
service
set,
what
we
are
doing
with
this
registration
and
this
backbone
router
is,
we
are
pretty
much
doing
the
exact
same
thing
at
layer
3.
D
Now,
if
you
wonder
why
that
11
did
the
way
they
did
well
think
if
I
mean
it
about
how
a
transport
proxy
operates.
You
know
those
loading
bridges.
What
they
do
is
when
they
don't
know
what
to
market
right
where
her
MAC
address
is
they
will
just
flood
the
network
with
a
broadcast
and
when,
when
a
packet
comes
back
from
that
MAC
address
that
they
will
of
you
on
the
bridge
and
say:
oh
that
MAC
addresses
on
this
port.
D
If
you
want
it
to
do
the
same
thing
at
a
your
to
on
Wi-Fi,
you
would
end
up
broadcasting
and
in
you
look
out
for
any
MAC
address
and
from
the
wireless
perspective,
that's
completely
unacceptable,
because
broadcast
is
actually
expensive,
under-delivered.
So
from
the
start,
dot
11
designer
said:
hey,
hey
I
can't
have
this
reactive
operation.
I
will
be
proactive,
so
part
of
what
you
do
with
an
association
process.
Is
you
actually
pre
establish
a
bridging
state
in
the
access
point,
so
when
there
is
a
lookup
a
broadcast?
D
The
access
point
already
knows
whether
the
MAC
address
is
attached
to
it
or
not.
So
it
will
actually
cancel
the
broadcast,
though
it
will
not
copy
it
on
the
wireless.
If
there
is
nobody,
interests
is
nowhere
side
right.
So
basically
you
avoid
the
cost
of
this
lookup
and
the
broadcast.
That's
associated
to
it.
What
we
do
we
need
five
for
five
and
the
backbone
router
is
we
do
the
exact
same
thing,
but
at
layer
3,
wait!
Five
for
five
becomes
the
Association
and
the
backbone
router.
Is
this
transparent
bridge
but
atlas
ray?
D
So
when
there
is
this
broadcast
happening,
you
know
and
nslookup,
not
the
access
point
to
lay
off
three
access
point
can
recognize
whether
the
IP
address
is
already
associated
register
to
it.
So
it
can
know
whether
it's
useful
to
just
cancel
the
recap
or
turn
it
into
unicast
or
just
reply
on
behalf
with
the
backbone
router.
You
can
do
all
those
operations
right.
So
we've
reached
the
same
state
with
this
this
method
in
ipv6.
Well,
you
you
can
avoid
the
cost
of
the
loop
gap
over
the
wireless
network.
D
If
you
operate
in
a
registration
mode,
you
must
think
again,
like
an
APU.
Nato
station
does
not
sound
a
packet
to
another
station
in
infrastructure
mode.
It
sends
the
packet
to
the
AP
and
the
AP
sends
it
to
the
other
station
because
you
have
to
live
within
the
broadcast
domain
on
the
access
point
in
at
layer
3
when
we
do
the
backbone
router
and
at
5:05
the
the
prefixes
are
advertised
as
not
on
link
because
you,
your
desired
operation.
Is
that
Li
the
same
thing?
You
want
the
packet
to
go
to
the
router.
D
That
is
we
access
point
and
the
router
forward
the
packet
back
to
the
destination
station
right
so
by
protecting
the
ownership
of
the
address
at
the
router,
using
some
crypto
mechanism,
proof
of
ownership,
etc.
We
can
provide
the
equivalent
service
of
sound
without
distributing
all
the
send
stack
to
every
devices.
It's
just
between
the
device
and
the
router.
D
It's
quite
easy
to
do
because,
as
your
register,
the
detrás
for
the
first
time,
then
you
can
give
the
token
instead
of
a
MAC
address
now
a
crypto
token,
and
you
can
use
it
for
proof
of
ownership
later.
So
if
your
IP
address
moves
from
earliest
really
to
another
la
3
AP,
the
AP
can
challenge
you
for
the
ownership
of
this
token,
which
is
stored
in
the
infrastructure,
and
so
you
can
prove
that
you're
still
the
guy
who
registered
this
address
for
the
first
time,
first-come
first-serve.
So
we
can
actually
deliver
the
savvy
properties.
D
We
can
do
so
that
res
validation,
I
will
only
for
the
packet
if
it
comes
from
the
rightful
owner
and
they
will
not
let
an
attacker
get
on
the
trace
that
belong
somebody
else,
because
he
cannot
prove
the
ownership.
So
that's
pretty
much
what
IA
pnd
does
so.
With
this
we
were
kind
of
done
emulating.
We
are
kind
of
done
emulating
an
aunt
of
an
infrastructure
ESS
at
layer
3.
D
If
you
work
on
overlays
and
big
fabrics,
you're
already
used
to
the
fact
that
if
you
want
to
establish
a
VX
land
between
an
ingress
and
egress,
you
have
to
resolve
the
egress
point
for
given
IP
address
right,
so
that
there
is
this
notion
that
there
is
the
central
database
somewhere
IMAP
server,
map-resolver
of
sort
which
knows
every
IP
address
on
the
network
and
varieties.
Because
you
need
you
need
to
establish
another
later
on
or
between
the
ingress
and
egress.
D
D
This
simple
interface
that
would
allow
any
IP
tries
to
tell
the
map
server
I'm
here
so
that
when
there
is
a
lookup
instead
of
having
to
broadcast
over
all
the
fabric,
you
can
actually
just
look
on
the
map
server,
map-resolver,
to
know
the
MAC
address
and
be
done,
and
with
this
you
save
all
this
broadcast
that
nd
is
doing
on
your
fabric,
so
that
was
that
was
something
which
just
is
a
simple
extension
to
what
we
have
in
wireless.
But
it's
really
not
done
for
wireless,
it's
redone
for
wire.
D
So
that's
that's
one
of
those
reasons
why
I
tend
to
think
that
the
work
which
I
started
as
a
six
low
document
code,
maybe
should
be
actually
a
six
man
document
right
and
that's
why
I'm
presenting
those
things,
this
presentation
and
the
unicast
look
at
that
six
band
tomorrow.
So
I
wanted
this
section
here.
Well,
this
presentation
was
to
give
them
the
story,
so
now
they
know
where
the
unique
a
circuit
comes
from,
but
for
people
you
know
I
talked
to
it's
in
Cisco,
it's
more
like
I'm
telling
them
a
with
a
remote
I'm.
D
After
is
a
stuttered
in
face,
enable
discovery
to
populate
the
map
server
so
that
you
know
the
devices
don't
have
to
talk
list
because
they
don't,
and
so
this
new
draft
is
actually
something
which
is
targeted
to
six
man
to
give
them
the
whole
story
to
show
them
that
you
know
in
a
world
where
the
plane
is
on.
That
is
not
the
only
case
where
you
have
a
lot
of
wireless
with
or
without
broadcast
emulation
at
the
rear
too.
While
you
have
data,
centers,
overlays,
etc.
D
The
methods
that
we
defined
in
last
century
for
a
yellow
cable
can
be
slightly
obsolete
and
that
we
may
want
to
actually
refresh
Andy
and
its
evolve
or
die.
Basically,
some
people
say:
okay,
let's
just
do
point
to
point
and
be
done
with
it
and
then
jeez
becomes
almost
nothing.
