►
From YouTube: IETF102-DMM-20180717-0930
Description
DMM meeting session at IETF102
2018/07/17 0930
https://datatracker.ietf.org/meeting/102/proceedings/
B
B
B
B
B
B
B
B
So
this
is
the
let's
review
the
status
of
each
sha
IETF
documents
or
first
one
is
the
on-demand
mobility
management
document.
So
this
completed
last
call
recent
to
iesg,
and
there
was
some
feedback
and
authors
have
not
yet
resolved
those
issues
yet,
but
today
you
know,
there's
going
to
be
a
presentation
and
I
think
Brian
Haberman
has
is
one
of
the
reviewers
he's
going
to
be
here
as
well
he's
going
to
give
his
feedback.
So
let's
wait
for
that.
That
is
one
document.
B
B
But
after
my
idea,
one
of
the
haven't
seen
you
know,
we
haven't
seen
many
discussions,
but
overall
this
on
a
good
track.
You
know
I
think
the
others
are
making
putting
a
lot
of
efforts
and
I
think
we
are
quite
you
know,
happy
with
the
progress
and
there's
another
document
which
is
the
the
the
pv6,
the
DMM
document
that
is
Carlos,
as
the
first
version
was
posted.
B
B
B
We
need
to
get
some.
You
know,
feedback
from
the
you
know
from
the
working
group.
We
need
more
discussions,
more
comments,
I
think
that
others
have
put
in
a
lot
of
efforts,
but
I
think
we
need
some
good
reviews
all
right
and
moving
on
I
think
this
another.
The
distributed,
mobility,
anchoring
I
think
this
is
I,
think
we
or
the
last
you
know
few
meetings
I.
Think
in
like
we
said.
Essentially,
you
were
not
happy
with
the
quality
of
this
document,
and
so
we
had
a
point.
B
They
editor
for
this
and
Carlos
kindly
you
know
two
core
and
he
did
a
complete
rewrite
at
the
document.
What
was
80
pages
came
down
to
literally,
like
you
know,
order
20
pages.
So
here
you
know
overall,
it's
a
quality
document
at
this
point,
but
at
the
same
time
it
also
reset
so
everything
in
the
sense
that
we
again
need
a
new
set
of
radius,
because
it's
we're
almost
at
a
0-0
version
but
because
you
know
I
think
it's
a
rewrite
of
the
document.
B
But
technically
the
document
is
stable,
right,
I
think,
but
we
need
some
reviewers
I
think
you
know
I
think
any
anywhere
any
volunteers.
Anybody
can
review
this
document.
It
provides
some.
You
know,
feedback
I,
think
that's
you
know.
Maybe
after
Carlos
presentation,
let's
you
know,
identify
some
reviews
and
I
think
the
one
more
the
ammon
ID
document
is
under
the
maintenance
category.
This
finally
is
going
to
be
published
as
RFC
83-71,
congratulations
to
Charlie,
Perkins
and
others.
I
think
I
did
from
what
I
remember
this
document.
B
Was
it's
been
there
for
a
while
I
believe
it
was
published
if
I
remember,
if
I
can
now,
my
daughter
was
born
and
now
she's
a
teenager.
So
it's
as
old
as
that.
So
so
a
license
statement.
I
think
there
were
few
LS
treatments
I.
Think
one
is.
There
is
a
3-2
BLS
request
from
City
for
working
group,
mainly
essentially
it's
an
update
on
the
work
that
they're
doing
with
respect
to
mobile
use
a
plain
optimization
and
they
wanted.
B
Think,
let's
discuss
that
and
finally,
I
think
these
are
the
new
set
of
you
know,
essentially
new
set
of
documents
that
were
published
after
idea,
101
right
a
few
or
mainly
the
survey
six
based,
you
know,
pocs
I
think
you
know
some
again,
mostly
around
user
plain,
optimization,
I
think
because
there
is
a
burning
topic
in
3gpp
and
I.
Think
you
know,
that's
that's
the
key
point
of
discussion
right
and
I.
Think,
let's
you
know,
some
of
these
documents
are
allotted
time
slot
and
you
know
we
will
be.
You
know
discussing
those
documents
and
I.
B
A
C
C
So
our
job
in
this
document
is
kind
of
a
semantic
selection,
basically
notes
that
are
actually
providing
reporting
services,
as
well
as
Jeff's,
basically
implementing
your
desire
for
policies
and
contractions,
so
from
101,
so
from
101
right.
What
we're
really
doing
is
we
had
a
few
actually
so
we've
had
constant
service
influence
if
we
eliminated
those
that
was
actually
discussed
in
one.
We
actually
put
that
out
the
mail.
E
C
Really
emulating
are
able
to
successfully
eliminate
the
influence
in
structures
or
sown
in
show
compliance
or
things
like
triple
yes
and
selection.
We
were
also
able
to
demonstrate
that
it
quiet
view.
The
selection
and
equality
agencies
of
the
information
model
is
better
and
appears
to
be
sufficient
with
what
we
currently
do
a
production.
What
some
people
are
constantly
don't
mean
to
so
we
had
review
the
yank
memorandum,
be
a
compliance
really
the
end
result.
There
was
basically
explaining
really
weird
in
MB
aid,
and
some
of
the
concept
applies.
C
C
Right,
we
did
add
more
examples.
We
only
had
a
few
in
the
beginning.
Part
of
this
was,
as
we
went
from
nine
to
ten.
It
was
a
major
rewrite,
and
so
we
want
to
make
sure
we
had
an
example
for
each
of
the
major
data
structures,
as
well
as
a
life
cycle,
as
we
talked
about
in
101
policy
base,
we're
a
template
based
design,
and
so
we
wanted
to
make
sure
that
we
were
completely
backed
up
in
terms
of
having
examples
in
the
document.
C
So
it's
a
little
bit
bigger,
but
what
we
did
was
we
basically
reduced
it
using
editing
and
then
added
more
back
from
the
examples.
So
it's
a
lightweight
update
to
the
spec.
So
if
you've
read
version
10,
it's
very
consistent
with
that
the
biggest
change,
as
we
talked
about
on
the
mailing
list,
with
service
groups
and
I'll
just
kind
of
blast
through
these
really
quick
in
the
interest
of
time.
C
But
what
we're
trying
to
do
here
is
really
enable
the
selection
so
once
again
we're
interested
in
making
sure
there's
information
available
for
this
to
be
done,
but
we're
not
worried
since
we're
not
specifying
a
protocol.
In
terms
of
who's
actually
doing
the
selection,
the
clients
are
agents,
so,
in
several
examples
that
we
did
in
terms
of
bouncing
its
3gpp
as
well
as
IETF
applications
in
production,
it
seems
to
hold
up
quite
well
as
we
talked
about
before
we're
template
based.
C
C
We
then
broke
out
the
entity
and
config
data
policy
really
hasn't
changed.
So
thank
you.
So
policy
really
hasn't
changed.
What's
actually
happened
here
is
by
adding
the
examples
right
we're
able
to
talk
a
bit
more
about
how
the
usage
applies
down
in
here.
We're
able
to
show
more
of
the
lifecycle
so
once
again
look
to
those
examples.
Those
are
going
to
help
you
out
quite
a
bit,
but
we
were
able
to
demonstrate
full
lifecycle
models
in
the
document.
Data
planes
once
again
haven't
changed.
We
do
right
with
the
policy
configuration.
C
This
was
something
we
did
not
have
time
to
talk
about
in
101,
just
as
a
reminder,
the
biggest
change
that
we
did
from
version
to
version
10
was
we
put
the
policy
templates
underneath
the
data
plane
and
eliminated
some
of
the
old
concepts
of
those
errors.
So
basically
the
system's
really
simple:
you
specify
your
policy
templates
to
install
them
on
the
data
plane
node,
you
set
up
a
DPM
policy
configuration
and
then
that's
your
pre
install
for
so
for
those
of
us
who
have
ever
been
wondering
how
pre-installed
rules
work.
C
This
would
be
kind
of
your
answer,
so
you
can
actually
pre
install
and
then,
when
it
comes
to
the
mobility
context,
you
can
then
reference
those
pre-installed
rules
that
are
here
on
the
data
point
node.
You
can
also
have
policies
on
the
data
play
node
that
just
apply
to
all
traffic.
So
this
is
just
a
general
policy
application.
You
don't
have
to
think
about
it
in
terms
of
a
mobility
context.
C
The
context
as
the
structure
implies
right
is
there
and
we
see
with
respect
to
kind
of
our
overall
design.
You'll
see
we
can
pull
in
mobile,
node
specific
policies.
The
DPN
specific
policies,
if
our
domain
specific
policies
or
templates,
if
you
actually
use
a
domain
to
actually
select
this.
This
is
for
those
of
us
that
worried
about
roaming.
C
You
have
the
specifically
assigned
policy,
that's
available
that
came
from
the
DPN,
so
we're
looking
for-
and
this
gives
you
your
traceability,
make
sure
it
was
actually
installed
and
then,
within
a
specific
service
data
flow,
you
can
actually
have
a
service
data
flow
policy
and
those
are
typically
dynamic.
Rules
that
are
built
on
the
fly
most
prevalent.
Application
is
halti,
but
you'll
see
that
it's
all
basically
there.
So
that's
it's
with
the
exception
of
monitors.
Those
haven't
changed
in
a
long
time.
Watch
what
you
want,
how
you
want
it.
C
E
C
C
Don't
get
the
data
back
in
a
response
on
monitors,
configure
results
and
result.
Notifications
have
not
changed
in
several
revisions
and
then
probably
the
big
thing
that
we
talked
about
in
version
10
was
how
operations
been
modified.
So
we
do
follow
yang
patch
style
operations
at
the
FPC
semantic
level,
and
this
allows
us
to
actually
do
some
interesting
things.
Probably
the
the
one
that's
the
most
interesting
is
the
move.
So
the
idea
of
moving
policies
between
data
plane
nodes
in
a
single
command,
as
opposed
to
kind
of
picking
them
up
telling
one
guy
he's
uninstalling.
C
It's
only
the
other
guy
that
he's
installing.
So
we're
able
to
actually
run
a
lot
of
this
in
the
design
the
inserts
and
the
merges
are
also
interesting.
So
it
allows
us
to
kind
of
set
up
a
situation
with
the
operations
work
and
minimize
the
over
the
wire
communication,
as
it's
eventually
realized
in
a
protocol.
C
Next
steps,
as
we've
talked
about
we're
done
guys,
so
we're
not
intending
to
add
anymore.
