►
From YouTube: IETF98-PAYLOAD-20170328-1030
Description
PAYLOAD meeting session at IETF98
2017/03/28 1030
A
This
is
the
payload
working
session
and,
for
some
reason,
I,
don't
see
more
in
the
room,
which
would
be
because
a
problem,
if
you
won't
come
back
because
here's
the
presentation-
okay,
so
we
start
with
the
this
okay.
So
this
is
the
payload
session.
The
chairs
are
myself
an
army
at
least
left
here,
I'm
not
sure,
but.
A
A
A
B
So
a
minus
rested
and
yeah.
So
yes,
before
just
last
week,
event
finally
got
to
go
through
the
number
of
comments.
I
get
the
North's
editor
on
all
questions
about
a
spec
and
and
sent
my
kind
of
proposal
for
fixes
for
certain
things.
I
mean
if
see.
This
document
has
been
at
11
hundred
and
sixty
some
days
since
it
was
approved
so
it
has
aged.
B
Yeah,
so
I
I
think
you
want
to
inform
a
working
group
of
what's
being
changed
here
because
there's
the
are
changes,
so
it's
that
does
impact
a
Content,
some
degree
etc.
There's
several
edits
to
it
in
regards
to
note
about,
for
example,
at
the
new
upcoming
RC
format,
etc,
and
a
few
hundred
things
which
have
there
are
things
that
has
changed
so
I
want
to
get
it
back
and
see
what
the
changes
are
really
when
our
settlers
also
has
come
through
it,
but
it
might
have
to
you
at
least
show
it
to
working
group.
So.
A
B
B
A
It
was
very
helpful
anyhow,
I,
didn't
know,
stages
to
everyone,
who's
writing,
payload,
because
we
could,
as
it
is
reassuring,
we
got
requests
about
how
to
do
that.
I
would
always
point
to
this
document
whatever,
because
it
had
the
structure
of
how
to
write
a
tale
of
a
veteran
support.
It
doesn't
matter,
it
was
an
RFC
or
it
was
a
working
group
document
which
was
already
as
the
right
content
in
which
how
to
derive
a
specification
and
payload
yeah,
and
thank
you
for
this
work,
because
it's
important
thing
for
this
work.
A
A
Would
that
work
for
them?
Those
changes
right.
It's
like
a
typical
payment
where
we
get
the
payload
from
that
done
that
we
are
being
used.
Others
bodies
and
we
just
need
to
go
to
the
process
and
see
that
the
tailor
that
they
awaited.
The
document
is
structure.
Then
the
content
makes
sense
and
it
works.
Ok,
so
expect
to
see
a
working-class
goal
as
we
get
a
new
revision
for
that
one
other
burn
status,
the
vp9
RTP
payload,
Jareth
I,
mentioned
that
it's
leads.
C
Texts
basically,
the
explanatory
text
about
sort
of
getting
the
back
on
vp9
overview
of
its
high
level
structure
and
whatnot,
which
nobody's
written
yet
and
I
thought
I,
had
a
volunteer
to
help
me
write
that,
but
yes
in
either,
so
we
will
try
to
get
that
than
the
other
thing.
Mrs.
a
bus,
much
more
significant
tactical
change
is
that
I
was
talking
to
somebody
the
other
day
who
expressed
some
interest
in
using
he
mentioned.
Oh
yeah
I'm.
He
wants
to
do
you
know
various
multicast
video
he's
interested
in
vp9
and
I
thought.
Oh
no.
C
That
means
I
have
to
support
mr
mt
mode
and
I
don't
want
any
I
didn't
want
to
support
mrng
mt
mode.
So
the
question
is:
do
we
right
now?
We
only
support
srst
for
the
scale
bowl
is
their
interest
in
mr
MP
or
mrs
t
modes
for
this
I
guess
mrs
t
know
mrs
to
you
because
well
I
guess
the
question
would
be.
I
know
that
I
think
I
had
some
recollection.
C
There
are
some
perk
architectures
that
one
at
mrs
t,
but
I'm
not
sure,
but
I
guess
you
don't
really
need
that
with
with
frame
marking
but
ya
mar
so
I
guess
the
question
is:
do
we
want
to
delay
this?