It's
becomes
lldp
or
whatever,
or
you
actually
refresh
it
and
modernize
it
so
that
you
can
use
still
use
it
and
build
large
subnets
with
it
I'm
in
favor
of
building
large
subnets
I'm
in
favor
of
because
we
do
it.
D
D
It's
no
surprise,
that's
what
we
do
every
day,
but
for
people
coming
from
a
more
traditional
world
of
6-man,
it's
more
of
a
shock,
because
some
people
never
wanted
to
hear
about
multi-link
subnet.
There
is
this
desire
that
the
subnet
is
a
link
and
the
link
is
a
subnet.
Some
people
don't
even
realize,
or
one
is
physical,
the
other
one
is
logical,
and
actually
you
could
map
the
logical
thing
and
bigger
than
an
Inca
and
smaller
than
the
link
for
people
because
of
Ethernet.
They
see
that
as
pretty
much
the
same
thing
anyway.
D
D
So
that
was
a
long
introduction,
but
you
told
me
at
yet
time
to
more
I
want
so
I
hope
that
well
I'm
happy
that
you're
here,
because
tomorrow,
I
won't
be
able
to
say
all
those
world,
so
basically
nd.
The
way
it
was
designed
was
for
point-to-point
and
transit
link,
it's
written
in
big
letter
in
the
draft,
and
that's
that's
because
it
uses
broadcast
use
it
layer
to
broadcast.
We
people
tell
me:
oh
it's,
you
know
it's
multicast
and
say
what
joke
is
that?
D
Because
there
is
no
such
a
thing
at
Lesley
automatic
as
well,
there
is
nobody
deployed
it
and
anyway
it's
not
my
aunt.
The
one
that
I
Triple
E
has
produced
is
not
meant
to
scale
to
to
add
the
power
of
24
bits
alright.
So
there
is
no
such
thing
as
a
layer
of
two
sorries
did
not
multicast
don't
dream.
So
when
we,
you
know,
with
our
good
face,
send
packets
to
two's
multicast
group.
The
first
thing
that
the
the
eye
area
does
is
place.
D
The
first
bit
of
the
MAC
address
to
one
and
the
thing
gets
broadcasted
through
the
fabric
you
there
is
no
layer,
two
magic
to
solve
the
problem
that
we
did
not
solve.
We
just
pushed
it
about
over
the
fence,
hoping
that
they
are
Tripoli
would
solve
this,
for
us
never
happened
right,
so
I
Wi-Fi
does
a
emulation
of
a
transit
link
and
that's
why
ipv6
of
our
Ethernet
can
be
translated
into
IP
six
over
Wi-Fi,
but
it
has
impacts
because
radios
are
not
wires,
in
particular
multicast
on
the
radio.
D
There
are
number
of
things
which
are
kind
of
interesting,
which
happen
when
you
send
a
multicast
versus
a
unicast.
When
you
send
a
multicast,
you
have
to
send
it
at
the
slowest
speed
possible,
because
you
want
to
reach
everybody
who
is
associated
to
you.
Ok,
it.
It
means
that
some
packets
can
takes
100
times
the
airspace
of
a
unicast
packet
that
could
be
set.
Much
faster.
D
Second
thing
is:
when
you
do
broadcast,
you
cannot
be
informed
right.
You
have
to
send
it.
Multi-Directional
omnidirectional
was
when
you
do
unicast,
you
can
be
in
form
and
then
multiplex,
no
MIMO,
multiple
of
multiple
packets.
So
when
I
hear
that
doing
p2p
doing
broadcast
over
IP
to
piece
is
like
doing
a
unicast
or
not
a
unicast,
it's
wrong,
because
the
fire
won't
do
the
same
thing.
D
If
you
tell
him
it's
a
broadcast
and
if
you
tell
unicast,
even
if
you
aim
at
a
single
person-
and
you
have
a
logical,
p2p
connection
with
that
person,
if
you
don't
tell
the
fie
that
it
is
a
unicast,
if
it
does
the
broadcast
bit,
it
will
tune
the
radio
differently,
it
will
be
slower,
it
will
be
omnidirectional,
okay.
So
it's
important
to
to
turn
any
package
again
into
a
unicast
okay.
D
That's
designed
for
cars,
where
you
don't
have
the
infrastructure
mode,
you
don't
have
VSS
and
VSS
you're
naked
you're,
just
like
on
15
for
ipv6
of
out
of
5,
meaning
that
the
broadcast
domain
are
those
of
the
sender,
not
those
of
the
access
point,
meaning
that
all
of
the
assumptions
that
we
made
that
there
is
a
single
broadcast
domain,
so
we
can
do
ipv6
and
D
the
way
it
stands.
All
this
is
broken,
doesn't
work
right.
D
Even
if
there
is
a
mac
layer,
emulation
of
the
transit
domain
of
a
fully
full
broadcast
domain,
everybody
is
in
it
hurts
it
hurts,
because
the
broadcast
is
slow.
It
hurts
because
when
there
are
so
many
broadcast,
your
batteries
is
rapidly
depleted.
I,
don't
know
if
you
found
it,
but
I
found
that
my
phone
depleted
twice
as
fast
at
the
ietf
them
at
home,
its
processing,
all
this
broadcast
its
processing,
so
many
broadcasts
that
the
phones
usually
have
protective
methods
by
which
to
filter
a
number
of
them.
D
If
the
filter
a
number
of
them
and
one
of
them
is
important.
Well,
it's
not
retried
like.
If
I
do
this
dad
over
my
my
fabric,
there
is
a
good
chance
that
the
number
of
the
phones
will
just
filter
it
out
because
they
keep
one
out
of
five
or
something
like
that.
Well,
the
end
result,
if
my
dad
operation
will
fail.
So
how
come
a
music
seems
to
work
because
the
duplication
don't
happen
right.
D
You
have
to
force
a
duplication
in
v6
to
resume
on,
if
you
like,
if
you
leave
it
to
make
a
try,
so
I'll
run
down,
you
know
all
the
heiress's.
We
are
from
six
months
the
chances
of
a
duplication
of
so
low
that
the
the
user
perspective
of
the
service
is
always
working,
but
in
fact
the
cooperation
is
below
behind
it
that
doesn't
work.
We
tried
it
at
the
IDF.
We
found
that
on
a
big
meeting
like
this
one,
that
out
of
five,
discovers
the
duplication.
It's
like
it's
completely
useless.
D
It
works
because
the
prime
does
not
occur
not
because
we
know
how
to
solve
it
right
so
having
the
infrastructure
know
the
addresses
of
everybody
in
a
registrar
mapserver.
Now
we
can
redo
the
dad
and
we
don't
have
to
have
the
radios
always
on
to
defend
himself,
because
we
have
an
infrastructure
service
for
doing
it.
D
Ok
and
there
are
number
of
other
mismatches,
actually
how
you
tell
me
when
them
take
too
much
time
right,
but
I
think
this
is
so
interesting
right.
I
want
to
share
with
you
I
want
to
make
sure
that
that's
your
in
you
know
on
the
right
page
and
all
those
things
because
it
doesn't
doesn't
meet
the
eye.
I
mean
it's
something
that
you
have
to
think
quite
through
before
you
read,
get
your
mind
on
it.
So
remember
further
mismatches
with
traditional
Andy,
the
first
one
is
that
you
have
an
office
roaming.
D
First
roaming
expect
that
you
can
just
get
there
and
and
the
infrastructure
will
always
know
where
you
are
at
leav
3.
That's
probably
true
right,
because
your
reassociate
rapidly
better,
they
are
too,
but
that
they
are
three.