We've
made
our
editing
passes.
We've
closed
all
of
our
action
items.
So
at
this
point,
what
we
really
need,
as
was
said,
was
we
need
reviewers
and
to
get
a
good
couple
set
of
reviews,
so
we
can
go
ahead
and
get
this
thing
to
last.
Call
okay,.
F
B
C
So
we
we
still
have
out
the
two
existing
implementations
and
and
they're
primarily
version
4,
based
because
a
lot
of
activity
happened
from
about
version
7
to
version
10
and
then,
as
we
said,
really
version
11
and
then,
when
we
made
the
area
with
12,
11
and
12,
a
really
error
correction
10.
So
the
codebase
has
been
out
for
it's
been
open
source
like
the
open
daylight,
one
I
think
for
now
a
year
and
a
half
and
we've
got
different
people
using
it
at
various
stages.
C
C
Absolutely
because,
right
now,
just
to
let
you
know
on
the
open
daylight
and
it's
actually
a
it's
it's
more
of
a
job
at
issue
and
and
right.
We,
this
is
the
first
time
we're
talking,
call
processing
with
yang
right,
so
that's
kind
of
a
new
concept.
So
right
now
we're
doing
about
a
thousand
TPS
with
a
triple
redundant.
You
know
Sdn
cluster
right
because
we
were
worried
about
five
9s
as
an
operator
all.
C
B
B
B
H
So,
basically,
briefly
to
remind
the
status
of
the
draft
and
the
history
where
we
started
with
this
and
where
we
are
now,
then
again,
I
will
very
quickly
summarize
the
operation
of
the
document
for
you
for
those
that
are
not
aware
and
then
I
will
basically
ask
for
previews,
which
is
the
main
point
of
actually
this
presentation
today.
So
so
in
terms
of
the
story
of
the
graph.
H
These
drug
wars
are
adopted
as
working
with
document
after
longer
meeting,
although
has
been
there
for
quite
a
while,
since
actually
since
the
beginning
of
the
DMM
working
group,
we
started
working
on
this
solution
and
we
even
had
a
couple
of
kind
of
demo
events
in
collocated
with
ITF
meetings.
So
we
have
implementation
available
on
this
document
and
the
solution
is
basically
a
Pima
base
stations
for
DMS,
so
very
simple.
H
We
basically
push
a
mobility
to
the
edge
and
allowed
to
have
more
than
one
core
working
simultaneously
for
a
given
mobile
node
by
the
mobile
node
having
multiple
prefixes
allocated.
So,
basically,
every
time
I'm
aware
now
that
attaches
to
a
network,
it
gets
an
ipv6
prefix,
anchored
that
at
that
access
router
and
it
continues
using
the
former
prefixes
that
were
allocated
in
previews
or
by
previous
access
routers.
H
H
So
here
is
a
bit
of
terminology.
So
we
have
a
couple
of
I
mean
the
uncor,
the
router.
We
call
it
mobility
and
called
an
access
router,
but
it's
basically
the
regular
router
that
is
concentrating
the
access,
router,
LMA
and
functionalities,
depending
on
the
prefix
and
the
and
the
mobile
node.
And
then
we
have
this
entity.
That
is
a
centralized
controller
entity,
but
we
cause
it
relies
mobility
database
and
basically,
as
I
mentioned,
keeping
the
dependence
from
the
mobility
of
the
VIP,
my
notes.
H
So
here
maybe
you
can
write.
Okay,
just
a
picture,
maybe
not
okay.
So
basically,
this
is
a
simple
picture,
summarizing
the
what
happens
when
the
device
attaches
to
the
network
attached
it
to
a
access
router
and
then
the
signal
is
sent
to
this
CMD
to
decentralize
control,
plane
entity,
which
is
the
one
that
creates
divining
and
replies
with
the
pba
to
the
to
Virata.
H
So
basically,
this
kind
of
as
the
CND
is
the
LMA
and
the
Mar
at
this
point
in
time
is
actually
not
even
a
mark
is
basically
the
encoding,
the
preface
that
is
locally
allocated
from
from
data
Explorer.
Then,
if
we
move
it
device
move,
maybe
you
can
also
put
a
picture
on
so
in
that
case,
if
the
device
attaches
to
a
different
exit
router,
it
will
also
signal
that
will
be
detected.
H
That
will
be
seen
added
to
the
to
the
scene
into
a
centralized
mobility
database
and
that
guy
will
be
the
one
signaling
to
the
different
anchors
that
are
available
or
are
active
at
a
time.
In
this
case
the
pivot
one
and
any
one,
the
new
one
will
knock
it
locally
and
corporate
fix
and
the
neck,
the
the
other
one
will
set
up
a
panel
with
a
new
point
of
attachment.
In
this
case
the
divine
we
have
two
addresses
one
on
call
at
mark
two.
There
is
no
need
for
any
tunnel.
H
Anything
is
a
circular
route
for
that
prefix
and
one
ankle
at
marwan.
The
traffic
will
be
related
to
the
current
location,
which
is
be
a
mark.
2,
so,
basically,
is
standing,
be
meep
to
support
multiple
angles
and
next
one,
which
is,
should
be
the
last
one.
So
this
in
this
division
we
seems
the
adoption
we
address
all
the
comments
we
got
actually
quite
a
few
revisions
from
during
the
option
call
from
different
guys,
so
we
have
results
reviews
in
the
current
version.
H
So
the
document,
although
has
been
presently
adopted,
is
point
mature
in
the
sense
that
we
had
a
lot
of
revisions
when
it
was
individual.
So
we
really
need
some
reviews
to
move
it
forward.
So
that's
what
I
would
like
to
do.
A
sixer
see
they
can
ask
for
reviews.
I
come
on
appeal
myself,
for
example,
for
the
previous
one
to
review
with
somebody
from
the
team
and
do
you
this
one,
this
kind
of
things,
I
guess
no!.
B
H
B
B
J
Go
so
good
morning,
this
is
pooja
from
Cisco.
This
is
an
update
on
drop
by
tfGM
I'm,
sorry,
six
mobile
airplane.
It
would
be
quite
fast,
so
the
summary
of
updates
from
version
one
to
version
two.
It
was
in
essence
just
to
clarify
some
of
the
questions
that
we
had
on
the
mailing
list.
So
first
one
of
them
was
to
clarify
the
type
of
PD
you
system
supported.
So
in
essence,
this
IP
v4
v6.
J
We
only
seek
serenade
on
a
structure
and
in
the
in
the
drop
you
actually
have,
which
is
obvious
expansions
in
supporting
which
kind
of
PD
uses
anti.
The
second
modification
that
we
introduced
in
a
draft
will
we
added
an
appendix
where
we
added
an
to
current
implementations,
so
we
have
an
open
source,
p4
code
that
is
implementing
the
sorry.
Six
functions
that
are
introduced
in
this
document
is
specific
to
mobility,
and
we
also
have
two
ongoing
talks
for
M,
Court
and
open
air
interface,
which
we
documented
in
a
separate
draft.
J
We
are
just
referencing
to
it.
We
also
added
some
references
to
the
what
technologies
are
used
for
networking
slicing.
So
this
is
referring
to
other
crafts
and
we
are
added
some
control
plan
considerations.
There
was
also
another
we
fix
some
typos
as
well,
and
there
was
additionally,
there
is
one
more
change
in
the
draft,
which
is
just
a
clarification
and
how
the
interworking
mechanism
works.
So
the
end
so
d
end
at
M.
J
That
could
be
six
that
t
function
and
how
does
it
work,
based
on
our
question
that
there
wasn't
a
draft
so
as
nexus
tips
what
the
authors,
what
we
are
working
on
is
on
how
to
to
support
a
different
gtp
you
messages
and
extension
headers.
So,
for
example,
if
ideas
container,
we
are
still
analyzing
how
to
do
this.
We
believe
that
maybe
we
do
not
need
to
modify
the
gtp
you
extensional
Heather,
but
this
is
a
still
being
discussed.
J
If
you
have
any
input
please
or
any
feedback
test,
please
come
to
us
and
then
there
was
also
another
question
on
how
we
support
different
types
of
EPF.
So,
for
example,
uplink
classifier
idea
for
branching
point
and
we
need
to
add
an
illustration
on
the
draft
and
on
this
topic
that
we
will
add
and
we
are
working
on-
that's
it
for
the
update.
Any
feedback
is
welcome,
so
Pablo.
J
So
I
don't
think
there
is
that
much
work
left
on
the
draft.
We
just
say
the
most
important
topic
is
just
which
was
then
the
function
coverage.
We
also
received
some
comments
on
the
motivation
of
this
work
and
for
this
reason
the
authors
have
worked
in
a
separate
raft
which
we
just
posted
last
Sunday
we
are
not
presenting.
That
is
explaining
the
use
cases
for
every
six
and
mobility,
and
that's
only.
K
J
F
Satoru
mastema
Softbank,
specifically,
this
document
does
not
leave
you
by
CPP
folks,
but
certain
for
some
of
the
field.
Work
really
hard
line
discussion
but
later
on
this
agenda
at
the
feet
of
upon
the
Year
Honor
Society.
So
then
those
will
be
as
kind
of
the
SL
physics
we
take
into
account
by
them.
L
Okay,
please
I'm,
also
working
in
fact
Istanbul
in
CPP
I
said
I.
Have
you
these
people
before
for
I,
don't
know
three
division,
maybe
in
what
I
think
most
a
lot
of
constant
revision
based
on
the
forty
TDP
you
function
and
the
faculty
there's
a
lot
of
a
new
function
have
been
introduced
by
the
faculty.
For
example,
we
forget
serious
now
asking
that
the
paper
had
not
aware
of
document
is
in
your
function.
Don't
need
any
consideration,
how
to
set
boundaries
and
do
excuse
me.
Can
you
please
repeat
your
question.
J
L
Said
there's
some
new
fifty
introduced
some
new
function
for
the
user
plan
function,
for
example
like
a
vanity
press
indication
and
qfo
I
first
follow
identity.
This
add
new
identity
have
been
introduced
for
the
new
William
Ltd
you
but
I
say
because
a
city
for
had
not
defined
anyway
you,
but
we
need
to
include
this
new
function
and
identity.
In
the
end,
you
use
a
plan
protocol
but
based
on
I,
checked
Boyan.
This
new
function
has
not
been
discussed
in
your
paper.
J
B
F
B
H
Okay,
hello,
this
campus
again,
so
I'm
really
update
on
the
issue
that
Molly
the
anchor
draft,
so
just
the
status
as
we
mentioned
at
the
beginning.