Do
the
work
for
mr
mt
because
I
I
don't
want
it.
So
maybe
the
answer
is
if,
in
the
future
somebody
wants
at
mr
mt,
we
can
define
it
as
a
bit,
sir,
and
the
alternative
payload
format.
So.
C
Basically,
I'll
host
what
he
wants
to
do
I
mean
he's
not
even
sure
he
wants
to
use
our
GP
for
what
he's
doing
yet.
So
it's
a
very
early
requirement
basically,
but
he
wants
to
do
very,
do
actual
multicast
streaming
where
you,
you
know
where
what
you
actually
want
to
turn
into
tuning
to
the
multicast
group
corresponding
to
the
layers
you
want,
so
that
would
require
mr
mt.
That's
the
original
motivation
for
all
the
MRRT
modes
is
multicast
yeah.
C
C
C
A
C
C
A
So
we're
working
for
the
toilet
in
order
to
go
to
last
hope.
Okay,
the
other
documents
that
we
have
in
the
workgroup
is
the
vc
to
high
quality,
and
it
is
a
new
Glascow.
Probably
we
don't
have
enough
in
here,
so
what
I
will
try
to
see
how
to
finish
this
one
currently,
there's
no
other
documents
there.
So
the
milestone,
you
have
a
milestone
that
already
expiring
now
about
the
North,
be
flexible.
A
Fact
there
for
a
flexible
back
and
probably
need
to
update
and
see
how
to
attract
what
is
the
new
milestone
for
the
vp9
will
also
need
to
update,
because
we
are
working
for
revision
and
they
hope
to
be
able
to
send
a
VC
to
high
quality
for
publication
by
lake.
So
these
are
Bionic
monsters
for
this
working
okay.
D
Yeah,
that's
how
a
girl
from
india
if
we
have
time,
I
would
like
to
talk
a
little
bit,
but
a
proposal
for
external
effects
frame,
which
is
in
fact
closely
related
to
the
presentation,
will
act
to
the
morning
so
using
different
types
of
HEC
cards
in
order
to
protect
well
time
flows.
So
if
we
are
five
minutes
for
that
at
the
end
of
the
session,
then
I
would
be
glad
to
talk
a
little
bit
parties.
D
A
D
A
E
E
C
E
Okay,
so
an
earlier
version,
we
extended
the
flux
back
draft
to
support
protecting
multiple
sources
streams.
So,
with
the
single
effect
stream,
you
can
protect
multiple
SSR
seas
and
when
we
did
that
it
expanded
the
payload
format.
Quite
a
bit,
and
so
we've
been.
We've
worked
at
reduced
that
in
this
version,
and
so
we've
actually
moved
the
those
vests
RC
count
and
the
list
of
us
are
seized
and
is
something
that
I
TP
is
advert
forever
and
it
reduces
the
overhead
by
by
a
32-bit
word.
E
Next
next
change
was
a
suggestion
to
remove
the
final
extension
bit
mccabe.
It
extends
the
mask
length,
so
you
can
start
off
with
a
16
bit
mask
and
then
you
can
extend
it
to
a
32
bit
mask.
Then
you
extend
it
to
a
64
bit
mask
and
the
that
cabe.
It
was
extending
each
one
of
those
masks
and
there
was
a
suggestion
to
remove
it
from
the
last
final
largest
mask,
and
so
we
saw
him
a
reason
to
resist
that.
E
It
precludes
having
larger
masks
in
the
future,
but
I
think
we're
fine
with
having
110
packets.
So
so
we
did
that
so
now
now
that
limit
is
110
packets,
one
flex
effect
packet
can
cover
up
to
110
source
packets
from
that
one
SRC,
and
then
what
was
just
discussed
in
the
previous
ABT
session
there
there's
some
complexities
around
when
you
use
srtp,
in
fact
together
which
one
you
do
first,
and
so
we
added
some
text
that
specifies
the
the
order
and
let
things
like
perc
work.
E
Alright
next
slide,
please
so
the
SSRC
count
and
list.
This
is
the
old
format
that
was
a
that
was
used
in
version
3.