When
you
have
this,
you
snoop
the
protocols
and
try
to
figure
out
well
the
latest
location
of
a
particular
devices,
and
we
are
doing
that
today,
this
network.
D
If
the
guy
moves
rapidly
with
that
11
now,
the
message
is
that
we
sound
about
the
fabric
to
say:
I've
seen
him
here,
I've
seen
him
here
might
arrive
in
disorder,
meaning
that
the
really
infrastructure,
I
think
is
here
and
actually
is
there.
Now,
if
you
use
that
for
verax
land-
and
you
lose
all
your
packets,
because
tunnels
are
not
rotted
to
right
place,
so
you
better
not
run
too
fast
at
the
first
message.
Second,
one
is
intermittent
connectivity.
If
you
want
to
save
your
battery,
you
don't
receive
all
packets
all
the
time
so
I.
D
D
D
Your
station,
you
just
joined
you,
send
a
few
packets
to
the
Google's
of
the
world
to
see
what
kind
of
connectivity
you
are
you
have
and
and
the
server
somewhere
and
the
infra
starts
responding
to
you,
the
router,
the
first
time
it
sees
one
of
those
packets.
It
will
punt
it
to
software.
The
software
will
send
an
NG
message,
waits
and
the
nurse
wait
for
another.
D
During
that
time,
if
more
packets
arrive,
the
packets
will
be
dropped
because
the
rotor
wants
to
protect
itself
right
and
after
the
DNA
comes
back,
maybe
one
packet
will
be
sent
to
the
station,
which
will
say:
hey,
I
have
a
very
little
mini
mated
connectivity
on
that
network.
Let
me
retry
in
five
seconds,
so
you
can
put
all
the
faults
you
want
a
day
or
two.
The
layer-3
did
not
proactively
set
the
ante
cache
on
the
router
before
the
first
packet
comes
from
the
infrastructure.
D
As
a
response,
you're
screwed
right,
you
can
say
40,
milliseconds
or
100
milliseconds
at
the
Mecca
yarn.
You
lose
10
seconds
at
303,
that's
kind
of
sad
right
so
that
that's
another
miss
from
ng,
because
n
D
was
designed
at
the
time
we
are
doing
some
software
operation
on
the
fly
of
the
received
packet
was
okay
right
now
we
do
everything
in
silicon
and
we
certainly
punching
a
packet
to
the
software
to
do
an
and
the
resolution
is
something
completely
absurd.
D
So
now
to
get
there,
we
need
to
think
a
little
bit.
The
first
thing
we
have
to
think
of
is:
what's
my
link
model.
What
you
have
at
the
file
level
is
this
broadcast
domain
right,
so
you've
got
you
get
mr.
high,
which
has
a
broadcast
domain
and
I
can
broadcast
on
this
and
you've
got
mr.
B,
which
has
this
broadcast
domain.
Is
that
my
IP
link?
D
Well,
some
people
say
it's
my
opinon,
it's
my
broadcast
domain,
so
quote-unquote
I
can
talk
to
whoever
is
mine,
but
in
my
broadcast
domain,
because
I
expect
that
to
be
reflexive,
which
is
true
or
not
true,
it
can
be
true
if
there
is
no
no
interviewer
and
we
have
the
the
same
antennas
and
we
have
the
same
power
levels
etc.
In
our
Wi-Fi
infrastructure.
D
We
find
that
there
is
a
ratio
1
to
10
between
the
quality
of
the
signal
in
both
directions
right,
even
if,
if
even
if
the
station
adapts
the
power
to
that
of
the
IP,
which
is
supposed
to
do
the
AP
suppose
say:
hey
I'm,
sending
to
you
at
this
power,
the
station
is
supposed
to
use
the
same
power
back,
but
actually
the
antennas
are
not
the
same.
Nothing
is
the
same,
and
even
the
quality
of
the
the
chain
is
not
the
same.
So
at
the
end
of
the
day,
you
don't
get
the
same
signal.
D
It's
not
really
bi-directional!
It's
not
reflexive,
but
it's
mostly
reflexive.
What
is
not
is
transitive.
So
basically,
you
can
be
out
this
link
when
two
people
are
in
each
other's
broadcast
domain.
So
is
that
what
you
call
the
link?
I
tend
to
think:
that's
what
I
want
to
call
a
leak?
Why?
Because
I
want
to
form
a
link,
local
address
yeah,
and
there
is
no
point
no
way
to
do
dad
once
and
for
all
on
on
the
radio
interface,
which
is
always
on,
say
my
my
a
CB
radio
in
my
car,
always
on
can
I.
D
Do
that
for
an
IP
address
in
the
morning
and
drive
through
the
world
I
mean
obviously
that
there's
nothing,
because
who
was
in
the
ranch
when
I
did
it
at
8:00
in
the
morning
at
home
and
then
I
drive
through
town,
if
there
is
duplication,
I
could
not
detect
it
from
home.
What's
the
point
and
I
never
turn
down?
My
interface
I
never
detected
anything.
How?
When
should
I
redo
that?
On
the
other
hand,
if
you
consider
that
your
link
is
when
you
can
actually
talk,
discover
mr.
D
B
and
talk
to
him
and
that
the
you
can
uniqueness
of
your
link,
local
has
to
be
asserted
only
between
those
two
and
that's
true
right
right.
If,
if
he
does
not
know
anybody
in
called
Pascal
and
I,
don't
know
anybody
called
Robert
I
can
call
him
Robert
and
I.
Don't
make
a
mistake,
and
you
can
call
me
Pascal.
That's
enough.
As
long
as
I
have
a
single
Robert
I
around
me,
I
can
talk
to
Robert
without
any
any
problem,
with
a
latent
duplication
and
misunderstanding.
D
So
the
way
we
do
it
in
in
five
four
five
is
exactly
like
that.
We
say
you
don't
have
to
assert
the
uniqueness
of
your
IP
address
once
and
for
all
that
doesn't
work.
You
have
to
do
it
individually
with
every
people
you
talk
to
so
the
link
layer.
A
little
low
contrast
is
not
actually
looked
up
on
this
central
registrar
that
you
are
for
uniqueness.
It's
just
between
you
and
your
first
hop.
Router
never
goes
anywhere
else
and
you
don't
swat
them.
D
But
if
you
do
a
write
over
of
mesh,
you
don't
wrap
to
link
locals
right
because
it's
what
hub.
So
we
really
see
the
link
as
being
this
point-to-point
thing
that
actually
the
we
build
a
p2p
subnet.
If
we
like-
and
that's
that's
the
me
more
thing
that
we
can
do,
then
we
need
to
so
this
is
kind
of
a
physical
which
ability
thing
right
so
link.
D
If
you
look
at
a
to
two
hundred,
it
kinds
of
words,
what
is
9p
link
as
if
it's
a
medium
right,
a
physical
thing
here
we
already
alpha
way
to
the
logical
thing:
it's
even
reading,
802
Android,
it's
not
very
clear.
What
should
the
concept
of
a
link
be?
Should
it
be
my
broadcast
domain,
or
should
it
be
this
peer-to-peer
point-to-point
connection,
but
then
because
I've
done
and
things
like
that,
it
makes
a
lot
of
more
sense
to
consider
it's.
D
The
point-to-point
connection
now
should
I
live
with
only
point
two
point:
seven
nights,
like
some
people
want
to
do
well.