We
have
a
significant
update
for
London,
so
we
went
from
more
than
forty
pages
to
around
50
so
trying
to
improve
their
ability
at
the
document
and
also
improve
the
chances
of
people
willing
to
take
a
look
at
it.
So
that
was
the
main
goal
at
when
we
did
that.
H
So
we
try
to
simplify
the
complexity,
simplified
that
Rowan's
try
to
have
a
less,
maybe
less
ambitious
scope,
but
more
useful,
and
since
then
we
we
had
a
review
from
Marco.
That
was
very,
very
useful
and
we
have
a
couple
of
revisions
in
since
then
trying
to
address
the
comments
from
from
Marco.
So
we
try
to
improve
the
focus
at
the
documents
or
less
PP
simple.
That
was
a
principle
that
we
we
try
to
do
or
with
we
have
in
mind.
H
Basically,
we
we
believe
that
another
document
is
much
easier
to
read
and
it's
more
concrete,
I
don't
know
Marco.
If
you
have
the
chance
to
take
a
look
at
the
revision,
that
will
be
helpful.
Did
you
in
the
future
in
you?
Can
we
can
discuss
offline
about
that,
and
also
the
drop
is
available
me
happening
based
somebody
wants
to
contribute,
we
can
do
it
directly
there.
H
So
this
is
the
the
table
of
contents
of
the
version
0-8,
so
the
one
that
we
beat
for
London-
and
this
is
the
table
of
content
of
the
version
that
we
have
in
version
10.
There
isn't
something
that
is
less,
because
there
is
a
type
own
version,
10,
something
that
shouldn't
be
there,
but
I
will
remove
that
after
the
meeting.
H
But
basically
what
we
try
to
do
is
to
identify
the
kind
of
three
on
Korean
cases
that
we
we
have,
which
are
the
what
we
call
in
America
case,
the
Mobility
case
in
which
we
use
traffic
radiation
and
the
mobility
case
in
which
we
have
fun
quarter
location.
This
basically
came
up
from
the
review
that
Marco
did
the
suggestion
that
he
he
proposed
to
keep
data
structure
or
to
use
data
structure.
So
these
are
the
three
cases
that
we
consider
so
they
know
many
cases.
There
is
no
need
of
an
arrest
continuity.
They
are.
H
The
mobility
is
either
even
either
sorry
not
require
or
sandal
by
on
another
layer
layer,
four
application
layer
whatever.
So,
basically
there
is
no
mobility
at
all.
They
know
every
time
it
moves.
Its
changes
addresses
that
simple
and
then
we
have
the
mobility
cases,
and
in
that
case,
in
that
we
have
two
sub
cases,
one
s
where
we
have
traffic
federation.
So
when
the
move,
one
node
moves
to
another
point
of
attachment,
the
traffic
for
the
preface
that
why
not
values
in
is
redirected
to
the
current
point
of
attachment.
H
So
this
basically
the
case
of
the
drafted
a
president
before
and
then
the
retailer
case
in
which
the
anchor
is
actually
relegated.
So
that's
more
complex
that
require
some
support
in
the
network
in
order
to
actually
move
the
anchor
of
the
demo
I
notice
you
see.
So
those
are
the
three
cases
that
we
described
in
the
in
the
document
so
as
before,
again
next
steps,
we
believe
that
argument
is
now
stable
in
the
sense
that
we
reduce
the
complexity.
We
got
a
initial
to
you
from
Marco,
basically
saying
if
I
understood,
probably
otherwise
correct
me.
H
If
I
is
at
your
point
that
we
were
on
the
right
track
in
terms
of
simplifying
the
document.
Now
we
fully
address
the
revision
of
Marco
with
in
more
reviews,
because,
as
we
mentioned,
this
really
had
a
lot
of
changes
since
before
London.
So
now
it's
something
that
is
not
simpler
to
read
a
much
shorter,
so
I
would
like
to
really,
as
you
guys
could
take
a
look
at
the
data
point,
because
otherwise
we
cannot
do
anything
else.
I
mean
from
my
stylist.
H
B
H
I
summarized
those
in
in
London
good
reviews:
I
mean
we
remove
all
the
stuff
related
to
your
icon,
mobility
and
Korean
things
with
Nemo
part.
We
also
simplify.
So
basically,
you
focus
on
very
simple
things
like
those
are
the
three
cases.
Those
are
the
things
that
you
need
to
do.
Those
are
the
entities
referring
to
this.
That
I
mean
early,
that
is
in
the
deployment
model,
I
think
document.
H
So
we
try
to
use
the
actual
terminology
that
is
kind
of
adapted
in
the
working
group,
and
it's
really
simple
so
I
mean
we
just
remove
all
the
fat
kind
of
thing
focus
on
the
three
main
cases
that
are
those
three
ones,
and
then
we
provide
some
pictures
and
like
okay
in
this
case,
this
is
the
entity
this
identity.
This
is
the
scenery
that
will
require
a
play.
B
B
M
I
won't
mean
ten
minutes.
Anyway,
we
had
a
discussion
early
this
morning,
I
just
to
remind
the
group.
The
document
went
through
a
preview
by
Brian
and
he
had
five
comments
will
listed
the
comments
on
the
list
and
the
initial
response,
and
since
then
we
had
some
email
exchange
trying
to
resolve
these
issues,
and
can
we
go
through
the
next
slide?
M
M
M
N
O
M
So
the
feeling
was
that
okay,
we
describe
the
extensions
for
the
socket
interface,
but
the
feeling
was
that
the
description
of
the
whole
picture,
meaning
okay,
an
application-
requires
something
from
using
the
socket
interface.
How
is
the
interaction
with
the
network
or
does
the
network?
What
does
it
reply
and
things
like
that
could
be
better
described
in
the
draft?
So
we
agree
and
initially
I
wasn't
sure
I
wanted
to
do
that,
because
we
didn't
want
to
impose
a
specific
solution
for
this
would
be
done.
M
There
could
be
different
alternatives,
but
eventually
we
agreed
that
maybe
better
explanation
could
help.
So
we
are
going
to
add
an
example
of
some
message
flow
indicating
that
this
is
an
example.
It's
not
the
only
message
flow
which
is
correct,
but
hopefully
that
will
improve
the
reader
help
people
understand
the
draft
better.
So
these
are
really
the
two
action
items
that
the
writers
have
and-
and
that
was
the
agreement
in
our
Monday
morning
session,
and
my
next
step
is
to
update
the
draft
openly
as
early
as
next
week
and
I'm,
not
sure
about
the
process.
M
O
Mascots
HT
consulting
I
lived
through
the
OSI
Wars
I'm,
a
Stax
person,
and
going
back
to
come
about
IP
session
to
you
just
say,
safe
session.
You
can't
really
say
transport
unless
you
can
assure
you
do
not
disrupt
the
transport
port
assignments,
which
is
an
OS
concept.
So
only
thing
you
can
say
I
believe,
is
session.
Continuity.
Ip
continuity
only
applies
if,
because
of
whatever
you
actually
maintain
the
IP,
addressing
which
you
actually
don't
so
being
a
survivor,
ragazza
I
would
say
that
you
should
just
say
session
continuity
and
leave
it
at
that.
J
J
O
Q
Q
My
my
basically
issues,
two
three
and
four
or
driven
by
the
fact
that
there
is
not
enough
descriptions
in
the
document
for
someone
to
actually
look
at
it
from
from
the
outside
and
say:
okay
I
understand
how
this
concept
is
supposed
to
work.
If
Danny
goes
in
and
does
he
put
any
puts
of
changes
to
kind
of
give
that
conceptual
example,
then
that
should
help
with
that
clarity,
because
you
know
it's
hard
enough
for
for
somebody
who
to
review
it
who's,
not
involved
in
the
working
group,
but
they're
involved
in
the
idea.
Q
If
you
ask
somebody
who's
actually
in
the
interior,
if
you
ask
somebody
to
review
this
from,
say
the
transport
area
or
the
security
area,
they
need
that
context
right,
and
so
you
I
would
much
rather
see
you
guys
fix
this
clarity
problem
here
sooner
rather
than
later,
so
that
you
don't
get.
You
know
three
discusses
from
three
different
areas
and
then
have
to
try
and
go
back
and
retrofit
everything
yeah.
B
Yeah
very
good
thanks,
fun,
I
think
that
helps
I,
think
Danny
I
think.
Definitely
there
is
work
for
this
I
think
you
know,
discuss
it's
rich,
but
Tom.
So
3d
movie
is
currently
reviewing
referencing
this
document
right,
not
this
document,
it's
referencing.
There
are
other
extension
extension
which
reference.
M
G
M
L
R
So
this
is
a
brief
update
on
the
floating
anchors
draft
and
since
we
have
ten
minutes
so
if
you
need
details,
please
have
a
look
to
the
draft
and
or
to
series
presentation
from
last
idea
at
the
moment.
So
for
today
we
just
want
to
highlight
a
little
bit
the
basic
principles
of
that
approach
and
how
it
fits
into
the
overall
picture
of
the
data
plane
discussion.
So
the
draft
with
the
basic
principles
and
operation
has
been
published
before
last
ITF
and
has
been
discussed
in
the
meanwhile.
R
We
received
some
very
wearable
comments,
which
cost
some
discussion
on
the
mailing
list.
Some
of
the
comments
went
a
little
bit
ahead
of
the
basic
principles,
but
very
important
to
see
how
that
concept
fits
when
being
deployed
with
a
cellular
architecture.
So
in
the
next
revision
we
plan
to
publish
soon
would
like
to
address
most
of
these
items
at
some
level.
R
R
Also
Network
slicing
for
customized
provisioning
of
network
should
be,
in
the
account
just
a
reference
to
current
study
and
3gpp
about
our
reliability
and
low
latency
communication,
which
really
discusses
also
the
dilemma
city
in
the
network
by
moving
and
devices,
but
also
changing
data
on
network,
so
that
peering
seems
to
be
important
in
particular,
for
edge
computation
and
services.
So
that's
our
main
objectives.
In
general,
we
look
at
annexes
independent
data
plane.
R
So
by
the
architecture
we
propose,
we
enable
optimized
routes
and
between
the
mobile
node
and
one
or
multiple
correspondence
services,
yeah
moving
from
centralized
or
fixed
anchor
to
decentralize
data
plane.
That's
the
original
requirement,
an
objective
of
the
DMM
working
group,
which
is
on
stage
since
a
while
and
last
but
not
least
in
the
draft
we
address
and
would
like
to
strengthen
the
discussion
on
how
that
approach
can
be
deployed
and
integrated
tightly
or
loosely
where
the
mobile
cellular
architecture.