You
see
there,
there
was
a
soc
count
field
in
the
feck
header
and
then
a
large
wasted,
reserved
area,
and
then
you
give
each
SSRC
sfc
underscore.
I
give
each
one
of
those
associes
before
the
rest
of
the
data
that
goes
with
that
SRC,
so
the
sequence
number
base
offset
and
the
mask
along
with
this
RC
those
would
be
repeated
as
blocks
for
each
SRC.
E
So
in
04
we
move
to
actually
use
the
RTP
header
that
we've
had
all
along
and
that's
our
see
list
and
count
are
naturally
things
that
are
already
handled
by
the
RTP
header
by
the
CSRC
count
and
see
SRC
list.
So
we
just
moved
those
directly
into
the
RTP
header.
So
basically,
you
can
think
of
these
feck.
This
spec
packet
has
contributing
sources
and
here's
this
our
seas
of
those
contributing
sources
and
here's
the
count
of
them.
So
it's
clearly
not
a
mixing
operation.
E
I
guess
you
could
consider
you
know
the
computation
of
the
fête
code
is
kind
of
a
mix
of
the,
but
logically
it's
these
are
contributing
sources.
The
source
packets
are
contributing
sources
to
this
fact
packet
and
so
we're
just
using
the
arts
p
header
as
it
was
intended
to
be,
instead
of
creating
new
effect
headers.
For
that.
E
Yes,
yes,
so
if
you
look
at
in
the
blue
up
top
that
start
to
be
header,
which
actually
was
never
documented
in
the
old
version
of
the
specs,
so
now
we
actually
have
to
be
show
the
arts
theater
and
then
the
fact
header,
which
is
what's
documented
all
over
in
the
spec.
Some
people
actually
confuse
that
to
be
the
RTP
header
and
it's
not.
E
This
is
the
feck
header,
that's
the
payload,
the
RTP
payload
after
arts
be
header,
and
that
is
where
you
do
the
recovery
operations,
so
the
source,
packets,
CC
count
and
csrc
count
and
CSRC
list
and
extensions
and
padding
and
all
those
are
recovered
by
that.
So
it's
part
of
the
feck
header
below
the
arts,
apierror.
E
So
so
to
cook
will
clarify
that
the
art
spider
and
blew
up
top
the
CC
there
is
used
before
it
used
to
just
be
0,
and
so
now
we're
actually
using
it
to
indicate
how
many
srcs
are
we
protecting
and
each
one
of
those
estar
sees
that
you're
protecting
would
go
into
the
CSRC
list.
Just
like
regular
are
ITP
I
know,
there's
a
corner
case
of
whether
or
not
we
want
to
allow
c
c
equals
0,
whether
or
not
we
want
to
allow
not
specifying
what
SRC
you're
protecting
and
in
you'll
see
a
later
slide.
E
E
I'll
get
that
I'll
get
to
that
on
and
actually
we
toyed
with
toyed
with
whether
or
not
you
know
some
some
some
fact
formats
don't
have
the
explicit
SRC
mapping,
and
so,
if
you
can
derive
it
through
other
means,
you
know
through
external
signaling
or
or
through
you
know
simple.
You
know
if
it's
a
simple
environment
where
there's
only
one
stream
that
it's
kind
of
unnecessary
to
signal
it.
So
we
haven't.
We
haven't
documented
that
yet
about
whether
or
not
that
should
be
allowed.
E
So
if
people
have
opinions
on
that,
but
there
is
one
case
where,
in
the
retransmission
format,
we
do
want
to
allow
c
c
equals
0
all
right
next
slide,
and
this
is
this.
Is
that
reason?
So
in
the
retransmission
format,
the
the
one
of
the
intense
of
the
retransmission
format
was
to
just
basically
send
the
entire
RTP
packet
again
as
this
payload,
including
its
header,
and
so
the
in
n,
0,
3
and
0
4.
E
There's
there's
not
a
change
to
the
retransmission
format,
but
I
wanted
to
highlight
here
that
this
is
different
from
the
regular
effect
format.