I
should
be
on
the
smart
grid
network,
with
thousands
of
nodes,
I
think
to
deploy
thousands
of
prefixes
for
just
two
sensors
talking
to
one
another
unreal,
a
that's
seems
a
bit
absurd.
You
could
do
it.
You
have
ample
room
in
v6
if
you
want
to
deploy
it
this
way.
Now
the
users
don't
want
to
deploy
it.
This
way
they
want
to
change.
They
just
expand.
D
Is
this
prefix
and
then
you
write
inside
it
using
ripple?
That's
how
we
deploy
those
things,
so
it's
still
useful
to
build
more
complex
network,
so
the
for
the
first
case
that
we
can
think
of
is
haven't
spoke.
This
is
the
case
of
an
MS,
your
roadside
unit,
and
you
have
two
cars
which
are
next
to
this
roadside
unit,
and
the
roadside
unit
can
actually
provide
connectivity
to
the
world.
D
So
what
you
see
in
this
model
is
that
the
roadside
unit
broadcast
domain
is
exactly
like
that
as
an
API
right,
you
can
say,
oh
I
can
broadcast
to
whether
my
hub
can
talk
to
and
then
I'll
get
the
equivalent
of
a
broadcast
and
and
and
make
a
subnet
of
it.
If
you,
if
you
say
not
on
link,
you
get
the
exact
same
service
after
you're
through
the
nightly
or
to
you
gold
every
times
through
the
hub,
and
then
you
go
to
your
destination.
D
So
the
simple
action
of
saying
not
on
link
gives
you
that
now
what's
more
interesting
is
the
prefixes
right.
You
can
be
on
the
same
prefix
as
this
guy,
because
you
never
talk
directly
to
him.
You
don't
need
to
be
in
his
broadcast
domain,
because
you
don't
do
that.
You
don't
do
Luka.
You
don't
do
that
because
eight
five,
four
five
registers,
your
heart
rise
to
the
hobby
and
the
hub
is
responsible
for
telling
this
address
is
duplicate
not
so
there
is
no
dad.
You
just
register
on
the
registration
succeeds.
D
Johan
the
registration
fails
because
there's
duplicate,
you
form
a
new,
tries
to
try
again,
but
you'll
never
have
to
do
that
again,
he's
in
charge
for
a
lifetime
enough
I'll,
look
up,
you
don't
look
at
you
talk
to
your
writer
and
it
will
clear
because
he
has
a
registration
on
the
other
side.
He
will
know
how
to
fold.
So
you
see
how
it's
five,
four
five
actually
solved:
the
prime
when
you
have
urban
spoken
activity
at
layer.
Two!
So
like
I
said
you
know,
CB,
you
don't
have
broadcast
emulation
at
Mac
layer.
H
D
So
let
me
see,
I
won't
cover
this,
so
the
last
model
that
we
see
a
lot
and
that's
how
we
build
our
mesh
of
smart
grid.
Things
is
this
mesh
here,
so
we
still
be
on
this
point-to-point
connection
and
then
we're
at
between
them
and
we
use
repo.
So
in
this
case
here
all
you
need
is
is
just
one
hopeful
warning,
so
you
don't
need
routing.
You
need
routing
as
soon
as
you've
got
to
hop
forward
right
and
that's
that's
when
we
actually
play
play
repo
and
without
five,
four
five.
D
What
you
do
then,
is
you
need
a
software
register.
We
call
a
6
lb
R,
which
comes
from
our
c
6725,
and
now
you
get
uniqueness
of
your
drives.
When
you've
got
uniqueness
of
your
address,
you
can
inject
it
in
the
writing
fabric
and
get
to
any
place
the
bottom
line.
Is
you
never
broadcast
right,
at
least
at
any
level?
Now
there
is.
There
is
a
last
situation.
D
Well,
actually,
your
multi
nning
subnet.
Is
it
erogenous,
meaning
that
you
have
a
different
quality
of
network
in
the
backbone
and
in
the
wireless
access?
So
in
this
case
is
just
mapping
again
a
Wi-Fi
network,
but
this
could
be
your
phone
mesh.
So
in
a
Wi-Fi
network
you
have
urban,
spoke,
the
API
or
EAP
and
all
the
stations
all
the
hosts
and-
and
you
have
this
backbone
and
the
backbone
router
here
is
doing
proxy
neighbor
discovery.
D
D
So
just
one
of
those
things
you
can
do
right
if
he's
sleeping
and
and
basically
the
backbone
water
could
reply
on
behalf
of
his
living
proxy.
If
there
is
like
a
dad,
so
you
can
do
all
those
sorts
of
operation
the
killer.
The
key
thing
is
on
the
backbone
it's
T
of
the
good
ol
nd,
so
you
do
broadcast
or
everything
it's
just
not
propagated
in
the
wireless
yeah.
H
D
From
this
group
is
just
more
like
trying
to
get
a
sense
that
we
are
done
with
what
we
could
do,
but
now
we
go
to
a
maintenance
mode.
If
we
go
to
maintenance
mode,
then
it's
probably
six-man
and
this
question
to
the
group
question
to
the
IDE.
But
it
seems
to
me
that
that
has
kind
of
illogical
transfer
when
the
technology
has
been
done
in
a
group
that
you
know
the
maintenance
is
not
as
redundant
a
group,
but
it's
six
men.
Unless
six
law
wants
to
keep
the
maintenance
of
it.
D
We
turned
America
see
here,
write
these
things.
We
turned
America
see
here,
but
then
do
we
do
we
want
to
give
you
know,
because
what
could
happen
to
this
is
get
generalize
to
more
interface.
That's
a
discussion
which
has
happened
on
six
men.
Do
we
generalize
that
or
more
interfaces?
It
makes
a
lot
of
sense
to
do
that,
but
then
it
would
not
be
six
level.
So
why
don't
we
say
a
it
goes
to
six
man,
they
maintain
it
and
they
generalize
it
to
more
interfaces.
That's
one
thing
on
the
table:
yeah.
E
Suresh
krisshnan,
so
not
speaking
as
a
tea
here,
but
it's
like
someone
who's
interested
in
this
topic,
so
this
bunch
of
us
who
did
like
similar
kind
of
work
in
six
men
before
so
eric
normal
Samhita
who's
sitting
right
behind
here.
So
what
are
the
things
that
has
like
not
been
addressed
like
very
well,
and
it's
very
hard
to
address
properly
is
the
backward
compatibility
with
the
stuff.
That's
there.
Currently,
okay,.
D
E
Something
for
so,
like
everybody
uses
this
as
the
default.
Nd
is
like
a
lot
of
work,
so
I
I'm,
not
you
could
think
no
right,
like
you
know,
that's
kind
of
what
we
thought
too.
So
like
I,
select
authors,
and
it
just
takes
a
lot
of
time
so
I
wouldn't
assume
that
this
will
happen
soon.
So
it
could
happen
right
like,
but
I
think
it's
gonna
take
its
time.
E
Like
you
know
this,
like
the
dhcpv6
discussion
right,
like
you
know,
you
had
this
discussion
right,
like
four
people
say:
oh
I
want
to
run
dhcpv6
like
a
random,
IAD,
Lance
right,
and
so
those
kind
of
discussions
will
happen
in
six-man
because,
like
people
have
different
views
and
that's
really
the
goal
of
six-man
right,
it
should
bring
all
this
perspectives
together.