R
R
So
by
applying
default
routes
in
between
one
or
multiple
correspondent
edges.
On
the
mobile
knowledge,
we
have
a
low
number
of
data
plane
nodes
which
are
impacted
so
in
case
the
mobile
moves
and
gets
a
new
mobile
edge
being
assigned.
We
apply
house
States
at
the
correspondent
edges
and
enable
optimized
roaring
in
between
the
multiple
services
and
the
mobile
node.
We
have
an
example
on
the
next
slide
here,
so
to
enable
this
traffic
steering
the
approach
is
compatible
with
any
of
the
currently
being
discussed
data
plane
protocols.
R
So
the
draft
currently
focuses
on
segment
routing,
because
here
we
have
the
flexibility
in
choosing
intermediate
hops,
between
the
corresponding
edge
and
the
mobile
edge,
but
other
proposals
like
locator
rewriting
or
any
ID,
lock
separation,
kind
of
approach.
I
can
apply
it
here.
These
states
that
the
correspondent
edge
are
of
or
can
be,
of
transient
nature.
So
whenever
the
mobile
node
gives
up
its
IP
address
and
IP
address
continually
is
not
required
anymore.
The
correspondent
edge
states
can
be
diminished
as
what
not
least
by
the
programmatic
flexibility.
R
We
can
go
beyond
the
actual
traffic
steering
and
additional
states
and
policies
for
metering
who
is
monitoring,
reporting
gating,
whatever
also
paging
support
can
be
taking
into
account
and
enforce
in
any
of
these
edges,
either
mobile
or
correspondent.
A
brief
example
here
very
similar
to
what's
represented
last
idea,
so
the
control
plane-
let's
not
focus
too
much
on
that
here,
but
we
have
the
control
plane,
leveraging
topology
awareness
and
location
awareness
on
the
data
plane.
Let's
assume
the
mobile
node
attaches
to
a
nexus
gateway
to
purchase
its
IP
address
anchor.
R
So
here
in
the
example
there,
with
the
green
IP
address,
with
prefix
2
from
gateway,
2
is
the
one
that's
to
be
maintained,
so
IP
address
continuty
has
to
be
supported
when
the
mobile
node
gets
a
new
access
gateway
being
assigned.
In
addition,
the
mobile
node
may
receive
a
local
prefix
which
is
route
over
but
valid
only
as
long
as
the
mobile
node
is
attached
to
access
gateway
to.
R
So
in
that
example,
the
mobile
node
has
communication
with
two
correspondent
nodes-
one
you
can
see
here
and
the
other
one
here
so
since
the
IP
address
is
from
X's
gateway
to
whose
network
default
routes
apply
in
between
correspondent
nodes
and
the
mobile
node.
So
what
happens
if
the
mobile
moves
and
gets
assigned
a
new
access
gateway
so
further
requirement
on
the
local
prefix,
the
yellow
one
is
deprecated
and
the
mobile
node
gets
a
new
one,
but
the
green
IP
address
has
to
be
maintained.
R
So
we
call
that
IP
address
continuity,
somehow
related
to
a
previous
discussion
with
Danny's
drafts,
or
maybe
in
terms
of
terminology
we
have
to
line
it
here.
Important
point
is
that
the
control
plane
can
leverage
the
programmability
of
the
correspondent
edges
by
enforcing
additional
policies
for
traffic
steering.
So
what
we
do
is
basically
enforcing
States
at
the
correspondent
edges
enabling
segment
routing,
ideally
located
separation
approaches
tunneling,
whatever
you
want
to
steer
the
traffic
to
the
Mobile's
current
access
gateway.
So
by
that
we
can
enable
optimized
routing.
R
The
feedback
we
received
was
very
valuable.
So
since
the
graph
focuses
on
segment
routing,
there
was
a
request
to
the
visa
section
about
how
that
works.
With
other
data
plane
protocols
like
iolaus,
then
there
was
a
request
to
add
more
information
about
additional
functional
or
non-functional
aspects
like
device,
dormancy
and
paging.
R
So
that's
very
important
if
you
integrate
that
approach
with
a
cellular
network
or
the
mobile
may
enter
this
continuous
reception
mode
and
the
exact
location
of
the
mobile
node
is
not
available
in
all
the
day,
looking
nodes.
So
that's
something
we
consider
very
high
importance
and
we
will
add
a
section
on
that
here.
R
So
the
big
question
on
the
mailing
list
was
how
the
correspondent
edges
get
information
from
the
control
plane
for
traffic
steering.
So
the
current
route
focuses
on
a
reactive
approach
which
is
work
working,
but
improvements
can
be
made
here.
So
we
are
currently
discussing
the
evaluating
different,
proactive
approaches
just
to
avoid
factory
ordering
and
that
kind
of
things
have
been
addressed
in
the
feedback
we
received.
So
that's
very
well
ability
to
update
the
draft.
R
There
was
a
question
of
so.
If
we
move
the
exes
edge
to
the
very
much
extreme,
so
two
base
station
or
something
that
axis
edge,
maybe
axis
specific,
but
we
could
clarify
that
if
we
have
a
loose
coupling
between
the
programmable
data
plane,
node
at
the
edge
and
the
actual
radio
technology
or
fixed
on
technology,
so
there
is
no
no
tight
binding
between
that
approach
that
we
present
in
the
draft
and
the
actual
access
technology
here
same
the
interfaces.
R
So
how
do
the
correspondent
edges
receive
the
policies
for
traffic
steering
and
additional
ones
for
metering
whatever
from
the
control
plan?
So
in
the
view
of
cellular
integration,
cellular
technologies
and
architects
are
very
particular,
but
we
have
some
ideas
which
we
would
like
to
discuss
with
you,
folks
how
that
picture
could
fit
into
a
3gpp
next-gen
architecture
as
an
example-
and
this
is
also
pretty
much
related
to
some
proposals
that
are
currently
being
discussed
in
3gpp,
where
the
data
network
plays
a
role
in
influencing
the
traffic
steering
policies.
R
Last
but
not
least
so,
correspondent
edges
have
States.
When
do
we
actually
delete
the
states?
Is
that
something
in
force
by
the
control
plane?
Is
it
soft
state?
So
there
are
a
few
things
that
should
still
be
thought
about
and
actually
captured
in.
The
draft,
so
more
comments
are,
of
course,
appreciated
and
hope
during
this
week
and
then
online
we
can
have
more
discussion
about
these.
Ok.
K
B
S
R
T
B
F
This
is
what
I
presently
in
the
society
meeting
in
London
on.
What
we
asked
to
IFD
is
to
provide
any
information
regarding
the
user
Prime
protocol.
So
then
you
can
see
the
mini
activity
on
this.
So
I
really
appreciate
all
the
effort
on
the
ITF
side
so
that
the
discussion
and
out
comedy
thought
about
discussing
videos.
So
if
you
can
show
the
draft
autism
in
statement
or.
F
B
F
Have
so
other,
as
we
equated
to
ITF,
had
to
provide
any
information
and
with
the
some
additional
information
we
will
work
on
the
to
this
figure
out
the
criteria
to
compare
the
candidate
solution
and
then
need
to
be
stopped.
The
award
of
the
candidate
from
the
ITF
side
so
I
think
a
dirty
Gary,
probably
techies
under
content
are
including
the
the
document
are
only
pointing
to
the
data.
I
internet
dwarf
would
be
a
suspicion
suitable
for
Dia
and
he
is
about
to
die
the
CPP
for
us.
Thank
so.
B
N
E
N
B
Fair
point
are:
do
I
think
this
is
what
you
know,
I
think,
last
time
what
we
discussed.
We
are
not
pointing
to
any
solution
documents.
If
you
look
at
these
two
individual
documents
that
we
referenced
I
think
you're
differencing
three
documents:
one
is
a
working
group
document
which
is
on
track
to
become
a
standard.
I
think
that
yeah.
B
That
setting
a
right
level
absolutely
so.
The
next
two
documents
are
a
requirement.
Analysis,
yeah
and
you
know
comparison
of
various
approaches,
so
these
are
neutral
to
any
one
specific
approach,
so
I
think,
given
that
these
documents
are
neutral-
and
we
are
stating
that
you
know
IDF
is
not
a
couple
of
points-
we
are
stating
that
idea,
which
is
not
going
to
pick
any
one
document.
As
the
document
for
this,
we
are
going
to
standardize,
potentially
maybe
more
than
one
solution
more
than
one
protocol
right,
but
the
analysis.
B
N
Just
to
me,
I,
don't
know
how
a
document
can
be
a
completely
neutral
analysis,
except
you
know
if
it
hasn't
been
through
the
process,
it
is
probably
incomplete
and
hasn't
been
fully
commented
on
because
it
wasn't
adopted
by
the
working
group.
So
that,
for
me,
is
a
bit
of
a
red
flag
and,
like
I,
said
you're
setting
the
wrong
level
of
expectation
as
to
how
this
stuff
is
going
to
progress
in
the
ietf.
You
know
I
3
GPP,
you
know
they're
trying
to
work.
N
P
P
The
treatment
yeah
hi
see
this
is
Uma
from
Bobby,
so
this
response
is
good,
very
good
actually.
But
the
only
comment
is:
please
say
that
this
is
initial
response.
There
are
various
proposals
that
need
to
be
thoroughly
discussed.
Some
of
the
proposals
recently
published
and
some
of
the
some
of
the
proposals
are
published.
Like
you
know,
last
five
ETF,
but
not
significant
discussion
happened.
P
B
More,
we
posted
the
text
to
the
working
loop,
for
you
know
for
any
changes
right,
I
think
so
we
are
not
editing
this
at
this
point
right,
I
think
you
know.
If
there
were
any
comments,
we
gave
a
deadline
of
last
Friday
right
to
make
any
suggestions
comes,
so
we
are
just
going
by.
You
know
what
was
discussed
in
the
manor.
So
if
you
comment
notes,
maybe
I
I
don't
know
if
they
say
you
need
to
extend
that.
You
know
we
can
probably
extend
it
for
another
few
days
for
additional
comments.
B
P
U
U
Any
particular
solution
is
basically
completely
studying
that
the
3G
requirements
and
that's
why
it's
usual
as
well
as
far
as
the
optimization
optimized
mobile
user
plan
is
concerned,
there
are
different
solutions
that
you're
considering
in
that
in
that
in
that
draft,
however,
we
are
not
particularly
advocating
a
particular
solution,
because
the
the
the
final
decision
actually
is
based
on
the
analysis
that
we
TPP
will
will
conduct
on
its
own
and,
as
a
result,
all
we
are
doing
in
there.