When
you
have
CSRC
lists,
because
here
we
have
no
CSRC
list
c
c
equals
0
and
the
SSRC
that
you're
protecting
is
in
the
is
in
the
payload,
because
you're
just
basically
taking
the
entire
arts,
be
packet,
that's
being
retransmitted
and
sticking
it
in
as
payload.
So
you
have
a
full
12
line.
Rtp
header,
followed
by
csrc
lists
and
extensions
and
and
RTP
payload
from
the
original
packet.
E
So
this
could
this
could
have
been
done
either
way.
We
could
have
required
the
CC
equals
1
here
and
had
it
in
the
csrc
list
and
omit
the
s
original
SRC
from
the
packet,
but
it
seemed
like
unnecessary
manipulation
of
the
of
the
original
packet.
When
you
could
just
blindly,
you
know
copy
the
entire
packet,
so
this
seemed
better
and
actually
the
Jonathan's
earlier
point
about.
Do
you
want
to
disallow
c
c
c
equals
0?
E
Please,
if
you
see
the
the
first
two
bits
of
the
feck
header
r,
the
r
&
F
bits,
the
ARB,
it
is
the
retransmitted
it
and
the
F
bit
is
whether
the
it's
a
fixed
mask,
whether
the
mask
is,
you
know,
sequence,
numbers
or
whether
it's
a
lengths
number
of
columns
a
number
of
rows.
So
we
could
actually
remove
that
retransmission
bit.
If
we
wanted
to.
E
B
Minus
banawa,
seeing
how
it
looks
like
now,
I
wonder
about
why
not
use
different
payload
types
for
the
retransmission
if
FEC,
because
now
you
really
have
two
different
formats
and
you
have
a
feed
which
is
in
tension
about
this
and
that's
the
PD
fields
and
yeah.
Now
we'll
come
back
to
that
question
again,
we
have
kind
of
come
full.
E
C
Yeah
John,
panics
and
I
guess
that
having
to
be
the
same
format
does
is
basically
say
you
know
you
can't
pick
and
choose.
You
must
implement
both.
So
if
that's
the
thing
we
want
to
say,
then
obviously
you
could
dimming.
You
could
say
that
even
if
they're
different
pale
of
types,
that's
how
much
hard
to
enforce
rather
than
just
saying,
okay.
Well
this
this,
this
payload
type
means
you
support.
Either
format
right
on
receive
from.
E
A
purist
view
I
would
agree
with
Magnus
that
you
know
that
there's
logically
there's
not
a
really
good
reason
to
buy
them
from
a
realist
point
of
view
from
real
world
implementations
point
of
view,
and
just
you
know,
you
know
a
time
you
know
you
know
time
to
get
this
out
point
of
view.
E
It's
still
there
so
the
question
is
the
Lumia
me
restate
Jonathan's
original
question
that
he
asked:
should
we
forbid
c
c
equals
0
forbid,
having
no,
no,
no
estar
sees
signaled
in
the
effect
itself,
if
there's
a
case
where
we
want
for
regular
effect
packets,
not
for
retransmission
packets,
but
for
regular
effect
packets.
If
there's
a
case
where
we
want
to
allow
someone
to
specify
through
some
other
means,
what
the
what
the
sources
are
see
is,
then
that
would
be
a
reason
to
allow
c
c
equals
0.
E
E
C
G
E
E
Okay,
no
next
one
and
so
give
that's
the
retransmission
format.
Okay,
so
the
other
change
which
is
I
think
pretty
minor.
We
just
removed
the
last
k
bit
so
in
the
biggest
mask
you
can
have
that
64
bit
field
of
mask
46
2109,
who
would
just
removed
that
cake
k
bit
to
not
allow
any
further
extension.
So
now
you
could
protect
up
210
packets
instead
109
all
right
next
slide.
E
So
you
can't
even
the
middle
box
can't
even
tell
what
source
packets
are
protected
by
this
fact
packet.
So
it
would
have
to
blindly
forward
all
affect
all
the
time
and
would
have
no
idea
what
what
that
fact
packet
is
even
protecting,
which
is
clearly
not
not
optimal.
Middle
box
would
not
be
able
to
optimally
tune,
affect
rates
and
characteristics
for
each
each
participant
would
have
to
do
worst
case
spec
for
the
entire
conference.