So
here
you,
if
you
go
into
any
of
this
like
like
I,
would
call
it
like
constrain
node
cluster
working
groups,
like
you
know,
you'll,
find
mostly
support
here
right
but
like.
E
If
you
go
to
the
wider
community,
there's
people
are
like
different
operational
models
and
and
different
kind
of
constraints,
who
might
have
other
things
to
bring
up
and
any
I'm,
not
saying
it's
possible
or
not
to
do
I'm.
Just
saying
it's
like,
don't
think
it's
gonna
be
a
short-term
thing
like
this
is
gonna
get
done.
That's
what
I'm
running
it!
You.
D
D
This
interesting
discussion,
because
initially
it's
for
like
maintaining
the
ultras,
and
it
makes
a
lot
of
sense
to
continue
here,
but
when
you
have
to
put
them
to
different
things
or
when
you
have
to
do
new
functions
like
this
one,
in
particular,
that's
when
it
becomes
a
lot
less
clear.
So
that's
why
it's
interesting
to
see.
You
know
how
every
logic
logic
applies
to
whom
so
unique
eyes
look
up
is
is
just
new
right.
It's
just
a
personal
submission
at
this
moment
was
new
at
last
a
chair.
So
it's
like
three
four
months
old.
D
What's
it
about
it's
about
scaling,
even
your
backbone
infrastructure,
it's
about
enabling
overlay
structures
like
the
exelon
or
in
vo3,
whatever
you
play
as
an
overlay
I've
been
to
all
the
little
train
myself,
because
they
don't
have
enough
pieces
to
try
that
that
there
are
limitations
to
what
you
can
do
with
just
nd
on
the
backbone,
even
wired,
and
that's
due
to
the
broadcast
storm
that
nd
Wilcox,
because
each
time
you
have
mostly
if
those
devices
are
wireless,
so
they
come,
they
go.
They
join
a
different
ap.
You've
done
them
to
sleep.
D
You
start
them
over,
they
will
they
will
form
a
new
address.
All
this
creates
a
background
pressure
of
broadcast
on
your
Wi-Fi.
So
that's
one
thing,
but
it
also
creates
that
that
pressure
on
the
infrastructure
itself,
meaning
that
you
have
to
refresh
in
the
mapping
servers
and
all
those
things
you
have
to
snap
the
back
yet,
and
sometimes
we
do
it,
try
it
sometime,
we
fail
bottom
line.
D
Is
all
those
broadcast
hurt,
mostly
if
you
start
distributing
your
layer
to
fabric
overlay
or
three,
like
you
have
multiple
sites
with
the
same
layer
to
the
main
that
hurts
when
you
have
all
those
broadcasts
right
at
the
began.
You
build
this
thing
that
the
DES
Moine
hurts.
So
broadcasts
was
okay
on
a
single
yellow
wire.
It's
much
much
less,
okay
on
the
multi-site
layer
to
the
main
so
scaling
a
network
like
this
to
tens
of
thousands
of
node
is
something
that
we
want
to
build.
D
D
And
I
guess
modern
topologies
in
they
want.
Is
this
this
resolver
to
be
able
to
just
build
the
overlay
in
the
first
place?
The
way
we
do
that
today
is
we
snoop
the
protocols,
so
we
actually,
we
have
to
program
the
chica
intercept
the
packets
in
outerwear
and
de
packets.
Recognize
them
push
them
to
software.
We
call
that
pants,
so
we
bound
to
software.
Now
we
analyze
the
packet.
We
do
something
like
a
reverse
engineering
of
the
protocol.
D
Try
to
figure
out
who
is
where
based
on
slipping
those
packets,
so
we
Snooper
and
EU
is
not
DHCP
who
is
not
actually
data
packets
if
the
source
is
not
known
to
us,
so
we
can
actually
trigger
like
and
deep
I
guess
to
see
was
there.
So
we
do
all
the
sorts
of
games
because
we
don't
know
who's
where
and
we
need
to
know
who's
where
to
be
able
to
to
enable
our
overlay.
D
It's
high
time
that
there's
a
protocol
that
tells
us
precisely
where
every
IP
addresses,
so
we
can
have
an
overlay
that
works
for
all
our
customers.
That's
what
we
want
to
be
honest.
We
want
to
build
a
good
overlay
service
and
it
could
spend
over
multiple
domains,
multiple
overlay
or
three
right.
You
span
over
your
cloud
or
different
places
for
that.
We
need
a
registration
protocol
and
we
need
it
on
Ethernet.
We
don't
need
it
only
on
on
wireless
devices.
D
Not
only
do
we
need
a
registration,
but
we
need
to
centralize
all
this
information,
so
we
can
do
look
up
because
what
kills
us
today
is
all
those
lookups
which
are
broadcast
about
the
whole
fabric.
So
if
you
could
have,
if
you
can
have
a
central
registrar,
but
we
do
have,
but
that
can
be
queried
by
any
host
through
a
started
interface,
then
it's
a
lot
better.
A
host
today
will
not
totally
as
our
map
server
talks
Lisp
right.
D
D
For
that
you
need
to
locate
the
map
server
big
deal,
and
then
you
just
ask
him:
if
there
was
a
registration
for
that
address,
he
will
tell
you
if
there
was
none
in
a
mixed
environment,
you
will
still
need
to
do
Ana's
look
up,
but
at
least
you
tried
and
you
see
the
sort
of
game
you
get
is
the
more
devices
play
by
the
new
rule,
the
less
broadcast
that
will
be
for
everybody.
It's
a
game.
D
Theory
thing
the
more
players,
the
more
we
win
right,
because
the
more
lookups
will
be
answered
unicast
and
there
will
be
less
broadcast
for
everybody.
So
by
playing
the
game
of
registration,
you
actually
help
everybody
else,
not
really
yourself
good.
So
that's
what
the
diss
unicast
thing
is
all
about,
so
you
can
do
an
NS
directly
to
him,
but
if
you're
behind
the
wireless
interface,
for
instance-
and
you
already
do
dns
to
your
router,
then
the
router
will
do
it
on
your
behalf.
That's
the
model
of
6775
and
H
505.
D
D
Now
we
say
eh,
why
don't
we
share
this
thing
for
everybody,
including
the
way
out
guys
so
for
us,
in
mostly
about
displacing
a
function,
but
also
adding
this
little
API
to
do
look
up
because
fall
night
was
mostly
used
for
dad.
Now
we
do
address
resolution
and
you
see
the
goal
is
to
take
a
6
low
fabric.
This
is
6
of
ER
and
give
it
to
six
men.
D
So
now
the
question
I
asked
to
Suresh,
you
know
earlier
becomes
relevant
again
because
the
case
has
changed,
so
this
is
6
low
methods
but
apply
to
real
6
man
case.
So
where
should
it
happen
now?
I
have
my
answer.
So
this
is
this
is
now
for
without
this
this
can
interest.
This
is
the
first
use
case
again
the
Tokyo
Station
game
that
I
talked
about
earlier.
This
is
how
things
should
work.
If
you
do
it
this
way,
then
you
can
infer
it.
D
Your
traffic
almost
instantly
instantly
enabled,
as
you
go
in
the
station,
so
basically
what
the
wireless
guy
does
when
he
accesses
network
is,
is
DNA,
tries
to
real
to
to
locate
a
router
and
tries
to
first
unique
as
to
his
own
router.
Well,
that
won't
work,
so
it
will
send
a
multicast
to
all
routers
and
the
cool
thing
is.