We
are
trying
to
compare
different
protocols
with
existing
gtp
scope
and
and
and
behavior
that's
why?
U
Even
in
the
in
the
optimum
is
mobile
user
plain,
we
stood
away
from
doing
any
any
or
prove
providing
any
any
final
decision
in
which
one
of
these
things
we
are
favored
in
terms
of
3gpp
study
items,
so
they
will
pick
up
what
what
needs
to
be
done.
They
will
conduct
this
study.
We
are
made
merely
providing
information
in
there.
Okay,.
V
We
have
very
tough
timelines
in
suite
EPP,
but
nevertheless
we
started
to
study
now
and
if
you
take
longer,
then
maybe
it's
not
really
sixteen,
where
this
will
be
adopted,
but
really
seventeen,
if,
if
there's
an
agreement
to
adopt
anything
from
EMM
Group
outcome,
but
I
wouldn't
go
into
a
discussion
where
practically
our
timelines
to
switch
BP
timelines.
Push
you
too
hard
to
complete
your
work.
Make
sure
that
your
work
is
stable
and
most
of
all
makes
sure
that
the
informal
and
formal
communication
between
this
group
and
ct4
is
well
in
the
life.
Thank
you.
V
M
Is
Danny
so
just
to
add
a
little
bit
to
what
Derek
said
and
to
answer
Dave
3gpp
is
well
aware
of
the
statuses
of
the
different
drafts
in
ITF,
so
they
understand
the
difference
between
an
individual
draft
and
a
working
group
draft,
so
they
won't
be
confused
by
whatever
we
provide
them
and
I
also
agree
that
we
need
to
continue
the
work,
especially
as
it's
of
interest
for
ct4
and
I.
Think
Gehrig
is
being
a
middle
polite
or
over
polite.
M
N
Dave
again
just
one
other
thing:
I
remember:
we
had
the
discussion
of
what
class
of
information
we
would
Lee
A's
in
London
and
what
it
was
is.
We
would
not
provide
analysis.
We
would
simply
provide
raw
information
and,
let's
see
t4,
do
the
analysis,
so
we
seem
to
have
be
going
backwards
on
that.
That.
B
Is
not
true,
Dave
I
think
if
you
read
the
response
that
you
know
that
what
you're
seeing
on
the
screen
we
are
clearly
stating
that
you
know
we
are
going
to
standardize
now,
potentially
more
than
solution
or
approach.
We
are
not
going
to
say
you're
only
going
to
characterize
the
solution
approach,
but
you're
not
going
to
say
that
this
is
a
ticket.
V
V
One
issue
is,
of
course,
how
fast
you
want
this
work
to
conclude
and
to
possibly
be
adopted
in
in
3gpp.
That
is
up
to
you.
How
fast
you
can
do
the
work
and
I
don't
want
to
influence
anything
on
that
I
think
you're
progressing
at
the
moment
at
a
very
high
speed,
and
you
have
to
be
careful
that
your
work
still
stay
stable
right.
The
other
thing
is
that
the
the
study
we
are
conducting
now
in
CT
4
relies
on
that.
You
have
a
certain
amount
of
stability
in
your
work
and
clarity.
V
L
I
think
people
have
a
purpose
to
issue
one
use
that
as
VP,
we
have
very
quickly
working
down
then
ITF
other
assay.
This
11
says
that
this
celebrity
TPU
for
diabetes,
16
in
understanding
even
for
the
Whistling.
The
faulty
Santa's
is
not
a
stable
enough
in
a
settled
over
20
study.
I
think
is
kinda
studied
in
the
assay
to
there's.
L
A
lot
of
new
feature
was
introduced
in
a
with
16,
for
example,
data
that
Travis
there
impotent
even
accessed
an
alley
between
50
40
and
the
Wi-Fi
access,
and
there's
also
actually
will
have
additional
less
in
communication.
In
that
way,
they
were
introduced
a
to
apparel
kit,
video
connection
for
1ue
to
improve
the
reliability.
So
in
there's
a
lot
of
a
new
feature
and
function
point
is
cancer
and
studying
in
CVP,
so
I
think
it's
very
hard
for
our
our
the
amendment
to
follow
the
steps
or
the
purpose
in
in
sweetie
pee.
L
Even
we
working
in
ITF,
wrong,
heinous
sweetest
and
a
fan
is
very
powerful
for
the
progress
of
the
standard.
I
would
stand
us
so
our
previous
authority
that
we
provided
our
text
as
as
a
current
tourism,
and
that
we
continue
you
would
identify
the
real
requirement
upon
the
poverty
decision.
Then
we
focus
our
our
our
solution
in
current
5,
DVD's
beauty
isn't
or
enough
to
support
50
requirement.
Okay,
thank
you.
So
any.
V
To
come
again,
yeah
I
just
want
to
avoid
that
we
have
a
3gpp
discussion
in
this
room,
so
what
essay
2
is
currently
doing
is
studying
an
enormous
amount
of
new
feet.
Just
and
nobody
knows
what
will
be
the
outcome
of
that
right
and
it's
not
your
job
to
evaluate
it
at
this
moment,
and
you
can
look
at
that.
You
can
draw
some
conclusions
from
that.
V
That's
fine,
but
from
my
understanding,
let's
take
CT
for
as
the
single
entry
point
for
the
communication
between
the
MST
mm
on
this
topic
and
let
CT
force
evaluate
what
the
real
requirements
in
release
16
will
be.
This
will
change
over
the
time,
as
other
groups
in
3gpp
will
progress.
But
let's
keep
that
discussion
out
of
this
room.
Please
no.
D
B
You
know
think
oh
yeah
she's
next
one
I
think
with
that
I
think
we
are
going
to
extend
the
you
know
four,
so
I
think
whatever
you
guys
have
posted.
If
there
any
objections
to
the
ls
response,
please
you
know
send
it
to
the
mailer
and
I
think,
but
we
will
extend
the
deadline
by
another
few
days,
but
that
LS
response
will
go
out
this
car
this
week.
I
B
I
I
First
area
I
introduced
background
on
the
overview.
This
arises
work,
so
we
mainly
have
too
much
patience
on
this
process.
Work
first
is
if
showing
a
showing
understanding,
level
or
ITF
side
force
with
these
specifications,
it
will
be
helped
to
make
three
TPP
folks
recognize.
The
ITF
proposal
was
concerning,
and
it
will
be
helped
to
understand
Assaf
occasion
for
IT
folks
and
second,
we
are
clarifying
problem
spaces
and
improve
our
points.
I
It's
very
important
for
proposing
new
portables
to
switch
BP,
and
we
build
at
this
document
should
be
part
of
recently
right.
So
in
this
work
we
undereyes
the
gtp.
You
specification
from
TS
two
to
nine
to
eight
months
and
5g
architectural
equipment
from
TF
twenty
three
five,
one,
five,
thirty,
two
absolute
suite
at
cetera.
In
addition,
we
extracted
some
evaluation
aspects
for
studying
the
use
of
plant
protocol,
flaunt
analyzed
5g
architecture
requirements.
I
From
the
this
work.
We
provided
a
certain
gtp
you,
the
patient's
six
oxygen
requirements
for
user
plane
and
seven
operational
aspects.
In
addition,
we
could
find
some
potential
gaps
between
the
gtp
you
as
a
current
user,
plane
protocol
and
5g
architecture.
The
comments
so
in
the
following
few
strides
either
explained
context
of
a
nice
draft.
So
in
these
tools
right
shows
the
gtp
of
its
operations.
For
example
gtp.
You
is
basically
point-to-point
toneri
product
and
sometimes
supports
a
point-to-point
tunneling
and
in
the
5g
architecture
qfy
us
qfy
is
used
for
identifying
feeding
session
and
DDP.
I
This
is
this:
this
is
a
example
of
gtp
forward,
so
it
will
be
helped
to
understand
what
did
you
a
pocket
in
these
two
right
shows
article
requirement,
5g
architecture,
so,
for
example,
in
5
g
5g
architecture
defines
PT
session
or
four
types
of
gtp
bit
of
station
ipv4
IP
received
internet
and
a
structure
so
use
a
plane.
Protocol
passed
combat
this
PT
session
or
5g
architecture
arose
to.
I
I
Here's
a
plane
protocol
must
compare
ok,
qfi
information,
so
this
table
shows
average
effects
abrogation
aspects
extracted
from
of
teacher
government's,
so
as
I
mentioned
in
the
previous
right
use,
the
plane
protocol
must
convey
every
must
cover
every
decision,
time
and
earth
nature
of
the
past
use.
The
plane
protocol
will
be
required
to
support
much
point-to-point
protocol
toneri
to
reduce
a
money
load
of
management
of
apostate.
I
Us
controller
is
still
a
very
important
point
and
emulation,
so
the
first
rising
will
be.
We
will
have
a
we
assume
that
the
trans
pricing
we
may
have
big
challenges,
for
example
the
processing
leader
which
can
network
and
some
slices
share
the
same
physical
resources
such
as
ups
and
sister
beam
and
in
some
cases
some
several
Metro
slices
has
different
forum
forces
so
use
a
plane.
Protocol
may
be
liquid
to
conveys
such
a
reporting
a
defiant
swatting
policy
for
each
vices,
so
in
the
following.
I
So
these
to
thrive,
describes,
detail
details
of
the
feedback
and
you
can
find
original
document
from
bottom
link.
In
this
presentation,
I
will
introduce
is
a
2-2
characteristic
feedback.
First.
Is
things
about
QF
the
place
of
QA
fly
so
in
the
Alice's
work?
We
couldn't
find
the
description
about
the
order
of
Q
extension
header,
but
in
the
recent
specification
document
describes
are
noted,
for
its
recommends
recommends
that
qfi
information
should
be
contained
in
the
first
extension
header
now.
I
For
the
PD
session,
so
from
T's
feedback
now
we
could
confirm
they're
our
query
gaps
between
gtp,
you
and
5g
architecture,
and
we
believe
that
this
mark
is
in
the
right
direction.
So,
lastly,
I've
talked
about
next
steps
or
this
document
so
for
improving
this
context,
so
we
need
a
more
feedback
and
review
or
feedback
from
both
ITF
and
3gpp
side,
and
we
beat
that
this.
It
should
be
a
part
of
later
deploy
and
also
if
this
document
is,
you
think
this
document
input
are
very
important
or
useful.