So
if
one
person
has
a
bad
uplink,
everyone's
a
downlink
would
get
penalized.
E
So
clearly,
that's
not
that's
not
an
optimal
situation.
We
think
that
architectures,
like
Burke,
would
benefit
from
having
middleboxes
to
be
able
to
look
at
the
fact
and
generate
their
own
fact.
So
the
proposal
is
that
for
for
cases
where
you
have
things
like
our
pitchers
like
Burke,
we
would
require
that
the
Flex
fact
order
is
effect
after
srtp,
so
that
means
middleboxes
can
receive
defect,
packets,
understand
what's
in
the
feck
header
and
and
generate
their
own
effect,
packets,
differentially
towards
different
receivers.
E
E
I
B
So
this
is
minus
M
I,
don't
think
you
can
mandate,
ire
of
them
may
be
either.
You
know
it
so
both
can
occur
and
depends
on
the
scenario
we're
using
it,
and
therefore
both
needs
to
be
loud
and
actually
order
of
equity.
Sister
piece
is
one
of
the
parameters
is
intended
to
be
negotiated
by
the
key
management
system
or
for
configuring.
A
serta
pisa.
G
E
B
And
then
it's
it's
not
I
mean
the
case
where
it's
need
might
be
the
other
ways.
I
guess
I
mean
perk.
Is
the
argument
for
having
it
anyway,
but
I
assume
the
fake
fix.
It's
not
unlikely,
be
used
in
other
context.
Also
there,
where
you
have
only
point
to
point
and
but
you
don't
need
it
so
and
there
you
should
protect
the
the
regency
ed
effect,
packets.
Also,
the.
E
So,
just
to
clarify
everything
would
still
be
integrity,
protected
and
encrypted
in
any
environment,
whether
it's,
whether
it's
/
corn
on
park,
you
still
get
integrity,
protection
and
and
in
encryption
of
anything
that
you
wanted
for
the
media
payload
itself
for
the
effect
packets
you
can
think
of
the
pic
packets
is
just
being
almost
like
a
line
code
for
that
one.
For
that
link,
all
you're
recovering
is
the
is
whatever
was
transmitted
on
that
link.
You
don't
recover
anything
about
the
about
whether
that
packet
was
authentic.
E
He
has
stuff
to
do
the
SB
authentication
of
that
packet
after
you
do
the
thick
recovery.
So
the
worst
thing
you
can
do
it
be
yeah,
an
attack
that
that
that
forces
someone
to
do
extra
extra
federal
computation,
but
you
can
do
that
anyway.
So
basically,
a
Doss
attack
using
fact,
but
you
can,
you
can
do
it
doesn't
take
any
with
with
even
real
meteorite
or
with
real
package.
You
can
do
a
an
authentication
attack
with
adults.
E
Attack
right
I,
send
you
a
bunch
of
us
RTP
packets
and
because
the
order
is
is
is
SFTP.
In
effect,
you
have
to
try
to
do
srtp,
gmac
or
something
or
you
know,
h,
mac
to
decrypt
it
in
it's
a
to
figure
out
if
it's
authenticated,
so
that
operation
can
also
be.
You
know
pretty
time,
because
we
actually
it's
more
time
consuming
than
X
or
so
that
would
even
be
a
harder
attack.
Victor.
So.
I
E
No,
no
we're
not
updating
kirsten
was
this
is
this
is
a
suspenseful?
What
the
default
order
is
for
this
payload
format.
3711
says
that
other
specs
can,
you
know
you
know,
can
set
their
own
signal
or
specify
their
own,
and
so
here
we're
specifying
the
default
and
I
think
Magnus
is
arguing
that
we
should
also
maybe
figure
out
where
it
should
be
signaled
in.
I
I
You
need
to
say
that
you
wore
order.
Oh
yeah,
okay,
so
the
other
question
was
do.
Should
we
actually
go
through
three
7-eleven
or
something
that'll
like
that
and
go
and
see
like
what
attacks
might
be
possible
on
this?