D
It
is
a
multicast
but
remember
at
leav
its
first
a
unique
as
to
the
AP
right
and
because
we
have
a
layer,
3
AP,
it
will
actually
go
all
the
way
to
layer,
3
and
what
the
AP
will
say.
Oh
no,
no
don't
broadcast
it.
I
can
handle,
so
it
will
never
be
broadcasted
on
the
under
layer.
It
will
be
absorbed
by
the
lioska
AP,
which
will
actually
answer
by
a
unicast
arrive.
So
in
this
phase
here
you
don't
have
a
broadcast.
That's
the
first
thing
you
need
to
realize.
D
There
is
no
matter
your
broadcast
on
a
wireless.
Then
there
is
this
NSE
arrow,
which
is
the
registration
itself.
Arrow
is
the
address
registration
option
and
that's
when
the
6
backbone
router
here
will
form
the
nd
cache
before
there
is
any
packet
for
it.
So
now
the
MAC
address
of
the
station
here
is
known
if
you're
lucky
with
the
new
draft
you,
actually
you
get
your
registration
to
the
6
MBA,
which
tells
you
oh,
it's
unique.
D
Ok,
you've
passed
that
at
this
point,
that
is
done,
but
if
you're
in
a
mixed
environment,
where
there
are
some
notes
which
register
to
the
6
lb
on
some,
which
did
not,
which
is
your
question
area,
you
still
have
to
run
that
so
this
would.
This
will
mostly
respond
in
an
interesting
manner.
If
there
is
a
duplicate
but
is
if
there
is
none,
then
if
you
are
in
mixed
environment,
you
still
did
to
do
that.
So
it's
a
broadcaster,
but
you
don't
have
to
wait
for
the
answer.
D
No,
but
normally
you
should
wait
one
second,
but
we
have
a
dad
so
optimistic.
That
is
what
I
represented
here.
So
you
send
in
our
ass
with
no
SLA.
Oh
because
you
cannot
wait,
that's
the
RFC
and
if
the
router
doesn't
have
a
neighbour
cache
for
you,
then
it
will
try
to
respond
unicast
if
he
tries
to
respond
Unocal
that
mean
you
will
look
you
up
and
that
you
can
respond
by
your
dad.
Ok.
D
So
with
this
little
nslookup,
which
is
also
broadcast,
you
respond
with
adonai,
you
being
the
backbone
router,
the
proxy
right
and
and
now
the
default
router
as
a
state
for
you,
and
now
the
packets
can
flow
both
directions.
Once
again
let
the
items
are
to
chaos.
I
mean
the
traffic
I
started
for
a
long
time.
D
This
is
how
you
can
make
this
six
work
in
a
matter
of
split-second,
so
that's
pretty
much
what
we're
after
so
we
also
change
the
Dow
tag,
because
if
you
see
this
picture
in
this
case,
the
six
thousand
is
not
directly
connected
to
the
internet.
It's
a
real
nice
guy,
so
the
request
is
actually
done
by
the
sixth
via
on
the
elf
and
the
lookup
could
be
that
way
as
well.
So
we
extended
the
dyrdek
exchange,
which
is
normally
phone
for
that,
so
that
you
can
also
do
a
gap
with
it.
Otherwise,
it's
well.
D
D
F
D
J
Okay,
this
is
sort
of
an
experiment
on
my
my
part,
I'm
working
with
a
Zhang
Zhang
and
Lee
Gwang
bong,
at
least
that's
my
attempt
at
pronouncing
their
names
and
I
apologize
for
how
bad
it
was
on
this.
The
original
idea
came
from
Zhang,
actually
not
from
me,
and
it's
called
asymmetric
a
ipv6
because
it
has
to
be
called
something.
J
J
Additionally,
the
edge
router,
the
first
top
router
may
be
very
constrained,
and
the
thesis
behind
this
work
is
that
in
a
very
constrained
router,
the
traditional
compression
and
decompression
algorithms
use
resources
that
you
don't
want
to
use
and
that
sacrilege
128-bit
addresses
consume
memory.
You
know
they
consumer
me
in
ribs
and
films,
and
they
consume
memory
in
buffers
and
so
on.
So
the
approach
is
to
tackle
those
two
problems
simultaneously.
J
How
this
would
mix
and
play
with
other
solutions
is
definitely
for
for
further
study.
I
should
say
that
right
at
the
front,
otherwise
Pascal
will
stop
bouncing
up
and
down
I
said
suspect,
so
the
method
is
to
within
a
domain
there
has
to
be
a
finite
domain
within
which
supplies
define
an
address
length
or
call
it
N
and
all
addresses
inside
the
domain
are
assumed
to
have
a
common
prefix
of
128
minus
and
bits.
J
J
So
there's
going
to
be
some
slightly
strange
business
going
on
and
the
router
that
sits
between
the
domain
and
full-length
IP
unnecessary
head
about
simply
missed
out.
That's
so
a
fairly
obvious
thing
to
do,
and
to
do
all
that
the
idea
is
to
use
what
we
call
flexible
header
encoding.
So
the
headers
are
not
in
a
fixed
format.
They're
enos,
somewhat
flexible
format,
fast
Carles
begins
you
jump
up
and
down.
K
J
Within
the
domain,
the
first
thing
you
have
to
do
is
decide
what
the
address
length
is.
Everyone
has
to
agree
on
that.
Obviously,
that
could
be
configured
by
a
manufacturer.
It
could
be
configured
by
operators
could
be
configured
by
users
actually
or
it's
parameterised
in
some
way,
and
a
gateway
between
the
this
domain
and
the
rest
of
the
world
communicates
their
dress,
lengths
to
everybody
else,
or
it
could
even
be
negotiated,
so
that
this
is
still
completely
open
right.
J
This
is
getting
a
bit
specific
and
I
must
say,
is
cramping
in
the
room.
Hope
so
I
didn't
do
this
slide,
and
this
is
a
suggestion
that
to
make
this
work
we
would
have
to.
We
could
add
flag
to
the
standard,
ipv6
RA,
which
says
that
some,
if
you
receive
the
sorry,
you
can
use
symmetric
ipv6
mechanism
and
the
prefix
would
be
modified.
Prefix
information
on
to
tell
you
what
the
prefix
was
and
what
its
length
was.
J
J
J
So
why
not
use
those
bits
as
well
since
they're,
actually
no
use
in
an
ipv6
packet,
so
that
gives
you
that
gives
you
basically
12
bits
in
which
you
can
indicate
which
fields
are
actually
present
in
the
header.
So
this
example
there's
a
payload
length.
There's
a
next
header
standard,
ipv6
next
header,
there's
a
hop
limit
and
there's
a
truncated
destination
address
truncated
because
it
is
under
the
domain
prefix.
You
don't
need
to
include
the
domain,
prefix
bits
and
there's
no
source
address,
because
well,
people
know
your
source
address
anyway,
from
your
MAC
address.
Once.
J
J
J
L
J
E
Yeah
suresh
krisshnan,
so
no
excuse
for
not
knowing
chick
everybody
had
t-shirts
for
like
the
last
few
meetings.
No
heidi
hat,
I
think,
like
one
issue,
I
see
right
away
with
this.
Header
format
is
like
the
alignment
like
all
the
bit
fields
are
like
not
aligned
so
like
the
payload
lengths
like
the
16-bit
field,
with
no
alignment
same
with
the
hop
limit,
I
think
if
you're
putting
in
C's
yeah.