I
U
Came
from
wall
again,
so
everybody
in
this
group
to
actually
read
these
analysis
and
we're
trying
actually
to
use
these
analysis
as
a
base
for
or
our
own
work
in
terms
of
you
know,
conducting
competitor
like
competitive
analysis
between
different
different
different
approaches
that
you're
talking
about
the
other
dress.
So
this
will
basically
put
a
ground
work
together
so
that
we
can
have
a
common
ground
analyzing.
All
those
different
protocols
against
against
the
same
thing.
You
know
basically
gives
us
an
apple
to
Apple
comparison.
U
So
any
any
any
solution
that
comes
to
the
picture
basically
will
be,
will
be
looked
at
from
from
this
document
perspective
and
see
first
but
ere.
We
can
satisfied
all
the
all
the
analysis
and
and
the
comparison
that
you're
we're
trying
to
make
based
on
this
before
going
forward,
two
to
three
T
V
P
we
set
up.
U
These
are
the
set
of
basically
protocols
that
we
want
to
take
to
you
guys
to
look
at
the
first
to
our
own
work,
to
make
sure
that
you
know
everything
actually
is
in
order
before
we
go
to
3gpp
with
the
list
of
protocols
to
to
for
them
to
consider.
That's
basically
the
work
that
you're
trying
to
do
you.
Thank
you.
Thank.
N
U
Harassment
here
again
from
what
we
know,
I,
don't
think
we
are
trying
to
actually
do
their
thinking
for
them.
We
are
basically
taking
their
their
requirements
and
we
are
trying
to
do
our
homework
in
terms
of
putting
a
groundwork
for
our
own
protocols
that
you're
proposing.
Let's
say
you
have
five
protocols
that
we
are
suggesting
as
a
to
be
evaluated
by
3
GPP.
This
work
is
basically
there
to
make
sure
that
these
protocols
and
approaches
basically
can
do
what
ecology
requirements.
Ask
any
other
approaches
to
actually
do
gtp
can
do
it.
B
U
U
So
background
I
guess
you
guys
are
already
know
about
this
treaty
has
issued.
The
ct4
has
issued
a
new
study.
Welcome
to
us,
there
are
references
in
there
that
you
can
take
a
look.
The
new
revision
of
this
graph
has
gone
through
many
many
reviews,
and
we
have
done
a
lot
of
work
in
the
in
the
conference,
calls
and
receive
comments.
We
receive
also
comments
from
101,
and
this
new
revision
basically
is
a
reflection
of
all
those
comments
and-
and
we
are
trying
to
to
address
all
the
issues
that
came
up
during
those
reviews.
U
Several
protocols
are
basically
being
considered
here
from
SR
v6,
all
the
way
to
Hardwick
hybrid
ICN,
which
is
a
relatively
new
one
to
d2
the
document.
The
talk
the
document
has
been
prepared
there
as
a
submission
for
consideration
of
different
protocols
by
3gpp,
and
that's
what
how
we
are
positioning
this
again
you're
not
trying
to
actually
tell
3gpp
what
to
do.
We
are
just
presenting
the
facts
and
we're
trying
to
actually
say
these
are
the
these
are
from
our
perspective.
These
are
the
contenders
that
you
might
want
to
consider
in
your
evaluation.
U
The
evaluation
is
not
even
ours,
you're
not
trying
to
actually
do
with
the
evaluation
ourselves.
We
do,
however,
come
back
consider
evaluating
these
protocols
against
the
the
previous
strap
that
Suki
and
others
are
putting
together.
So
what
are
the
focuses
of
this
new
object?
It
basically
focuses
on
5g
architecture
from
the
mobility
management.
Dr.
Roman
architecture
now
has
been
added
to
the
draft
you're,
considering
the
roaming
architecture
as
well,
and
how
these
protocols
basically
behave
in
terms
of
roaming,
support
for
different
mobility
protocols
within
the
operator
networks
and
then
slicing,
for
example.
That's
another
consideration.
U
We
have
made
in
the
new
division
again
the
data
player
architecture
model
for
m9,
because
the
study
all
right-
and
this
is
a
very
important
they
study
item-
that
cc4
issued-
is-
is
only
applicable
to
the
n9
interface,
the
interface
between
the
UPF,
so
n
3,
&,
6,
&,
3
or
4
&
2.
They
are
all
out
of
the
out
of
the
out
of
the
scope.
We
are
only
concentrating
on
the
n9
to
place
for
now.
That's
what
is
in
scope
and
we
have
the
last
one
is.
U
The
last
portion
is
a
bit
controversial
because
we
made
it.
We
made
those
classifications
with
of
the
protocols
that
debate
is
still
ongoing
and
we
might
have
to
revise
it
at
the
idea
at
the
end
of
this
reviewed
approaches.
So,
generally
speaking,
that
the
first
two,
the
first
two
actually
are,
the
are
the
problem
in
the
models.
U
There
are
two
basic
models
that
we
considered:
one
of
them
is
interworking
and
the
other
one
is
the
integrated
model
in
the
interworking
model.
The
the
idea
is
to
leave
the
3gpp
control
plane
intact
and
for
basically
will
remain
the
same
and
three
remains
the
same
and
and
and
and
four
and
six
and
will
remain
the
same
and
the
only
the
only
thing
that
we
are
changing
is
the
n9.
U
U
The
the
one
of
the
one
of
the
biggest
objective
is
probably
migration
from
existing
network
to
to
the
next,
and
this
interworking
model
is
aimed
to
to
actually
address
that
how
you
actually
go
from
the
existing
gtp
based
model
to
anything
in
the
future.
Basically,
that's
the
bird.
The
second
one
is
the
integrated
model,
which
basically
talks
about
how
ups
will
get
integrated
with
the
new
protocols
and
basically
they
we
start
going
toward
the
next
generation
of
n9
in
mobile
network.
U
The
draft
also,
as
I
mentioned,
it
has
gone
through
some
sort
of
classifications.
In
terms
of
you
know
these
protocols,
some
of
them
are
locator
based
on
them
on
lock,
ID,
and
some
of
them
are
ID
based
again,
this
is
under
debate
whether
we
want
to
actually
classify
these
protocols
as
such.
There
are
many
different
classifications,
and-
and
this
might
change
as
we
go
into
the
future
because
again
they're
something
like
SR
v6
and
lists-
can
actually
get
get
classified
as
differently
into
the
into
the
model,
and
that's
why
the
debate
is
sort
of
continues.
U
Last
but
not
least,
there
is
a
there
is
a
there's
another
draft
that
my
colleagues
will
talk
about,
which
takes
advantage
of
3gpp
own
release,
15
feature,
which
is
called
an
uplink
classifier
again.
This
is
this
is
another
important
aspect
that
we
considered
in
this
new
draft,
which
basically
allows
us
to
to
fit
the
model
into
interworking.
Essentially,
even
the
new
new
slices
in
body
start
appearing,
we
can
use
the
uplink
classifier
as
a
filter
at
the
at
the
G
note
B
to
add
the
two
to
separate
the
traffic
in
this
manner.
U
If
we
have
a
blank
canvas
slice,
we
can
basically
set
the
n9
to
a
different
approach
or
a
different
protocol
only
for
that
particular
slice
based
on
the
filter
that
we
put
place
in
the
in
the
you
SEL.
So
this
is
this.
Is
this
is
again
another
approach
toward
the
the
migration?
Basically,
which
is
in
any
network.
We
find
that
it's
the
most
important
part
you.
You
cannot
turn
on
the
switch
and
convert
the
old
network
to
the
new
one,
just
like
that.
U
U
U
P
Have
a
question:
my
name
is
Uma
I'm,
with
this
document
from
the
day
one
last
year,
December
with
Kalyani,
but
I
am
NOT
a
co-author
I'm,
not
a
contributor
in
this
document,
but
I'm
everywhere
of
all
the
progression
of
this
document,
I
mean
aware
of
is
to
be
in
the
offline
calls.
So
my
question
is
the
service.
P
X
is
good,
very
good
choice,
but
only
one
comment:
I,
like
a
survey
six,
because
people,
if
a
give
a
comment,
people
are
thinking
that
it's
not
like
that
it's
good,
but
the
problem
with
the
drop
in
gtp
drop
in
replacement.
The
problem
with
that
approach
is,
you
are
mixing
underlay
with
overlay
because
you
are
putting
PID
ripping
off
the
TA
ID
from
the
gtp
you
header
and
encoding
as
ipv6
eight.
P
The
problem
with
that
approach
is,
you
are
forcing
all
under
layers
to
be
a
service,
six
and
ipv6,
which
is
not,
which
is
not
you
intended
SR
MPLS.
This
will
radiate
a
plane
which
is
assault.
He
used
Assessor.
So
let's
not
dictate
operators
what
they
want
to
use.
I
talked
a
couple
of
operators
in
Europe
from
Xcor
Kimbrough
point
of
view.
So
there
are
lot
of
a
lot
of
existing
deployments
using
other
than
v6
ok.
So
we
need
to
be
cognizant
that
only
that
aspect.
Otherwise,
the
document
is
great
Lisa.
P
B
U
So
we
are
we
actually,
this
is.
This
is
something
I've
taught
off
with
with
the
Duomo
as
well,
and
we
will
short
consider
this
in
our
order.
Some
some
work
that
is
going
on
on
the
sidelines
to
actually
have
enjoyed
working
situation
between
SR,
v6
and
MPLS
and
other
networks
as
well,
so
they
are
coming
down
the
price
I'll
be
going
to
be
as
they
come,
and
you
will
basically
put
put
these
things
and
we
hope
the
document
accordingly.
Okay,
so.
B
U
I
want
to
be
absolutely
neutral,
absolutely
yeah.
So
again
the
essence
is
not
basically
to
to
to
favor
one
one
or
one
approach
with
the
other,
because
we
are
just
presenting
facts
and
what
can
be
done
using
different
approaches.
It's
the
3gpp
that
will
pick
up
these
these
draft
and
we'll
take
a
look
at
it
and
basically
they
can
come
back
and
say.
Okay,
probably
you
need
to
actually
give
me
some
more
information,
I'm
interested
in
this
one.
U
Please
give
you
more
information
about
about
this
approach
so
that
that's
how
you're
looking
at
this
whole
thing.
As
I
said
like
once,
we
actually
integrate
the
the
the
SR
v6
into
the
UPS,
then
we
can
actually
start
using
other
options
like
te
and
introduce
entropy
and
and,
for
example,
function
chaining
into
the
place
you
got
to
remember
the
function.