The
second
bit
was
the
FEC
header,
or
would
that
be
integrity
protector,
or
only
the
payload
that
it
carries?
So,
do
you
have
the
RTP
header,
then
you
have
to
pack
header
and
then
the
the
payload
so
I
guess
the
payloads
not
protected
by.
I
E
C
Jonathan
Maddox
I
mean
it
certainly
for
perk
arm.
I
agree
that
the
effect
needs
to
be
at
needs
to
be
done
after
the
end
and
encryption
yeah
after
the
undamped
I'm
consider
the
hop
by
hop
and
I'm
concerned
that
there's
a
lot
of
things.
C
You
know
for
the
non
perk
cases,
a
lot
of
things,
architectural
II
tend
to
have
the
smtp
processing
very
early
and
then
repair
flow
stuff
happening.
You
know
much,
you
know
significantly
later,
like
particularly
you
know,
all
the
/
stream
state
you
keep
that
it's
not
the
crypto
state
might
be
done
rather
deeper
down
your
pipeline,
so
forcing
this
stuff
earlier
is
would
be
a
hard
architectural
trange,
whereas
not
doing
this,
it
means
that
flex
package
to
drop
in
replacement
for
up
feck
or
r-tx
so
I'm
concerned.
C
J
E
E
B
So
yeah
I,
I
think
the
payload
format
should
actually
you
could
discuss
this,
but
it
shouldn't
specified
because
it's
actually
not
on
on
flag
fix
order
and
default.
I
think
needs
to
be
that
you're
not
touching
in
this
apt
or
to
save
ASAP
and
the
configuration
of
sot,
p
and
and
the
one
of
the
issues
that
is
applies
to
perk
perfect.
Is
that
detail
SS
OTP
defaults
and
has
no
in
currently
specified
signal
for
reversing
the
order
of
the
smtp
infecting
that's.
B
No,
don't
really
signal,
I
mean
the
birth
order
is
applied
on
sh
IP
level,
it's
not
to
provide
on
payload
type
level,
because
you
can't
really
do
this
operation
on
a
payload
type
specific
level.
You
need
do
it
because
it
affects
how
you
modernize
the
things
between
the
RTP
stack,
sfp,
module
and
eve.
Then
the
you
need
to
it.
Kind
of
applies
to
the
session
like
well,
are
not
on
the
individual
streams.
So
you
have
to
be
careful
here
and
that's
why?
I
think
it's
brilliant.
E
A
Needs
to
specify
our
kasha
flux
right
operator
as
a
second.
So
so
are
you
going
to
discuss
it
in
delco?
So
because
definitely
that
was
my
eye
when
I
was
listen
to
me.
I
said
this
is
a
parrot
issue
because
it
requires
the
pest.
If
you're
saying
need
to
be
in
the
signal
needs,
definitely
something
that
we
need
to
be
done
in
hurt,
because
that
the
signaling
is
audio
issues.
E
A
Again,
what
would
be
the
conclusion?
What
which
would
really
have
something
about
these
about
the
issue
of
when
to
do
the
srp
peas
before
after
defecting
the
payload
or
we
leave
it
to
3,
7,
11
and
2
and
2
3rd
I
mean
that's
a
question
because
it's
like
muggle
said
that
if
you
put
it
in
the
paper
that
when
you
can
do
it,
but
it's
not
the
place
to
yeah.
C
I
mean
I'm
jonathan,
I
mean
I
think,
two
magnets.
This
point
I
think
it's
going
to
be
I
mean
if
you
want
just
try
to.
You,
know,
try
to
use
flex
pack
but
fall
back
to
all
their
stuff
you're
going
to
offer
flex
ack
effect
and
RG
x.
And
you
know
you
don't
want
to
say
that
ok,
I
have
to
do
my
effect
processing
the
different
place
than
the
code,
depending
on
which
payload
format
I
get.
That
would
be
chaos,
but.
C
Well,
let
me
perfect
I
mean
work,
obviously
behaves
a
lot
differently
than
regular.
Refusing
your
doing
double
right,
so
I
mean
I'd,
say
that
you
know
when
you're
negotiating
that
you're
doing
double.