J
D
J
You
know.
So
if
something
arrives,
the
short
address
it
has
to
and
has
to
go
out,
he
has
to
expand
it
to
the
full
length
address
and
in
the
opposite
direction.
Coming
in
that
with
inside
the
domain,
you
could
communicate
entirely
using
the
shortened
addresses
and
in
some
cases,
with
no
source
address,
because
that's
implicit
anyway.
J
J
J
M
Hello,
come
home
from
Huawei
one
of
the
author
of
these
graphs.
The
main
requirement
from
very
simple
scenarios
like
supermarket
and
and
the
point-wise
are
very
simple
and
the
resource
limited,
so
they
will
test
the
one
to
talk
to
talk
with
each
other
and
only
maybe
only
with
crises,
source
and
destination.
M
So
here
to
talk
to
will
so
the
problem
for
how
to
address
and
the
points
and
there's
no
other
functions.
So
a
bv6
may
be
so
heavy.
For
this
scenario
and
I
also
know
there
are
many
perfect
compression
approaches
like
a
stick
and
six
oppa,
but
but
which
may
need
to
compress
compress
and
the
comprise
when,
when
community,
when
communicated
with
each
other
here,
we
just
wanted
to
mostly
reuse
at
v6
Sikkema
to
so
common
problems.
Thank
you.
D
D
First
I
mean
this
is
the
right
group,
because
that's
kind
of
the
prime
of
the
nucleoid
fall
temperatures
right,
so
you're
in
the
right
place,
and
obviously
we
did
a
lot
and
thought
in
many
different
directions
now
and
when
I
just
look
at
what
you
you
proposed
first,
there
is
this
argument
that
compression
is
difficult.
Well,
the
six
top
and
compression
difficult
is
a
hard
world,
but
I
think
saying
that
what
you
do
is
not
a
compression.
It's
actually
fallacy,
because
you're
doing
pretty
much
the
same
thing
as
we
are
doing.
Yeah.
F
D
So
saying
I'm
not
doing
compression
you're
doing
compression,
so
my
stuff
is
good.
Your
stuff
is
not
what
you're
doing
same
thing
as
we're
doing
now.
The
other
argument,
which
is
very
interesting,
is
can
I
process
the
packet
in
that
format,
like
a
that
rocket,
I
wrote
in
that
formatting
can
I
receive
the
packet
in
that
format
and
well
it's
true.
D
That's
that's
really
important,
and
if
you
look
at
it
once
create,
when
you
have
time
or
in
stats
sake,
you
realize
that
it's
completely
designed
to
be
forward
in
the
compressed
format
to
be
operated
in
the
compressed
format.
You
never
want
to
look
at
the
original
format,
but
its
equivalent,
and
just
like
it
is
saying
exactly
I
mean
we
shall
so
much
the
goals
I
mean
they
did.
We
did
not
want
to
have
to
transmit
address
in
the
for
want
to
transmit
them
in
a
compressed
version.
D
Now
we
did
not
want
to
enforce
the
same
compression
for
all
addresses
in
the
domain
like
for
now
you're
doing
we
can
actually
decide
address
poetry
so
much.
We
compress
it
and
we
can
actually
place
your
source
for
today,
which
is
very
in
low
power
devices
right.
Sometimes,
you
can't
have
state
or
out
of
our
multiple
hops,
so
we
also
thought
about.
Oh,
how
can
I
write
is
this
packet,
as
you
want
to
write
it
such
a
way
that
is
very
easy
to
afford
and,
like
you
said,
the
source
is
not
important.
D
What's
important
like
the
destination,
but
if
you
do
so,
swatting
is
not
just
the
first
of
destinations,
just
all
the
other
destinations,
and
then
you
start
singing
a
do.
I
like
the
SWOT
that
they
have
to
do
between
the
rotting
adder
and
the
IPO.
Now,
let's,
let's
remove
that
anyway,
I,
don't
care
about
keeping
the
old
chain,
so
why
don't
I
put
it
in
the
beginning
and
cut
stuff
so
realize
that
the
kind
of
thing
that
started
this
draft?
D
This
is
the
kind
of
thinking
we've
had
for
like
ten
years
and
and
we
weren't
very,
very
fine
that
direction
actually
so
so
I
really
want
to
work
with
you
on
understanding.
You
know.
Is
there
a
gap
that
we
have
not
covered
yet
so
far?
I've
not
seen
that
gap
in
your
presentation,
I
think
we're
covered
already
to
do.
Watch
what
you
want
to
achieve.
In
particular,
you
don't
have
to
expand
anything.
You
can
work
with
the
compressed
formats.
She
could
understand
the
conference
format.
Whyever
work
with
the
uncompressed
format.
J
D
I
said
I
think
it's
a
fallacy
to
Kali
to
that
that
patient
layer
you
care
I
mean,
but
it
is,
but
but
it
does
compression
the
compression.
That's
what
it
does.
You've
shown
me
to
format
when
it's
smaller
than
the
other
I
call
that
a
compression
right,
saying,
I
avoid
the
process,
and
now
you
can
just
say:
oh,
my
compression
is
simpler
than
yours.
Well,
we
can
discuss
a
beauty
contest
if
you
like,
and
we
can
improve
6lowpan
if
you
promise
that
we
can
do
better,
but
we
are
in
the
very
same
family.
D
We
have
very
same
goals
and
we've
pretty
much
achieve
what
you
claim
you're
after
and
for
some
things,
I
think
we
did
it
better
like
like
the
size
of
the
compress
thing
we
can.
We
are
much
more
elastic
now
for
other
things.
Maybe
you
have
a
better
portion.
We
can
improve
what
with
it
but
you're,
not
starting
from
scratch.
We've
had
a
lot
of
experience
on
that.
J
A
Carlos
Gomez
thank
you
for
bringing
this
work
and
for
the
presentation
as
a
participant
and
I'm
curious,
I,
don't
know
if
maybe
you
have
any
sort
of
evaluation,
some
kind
of
quantification
of
the
effect
of
minimizing
or
avoiding
the
compression,
because
I
understand
that
avoiding
the
compression
is
one
of
the
main
points
in
your
proposal.
So
I
don't
know
if
you
have
any
sort
of
evolution.
The.
N
Yeah,
actually,
the
decision
actually
is
a
motivation.
We
had
use
what
we
saw
in
6lowpan
versus
you
know
compression
and
decompression,
and
it
has
to
do
every
hope
when
it
goes
to
layers
way
without
him.
But
what
we
want
to
is
we
had
this
address,
be
able
to
build
up
the
routing
table
within
the
domain
without
any
camera
temperature
and
decompressing.
So
in
that
way,
actually
you
know
in
ways
in
the
domain
there's
only
necessary
address
as
limited
as
possible.
Yes,
so
that's
for
every
note
and
for
every
package.
D
You
can
be
at
the
writing
table
in
such
a
way
that
if
you
know,
for
instance,
that
you're
in
a
repo
to
your
deck,
just
to
give
you
an
example
because
we
built
for
that-
and-
and
you
know
you
know-
what's
the
the
size
of
the
bits
which
always
come
and
then
actually
we
have
a
context
that
we
distribute
through
our
eyes.
That
tells
us
what
this
context
is,
and
we
can
actually
remove
that
from
everything
and
now
and
in
all
the
packets.
D
You
see
only
the
rest
and
now
having
a
routing
table
which
sometimes
I'll
need
at
rest.