Chaining
is
something
that
we
see
in
the
IETF.
I
something
important
3
GPP
might
don't
actually
look
at
it
from
from
the
specifications
allow
for
multiple
UPF
in
the
india
standards.
U
T
U
U
U
B
U
I,
let
me
give
you
another
comment
so
today,
bi
have
we
have
a
whole
bunch
of
functions
are
actually
being
done
within
the
FCW
and
PG
double.
If
you
want
to
actually
separate
those
functions
and
put
them
in
their
own
ups,
then
you
have
to
actually
steer
traffic
to
those
duties
right.
So
that's
where
you
need
more
channels
and
that's
where
service
chaining
aspect
of
S
or
v6
come
to
the
picture,
not.
P
B
U
I
think
part
of
the
problem
is
the
terminology.
It
I
think
we
are
distinguishing.
If
you
serve,
is
shining
and
function
chaining,
so
service
rating
you're,
absolutely
right
and
six,
but
we
at
least
myself
I'm
considering
function
chaining
as
a
form
of
service
training
as
well.
So
that's
why
I
think
they
know
we
can
actually
put
that
app
to
apply
that
concept
into
the
in
mind
as
well.
Just.
W
A
quick
note
here:
service
chaining
in
a
service
if
context
is
more
like
cellulous
programming,
and
it
is
broader
concept
for
network
virtualization
and
other
things.
So
when
you
drop
cook
here
is
a
broader
context
in
terms
of
service
programming,
where
you
can
do
network
virtualization
at
wall,
which
would
be
part
of
this.
Thank.
R
You
very
quick,
very
quick
comment,
so
I
was
in
some
carts
at
the
very
beginning
of
this
work
and
I
understand
very
well.
Why
you
focus
on
in
nine
and
nine
and
six
we're
out
at
six.
So
now
you
bring
in
service
training
on
an
nine
which
I
also
think
it's
an
in
six
issue.
So
I
was
wondering
and
at
the
beginning,
I
asked
Kalani.
Why
not
in
including
n6
into
that
work,
because
if
you
so
so,
you
still
assume
the
the
PDI
session
anchor,
which
is
one
end
point
of
nine,
is
a
central
anchor.
R
So
if
you
move
that
anchor
more
and
more
to
the
edge,
you
mitigate
the
impact
of
nine,
but
you
move
the
problems
to
in
six
between
which
is
in
the
context
of
the
MFI
draft.
So
it's
related
problems.
And
now,
if
you
introduced
service
chaining,
it
applies
to
both
in
nine
and
at
six.
So
if
you
don't
introduce
it
to
the
draft,
let's
at
least
consider
it
in
the
discussion.
What
happens
if
you
move
the
PDS
session
anchor
to
the
edge
and
how
to
solve
the
problems
yam
in
six,
so
that.
U
Is
exactly
what
I
call
in
this
slide?
If
you
can
see,
I
didn't
call
it
service
chaining
at
all,
I
call
it
chaining
right
so
to
us.
It
is
to
me
it's
the
is
the
function
chaining
that
you
can
today.
If
you
look
at
the
s
gwgw
the
fact
that
these
guys
are
actually
tied
up
together,
it's
basically
create
a
chain
of
functions
right,
so
you
can
actually
separate
those
functions
and
put
them
in
different,
different
different
boxes.
U
And
basically
now
you
have
a
traffic
steering
problem
in
your
hand
that
you
have
to
actually
take
the
traffic
from
one
point
to
the
other
point
to
the
other
point,
to
the
other
point,
all
the
way
to
B
to
D
to
the
PGW
final
function.
That's
what
that's!
What
we
consider
chaining!
So
it's
not
a
service
chaining
that
n6
is,
is
trying
to
dress
it's
basically
the
the
multitude
of
functions
that
are
being
available
in
PGW
and
sgw,
and
there
is
a
way
to
actually
separate
those
those
those
functions
constraining.
U
T
So
this
habit,
Rodya
from
Cisco
I'm
gonna,
go
through
the
LOC
idea,
split
approaches
and
the
rest
of
the
document
first
list
locate
the
reservation
protocol.
I
would
like
to
clarify
something
here,
because
we
were
still
hearing
some
misunderstanding
on
this.
So
the
least
working
group
was
each
other
a
couple
of
years
ago,
and
one
of
the
main
topics
of
the
returning
was
to
separate
the
least
control
plane
from
the
least
they
deplane.
That
work
is
complete.
T
Now
the
documents
are
about
to
become
standards,
so
the
the
recipes
that
are
mentioned
and
they
under
the
Queen
at
about
become
a
standards.
Today.
When
someone
says
that
these
you
see
lives,
that
doesn't
necessarily
mean
that
there
isn't
the
least
data
plane.
So
I
said
on
the
on
the
first
one
and
they
on
the
slide.
T
Today,
you
can
use
the
least
concern
plane
to
support
many
different
data
planes,
and
actually
there
are
a
few
documents
that
have
innocently
parties
that
go
in
the
deliver
that
so
one
thing
to
notice
here
is
that
the
instrument
of
rank
has
been
around
for
over
10
years
now,
and
it's
pretty
mature
and
it's
reducing
quite
large
deployments
operational
experience
with
with
this
protocol,
and
it
tackles
mobility
aspects
deeply
in
the
in
the
in
the
protocol.
There
are
several
documents
that
go
in
detail
over
different
axis
of
a
mobility
with
list.
T
There
is
one
in
particular
on
list
for
mobile
networks
and
then
on
the
list,
data
plane.
That
has
something
you
can
use
with
the
list
control
plane,
but
is
not
the
only
option.
Basically,
discipline
do
have
hyssop
in
IP
encapsulation,
with
a
fix
16
byte
here
in
in
the
middle,
using
UTM
calculation
are
not
the
critical
stuff.
T
So
two
recent
documents
that
relevance
for
this
work
and
for
these
working
group,
the
least
control
plane
for
for
I'll,
a
highly
we'll
talk
a
bit
toward
and
the
least
canto
brief
for
a
services
that
we
have
recently
published.
So
the
last
one
is
controlling
for
SL
v6.
We
are
going
to
present
it
on
the
least
working
group
on
Thursday,
so
you
guys
are
free
to
common
commander
and
then
on
the
on
the
right
side
of
this
slide.
You
can
see
the
the
models
that
Alice
was
talking
about.
T
T
Ila
another
protocol,
it
does
others
right
instead
of
encapsulation.
There
are
no
documents
done
by
her,
but
I'm
sure
that
he
will
be
more
than
happy
to
discuss
it
with
you.
A
highly
ICN
and
you're
gonna
Luca
will
present
later
today.
Here
I
am
for
mobility.
This
is
slice,
is
really
nice
or
invite
you
to
take
a
look
in
details
on
the
runtime.
You
can
use
different
slices
to
have
all
these
approaches
persisting
if
you
wish
to
do
some
and
finally,
we
hope
in
to
get
feedback
from
you
guys.
Okay,.
D
Know
just
one
quick
clarification
for
the
3gpp
people
in
the
world
in
the
room
and
the
world.
I
guess
the
list.
Mobile
networks,
spec
is
specking.
The
N
3
+
6
+
n9
interface,
and
the
most
important
feature
is
to
get
shortest
paths
between
G
note
B's.
So
to
you
ease,
have
the
lowest
latency,
that's
what
the
focus
is
not
so
when
we
did
the
comparison
among
and
focused
on
n9.
We
just
want
to
make
it
clear
that
the
list
proposals
showing
all
those
interfaces.
B
I
So
I'll
be
wrong
again,
so
in
this
presentation,
I
will
introduce
an
approach
to
introduce
ideal
expression
architecture
into
OMG
system,
mostly
I'm,
going
to
use
the
background,
and
purpose
of
this
document
is
approach.
So,
as
we,
you
already
know,
the
ID
locks
and
separation
protocols
is
expected
to
enhance
mobility
in
mobile
network
and
we
we
are
proceeding
to
introduce
such
protocols
to
the
CP.
On
the
other.
Meanwhile,
specification
of
five
the
width
is
15
is
already
fixed
and
60p
B
may
decide
to
change
any
interfaces
so
this
in
internet
draft.
I
So
this
slide
shows
overview
and
summary
of
this
document.
So
this
approach,
I
use
is
you
are
in
crossfire,
which
is
a
fundamental
fundamental
function
of
EF
and
provide
enable
to
introduce
I
do
capture.
We
know
a
low
impact
to
the
FRG
as
specifications
and
each
this
will
we
provide
will
be
able
to
provide
a
vibration,
a
good
vibration
path
to
the
UPF
support.
Id,
look
mekinese
Netley.
I
I
I
This
figure
shows
challenges
in
the
UE
to
distributed
to
data
network
communication,
so
in
if
you
e,
moved
to
the
another
site
during
any
session.
So
you,
the
traffic,
afforded
to
the
little
function
deployed
in
the
data
network,
be
a
central
UPF.
It's
also
made
cause
I
shall
today,
so
this
proposal
provides
solution
to
use
such
additional
delay.
I
So
this
surprise
shows
a
key
concept
of
this
approach,
so
this
approach
uses
the
open,
crossfire
and
open
crossfire
type
out
just
12
traffic,
which
needs
to
be
afforded
with
Coris
path
to
the
ideal
domain
as
external
network.
So
this
is
over
B
or
with
this
architecture.
So
idyllic
node
is
deployed
between
distributed
UPF
and
data
network,
and
you
LCL
monitors
the
traffic
and
diapered
required
traffic
and
idyllic
not
receive
the
packet
and
for
that
to
the
appropriate
destination
and
a
yoke
node.
I
I
B
B
E
B
E
B
X
B
U
Couple
of
things,
so,
yes
you're,
absolutely
right.
We
have
to
actually
used
to
work
on
this
and
make
sure
you
know
we
have
a
symmetrical
path
between
the
opting
and
down
like
the
other
thing
is
that
is
alluded
to
to
yank
or
less
mobility.
I
would
say
that
you
know
for
a
man
called
less
mobility,
therefore,
more
work
to
be
done,
which
they
today
the
path
probably
is.
U
The
pathway
on
Karla's
mobility
is
very
difficult
to
actually
achieve
so
and
but
but
I
agree
that
we
have
to
actually
work
on
this
approach
and
see
how
we
can
actually
come
up
with
the
solution
to
to
to
make
the
both
paths
symmetrical
because
know
for
exactly
if
you
were
going
for.