That
means
that
you
need
to
do
your
repair
flows
of
it
differently
and
sort
of
inherently,
because,
obviously
you
have
to
do
it
again
after
the
end
and
whether
you
do
it
before
after
hop-by-hop.
That's
something
perks,
gonna
me
to
decide
so.
E
A
E
J
J
A
E
Next
slide,
please
anything
else
on
the
s,
so
the
result
was
we're
going
to
discuss
it
in
fact,
and
we
hopefully
will
be
able
to
remove
anything
related
SR
to
be
out
of
this
spec
and
and
let
em
be
documented
somewhere
else
next
slide.
So
that's
it.
The
only
other
open
issue
is,
I
think
that
the
protection
recovery
procedures
need
to
be
clarified
a
little
bit.
Some
people
could
take
a
read
through
it
and
see.
What's
clearing,
what's
not
clear
to
them.
That
would
be
helpful,
so
otherwise
I
think
we're
done
and
ready.
Fer.
A
G
A
One
we
have
to
wait
to
the
decision
what
after
perv,
which
means
that,
anyhow,
in
order
to
go
to
work,
you
last
call,
we
still
need
to
get
those
two
seem
to
finalize
the
third
me:
sugar,
yep,
okay,
yep,
great
okay!
Thank
you
any
other.
So
this
is
the
last
presentation
for
our
presentation.
Any
other
issues.
I
want
to
be
stuff
together.
I
D
Soccer
again,
just
a
few
minutes
I
will
be
brief.
I
wanted
to
introduce
very
briefly
what
we're
doing
it
check
frame.
So
I
don't
know
how
many
of
you
are
aware
of
fact,
friend
she's,
a
protocol
that
was
specified
by
the
fact
frame
working
group,
but
not
in
this
area
in
the
transport
area,
which
is
probably
the
reason
why
we
are
not.
There
were
not
so
many
exchanges
between
us,
so
it
was
done
in
ic
63,
well,
maybe
in
2011,
and
not
remember
exactly
but
I
wrong
these
days.
D
Since
then,
the
working
group
has
been
closed.
So
why
am
today
is
that
we
are
extending
this
protocol
back
4m
protocol
towards
new
types
of
FEC
schemes
actually
cuts,
namely
congressional
codes,
so
that
there
are
two
interesting
things
I
think
in
this
work,
which
is
by
the
way
joint
walkways
alley,
began
first
of
all,
a
different
actor
and
if
I
can
discuss
with
some
of
you,
what
a
difference
is
the
benefits
of
doing
a
within
article
below
a
typical
friends,
somewhat
shim
layer
that
fits
between
OTP
and
unity,
then
I
would
be
very
happy.
D
That's
the
first
point
and
second
point
is
a
deuce
schemes
that
we
call
congressional
schemes
FEC
codes.
They
were
in
totally
different
way
from
what
we
had
here
with
XO
codes
or
records
or
with
lemon
curd,
a
lipsy
cut
all
of
them.
This
is
those
are
not
block
codes.
There
are
convolutional
codes,
which
means
that
there
is
a
sliding
incoming
window
that
will
slide
over
the
source
flow
and
whenever
you
need
to
produce
a
repair
packet,
but
you
just
have
to
create
linear
combination,
the
sauce
packets
that
are
within
this
and
curling
window.
D
Unless
it
you
sell
it.
So
it's
totally
different
and
we
carried
out
extensive
personal
simulations
and
there
is
a
key
benefits
in
using
those
cuts
in
terms
of
improved
robustness
and
reduced
end-to-end
latency
in
terms
of
FCA
calling
in
different.
So
if
you
are
interested
first
of
all,
there
will
be
a
presentation.
D
A
Hello,
okay,
so
interested
in
that
I
can
go
to
this
session.
Only
the
other
thing
we
forgive
others
followed
protocol
scheme.
We
decided
it's
okay,
you
can
do
it
in
afterwards
in
a
second
in
the
documents
that
explain
how
to
do
it
with
other
scheme.
That's
that's!
What
we'll
discuss
was
when
we
discussed,
which
way
to
go
here.
That
was
the
conclusion
that
the
working
group
ever
thought:
I'd.
Okay,
hey.