It's
completely
an
implementation,
it's
not
visible
from
the
outside,
so
you
can
really
build
it.
We
never
said
don't
do
that!
It's
just
copy
to
your
back
to
to
the
outside.
We've
processed
the
packet.
We
can
process
the
packet
in
a
compressed
format
and
we
don't
mostly
relate
for
eight
one.
D
Three,
eight
right,
we
just
snip
the
first
of
the
destination
addresses
which
is
just
like
a
rotten
either
just
need
the
first
and
fold,
it's
very
simple
and
if
your
table
watching
table
contains
eight
bytes
power
twice
because
you
know
the
first
date
a
command.
That's
that's
fine!
It's
your
implementation!
So
I
can
we
do
that
already
now
we
could
describe.
You
know
that
operation
inside
the
box,
the
format's
we
have
already
enabled.
O
N
Husker
I
could
agree
on.
You
know
what
to
claim
the
6lowpan
actually
could
do
that,
but
actually
that's
a
problem
because
the
6lowpan
actually
described
a
framework
could
have
many
different
subgroup
of
it,
but
to
do
that,
each
of
them
the
actor,
a
basic
slope,
a
framework
that
become
complicated.
So
you
can
say
what
we're
doing
is
a
simplified.
Brushing
of
you
know:
6lowpan,
that's
fine,
I,
don't
care
what
I
care
is.
You
know
those
constraint
note
could
do
it
in
the
simplest
way
and
save
the
energy
for
them.
D
To
be
proven,
right,
I
mean
think
for
a
second
about
the
dalvik
work
that
we
did,
that
custom
did
for
fragments,
it
was
enabled
by
6lowpan,
but
nobody
wrote
that
you
could
have
this
behavior,
so
those
documents
and
fragmentation
fragment,
forwarding
and
and
they're
weak
draft.
Where
are
useful,
they
will
be
proceeded,
but
they
did
not
need
to
change
the
format.
So
what
I'm
saying
is,
if
you
could
have
a
simplest,
behavior
or
a
subset
of
the
behavior
by
documenting
that
behavior?
D
That's
auto
gon,
also
providing
new
formats
but
I'm,
claiming
so
far
I'm,
not
saying
that
we
have
documented
the
behavior
that
you
talked
about
and
I
did
say
it's
not
useful
to
document
it.
I
said
you
don't
necessarily
want
to
change
the
formats
to
get
it
because
the
format's
we
have
allow
it
so
far
for
what
I've
seen
right.
It
may
be
in
the
details
which
show
me
that
we
miss
something
and
then
you
can
elaborate
on
the
format's.
You
have
look
at
81
38
as
a
starting
point
about
the
format's.
We
have
yeah.
J
H
J
M
How
would
the
last
comments
here
we
want
to
argue
which
is
vaster
formatting
here
inside
we
want
to
say
the
compression,
maybe
it's
beautiful,
unanswered,
84
and
also
there
for
compression
context.
Is
it
very
important
here
we
think
if
we
can
standardize
context
in
the
in
the
drafts
and
then
on
running
on
running
environment?
We
don't
need
that
anymore.
So
we
assume
we
can
config
lanes
of
the
road
twice
when
deploy
when
deploy
an
employer
devices.
M
P
H
P
This
is
a
deadline.
Draft
is
about
a
new
routing
header,
pipe
that
carries
information
about
to
enable
better
delivery
of
packets
that
have
the
delivery
constraint
on
delivery
constraints.
So
that
means
that
the
devices
that
received
a
packet
with
this
header
and
make
the
right
decision
about
scheduling
their
delivery
to
the
next
next
top
requirement,
and
the
assumption
is
that
the
network
devices
are
time
synchronized
and
there's
a
little
bit
of
specification
to
don't
how
to
handle
different
time
zones
this.
P
E
Kelli-
and
this
is
Sarah
sure,
hey
yeah
yeah,
so
one
of
the
things
I
haven't
seen
as
a
response
to
Alice's
discuss
at
all,
so
I
saw
you
responded
to
Roman
and
Magnus,
but
I
haven't
seen
a
response
at
all,
so
I
was
kind
of
like
trying
to
match
up
to
the
responses,
but
I
didn't
see
any
so.
My
if
you
made
changes
like
I
think
it'll
be
good,
like
this
put
the
changes
into
a
response
to
her
discuss.
P
E
E
P
So,
as
mentioned
I,
there
was
a
new
newer
material
in
the
security
consideration
so
about
the
expected
deployment
scenarios.
Well,
actually,
don't
that's.
That
was
a
separate
request
who
for
us
to
describe
how,
in
our
expectation,
the
document
would
be
implemented
and
deployed
for
what
purposes.
So
so
there
was
some
discussion
about
that,
and
now,
with
a
lot
of
Corrections
made,
it's
amazing
how
many
little
places
are
to
make
mistakes
and
also
it.
It
was
a
pretty
good
amount
of
material
added.
P
Several
paragraphs
to
describe
really
how
to
go
about
using
shorter
time
ranges
so
that
you
don't
always
have
to
use
the
maximum
number
of
bits
to
specify
what
the
deadline
time
is,
or
the
origination
time
Delta
and
so
I
think
that,
hopefully,
this
revision
of
the
slides
that
has
a
page
about
that
which
I'll
describe
a
little
bit
more.
But
several
people
notice
that
there
was
basically
some
arithmetic
errors
and
all
those
things
were
corrected
and
as
well
as
for.
P
Clarifications
and
and
believe
it
or
not,
there
was
one
place
where
the
order
of
the
description
of
the
fields
didn't
match
the
order
of
the
fields
as
they
were
laid
out
in
the
routing
header.
So
so
all
those
things
were
done
and
I
think
that
we
have
responded
to
all
the
comments
and
discusses
and
suggestions
for
improvement
on
next
slide.
Please
so.
P
So
this
basically
is
the
same
as
it's
been,
except
for
slight
Corrections.
So,
for
example,
if
you
look
under
the
deadline
time
value,
the
minimum
amount
of
deadline,
time
is
shown.
You
know
the
minimum
can
I
say
yo.
If
you're
going
to
send
a
packet,
you
have
a
deadline
time,
so
the
minimum
DTL
the
deadline
time
length
is
zero,
which
means
that
you
can
have
a
one
hex
digit,
and
so
that.
P
So
that
shows
that
the
and
then
a
maximum
number
then
is
1650.
So
the
deadline
time
cannot
be
specified
in
a
small
number
of
bits
or
I
can
use
the
whole
64
bits.
The
origination
time
Delta
is
to
be
subtracted
from
the
deadline
time
and
that
can
be
anywhere
from
zero
to
twenty
eight
bits,
because
it's
it's
only
a
three
bit
field,
then
you
can
only
go
up
to
seven
and
seven
times.
Four
bits
is
28
bits
next
flag.
Next
slide.
Please.
R
P
P
I'm,
sorry
didn't
quite
get
them,
but
anyway
there
was
one
more
slide
in
the
corrected
presentation
set.
That
shows
how
wraparound
works
and
I
guess
I'm
not
going
to
have
a
chance
to
present
that
at
last
slide
here,
but
if
I
can
describe
it
verbally
what
it.
What
the
effect
of
it
is
is
to.
If
you
look
at
your
synchronized
timeline,
as
maybe
you
can
say,
it
starts
from
0
to
and
goes
to
infinity,
or
at
least
what
the
maximum
time
is.