You
are
LLC,
for
example,
that's
pretty
much
way
you
have
to
go
lightly.
C
Just
a
question:
it'll
take
the
answer
offline.
The
comment
was
made
at
3gpp
wanted
a
symmetrical
path.
My
question
is,
you
know:
is
that
a
hard
requirement?
And
if
so,
where
can
we
get
that
information,
because
that
would
really
guide
a
lot
of
us
looking
at
solutions
to
make
sure
we're
not
entertaining
anything
that
does
not
meet
that
base
requirement?
Good.
B
I
D
You
done
I
don't
want
to
interrupt.
You
I
just
want
to
make
a
comment
on
the
symmetric
comment.
Previously,
with
the
mapping
system.
You
can
get
symmetric
or
asymmetric
depending
on
what
set
of
locators.
You
register
the
mapping
system.
So
no
matter
what
the
requirement
is
I
think,
with
a
mapping
system
based
solution,
we
could
support
either
one
yep.
U
B
I
E
B
Okay,
so
few
people
I
think
not
I,
think
we
want
few
more
discussions
and
all
of
that
before
even
issued
an
option
called
but
I
think
two
key
points
or
I
think
you
know
you
got
a
lot
of
feedback.
There's
one
thing
already
right.
Second
thing:
I
think
regarding
the
LS
reply,
we
will
not
be
able
to
intervene.
You
know
refer
to
this
document
at
this
point,
because
it's
not
a
working
document.
D
B
I
think
I
think
from
all
point
of
view,
I
think
you
know
I
think
we
talked
in
the
past
as
well.
We
are
not.
We
are
going
to
I,
think
pick
up
individual
solutions
or
approaches,
and
we
are
going
to
standardize
them
right
again,
as
we
discussed
before.
We
are
not
saying
you
know
less
or
more.
You
know
whatever
like
in
ipv6.
Transitioning
is
a
great
example.
V
V
We
cannot
correct
your
papers
right,
you
trying
to
do
your
papers
here
in
a
consolidated
manner,
and
we
cannot
just
come
down
and
say
this
is
wrong
what
it
should
be
written
in
a
different
way
or
so
on
and
and
I
think
it's
also
not
beneficial
for
you
that
we
spending
in
3gpp
too
much
time
on
reviewing
your
analysis
of
our
documents
that
would
be
not
leading
to
any
goal.
So
don't
expect
too
much
from
us
on
that
part
on
alternative
solutions
to
to
Chi
TPU
that
you
have.
V
We
have
a
study
item
now,
of
course,
if
we
are
overloaded,
study
items
will
always
take
second
priority,
so
we
will
have
a
lot
of
load
in
in
release
16.
Nevertheless,
I
think
this
is
now
understood
that
this
is
important,
that
that
many
companies
are
pushing
for
that.
So
yes
bring
us
proposals
and
bring
us
different
proposals.
We
will
evaluate
them
and
we
will
most
likely
come
back
with
questions,
but
one
thing
in
general
and
that
is
really
now
set
as
sort
of
a
private
person
here.
V
If
you
want
to
get
your
proposal
to
get
more
attention
in
3gpp,
please
consider
to
come
to
ct4
and
work
there
with
us.
This
is
a
crucial
time
now,
if
you
think
your
solution
is
the
right
one
to
go,
then
please
bring
your
input
not
only
here,
but
also
come
to
CT
for
and
explain
the
background
of
your
ideas
there.
Thank
you
great.
U
Last
presentation
is
years
so
you're
running
into
your
quick,
quick,
quick,
quick
comment
on
what
Dino
was
saying,
so
maybe
we
can
actually
use
the
analysis,
strap
as
a
way
to
actually
chin
down
the
chin
down
the
list.
Maybe
so,
first
we
evaluate
it
against
what
we
did
in
terms
of
analysis,
and
then
we
saying
okay,
which
one
of
these
protocols
they
stand
their
own,
our
basically
sticks,
and
then
we
can
probably
trim
there
the
number
of
protocols
that
way.
Ok.
S
I
mean
I
was
a
big
commenter
earlier
I
just
want
to
skip
it,
which
is
that
this
solves
the
easier
problem
of
the
ankle
list,
but
I
wasn't
sure
if
anybody's
addressing
all
the
other
key
requirements
are,
which
is
really
what
the
anchor
model
provides.
The
services
of
a
quas
charging
quote
a
whole
bunch
of
stuff,
that's
happening
only.
You
move
that
out
of
the
way
who
is
actually
addressing
that
and
I.
Think
even
we
solved
this
problem.
It's
not.
S
P
Okay,
my
name
is
Zuma
I'm,
just
because
a
five
minutes
I'll
quickly
run
through
it.
I
will
let
you
read
my
draft.
This
is
put
forth
by
me
and
couple
of
folks
so
I'm
not
going
to
talk
about
the
background
we
talked
about.
You
know
multiple
drafts,
like
the
analysis,
draft,
gtp
analysis
staff
and
you
know
beta
sub
proposals.
So
we
are
talking
about
a
nine
interface
here.
So
the
background
is
for
the
nine
and
n
3
and
then
six
whatever
or
whatever
it
is.
P
So
the
thing
is
what
I'm
talking
about
here
is
so
various
SS
piece
like
3gpp,
has
defined
service
slice
types
as
indicated
here.
There
are
various
traffic
types
for
various
video
sessions
which
are
machine,
critical
or
real-time.
Today
is
only
EMV
b4
fixed
wireless
requirements
initially
for
Fiji,
but
in
future,
if
you
go
to
the
different
video
sessions
where
low,
latency
and
mission-critical
applications
are
there
today
for
transport
network
is
considered
not
not
considered
in
the
middle
either
returning
and
3
or
n9.
P
P
The
transport
and
I
will
talk
quickly
about
that
and
also
other
technologies
like
SR
v6s,
our
MPLS
or
RSVP
T.
How
this
can
be
integrated
into
the
fire
architecture.
This
approach
doesn't
replace
gtp
you.
It
works
with
gtp
U
1
in
3
interface,
and
it
works
with
any
encapsulation
approach.
You
choose
on
n
line
interface,
including
gtp
you.
P
So
this
is
the
service
based
architecture.
Various
cups
architecture,
basically
is
control,
plane
is
separated
and
data
plane
is
separated,
so
which
is
all
good
so
because
of
the
coverage
area.
P
gateways
to
be
thousand
kilometers
moved
to
hundred
kilometers
range
because
of
EPF
this
interface
a
new
problem.
So
this
new
concept
of
UPF
mobility
has
come
in
so
that
there
are
two
things
that
being
solved
there.
I
will
talk
about
SSA
mode,
1
mode,
2
mode
3
later
so.
P
But
thing
is
the
most
important
point
here
is
how
do
you
integrate
the
transport
network
in
between
between
anything,
face
online
interface?
There
are
two
ways
that
can
be
done.
One
is
discrete
approach
where
underlying
transport
network
is,
you
know
just
it's.
It's
independent
of
you
know
fiber
and
axis,
which
is
deployed
way
for
for
chase
like
4G
slices.
Basically,
we
have
only
two
barrels
they're
only
default
in
voiceover
voice
better,
so
it's
easy
there.
So
there's
not
the
hard
political
requirements
in
the
discrete
approach.
P
You
were
transport
network,
complete
links
and
node
information
it
fed
into
the
central
place,
and
it's
continuously
monitored
and
path.
Selection
happens
based
on
the
beauty
piece
or
sport.
These
GT
PhD,
once
gtp
a
packet
is
encapsulated
with
gtp
you
source
code
base
of
selection
of
the
transport
will
happen.
This
is
being
deployed
in
current
current,
a
couple
of
deployments,
including
in
US
and
UK
UK
in
couple
of
places.
P
Actually,
so
what
I'm
proposing
is
an
integrated
approach
where
this
is
not
required
for
all
the
PD
sessions,
where
this
mission
critical
and
mission-critical
real-time
sensor
sensitive
applications.
The
approach
is
you
put
the
PMF
transport
network
function
into
the
service
based
interface
and
create
two
interfaces
here.
They
I
should
say:
cell
site
router,
which
is
apart
from
G,
note
B,
which
is
front
hall.
It
can
be
in
some
places,
so
a
cell
set
router
can
be
part
of
the
G
note
P.
P
So
this
not
Bowl
interface
from
the
UPF
you'll
see
a
loop
here
for
a
branch
point
appear
for
the
anchor
PF,
the
sorbonne
interface
to
the
cell
set
router.
So
this
completely
enough,
the
advantage
of
putting
into
the
spi
is
when
the
mobility
happens.
Resources
on
the
transport
network
can
be
factored
for
some
of
the
PDA
sessions.
So
what
are
the
TNF
functionalities
I'm
looking
here,
so
the
NSS
I,
what
they
20
purified?
P
Zero
one
defined
can
get
that
80
transport
network
information,
while
making
the
dish,
including
mobility
and
X
and
mobility
or
enter
mobility,
and
for
wherever
it
is
required
for
some
of
the
slices
some
of
the
Phi
Q
eyes.
This
can
be
this
information
can
be
used
to
making
the
mobility
decisions
and
for
the
regular
non
mobility
cases
for
a
service
request.
P
I
I
don't
have
to
actual
to
ideal
case
also
whether
you
have
transport
network
resource
that
they
are
not
there,
you
can
be
verified
and
I
think
we're
out
of
time
just
a
few
final
thoughts.
Okay,
next
slide,
I
was
talking
about
just
give
me
one.
Okay,
I
was
talking
about
various
technologies.
Various
transport
technologies,
including
rsvp-te,
including
SR
MPLS,
including
SR
v6,
including
the
optimizations
PPR,
can
bring
in
have
been
discussed
in
the
summary.
This
proposal
leverages
the
existing
IETF
technologies
to
get
the
benefits
to
the
fising
some
of
the
pedia
sessions.
P
B
W
Z
Snoring,
so
in
the
meantime,
why
don't
you
start
your
presentation?
Yeah
Sheldon,
og
from
Cisco
and
in
this
presentation,
I
wanted
to
give
you
an
overview
of
what
we
proposed
to
manage
uncle
s,
mobility
for
the
use
of
hybrid
AZN,
so
I
badarian
has
been
introduced
by
a
Schmidt
in
his
to
his
presentation,
and
it
belongs
to
a
class
of
approaches
which
is
ID
based,
which
means
that
this
come
in
contrast.
Thank
you.
Z
So
this
is
done
to
new
ID
and
it
has
been
already
introduced
in
the
draft.
I
refer
in
the
slides
and
it's
a
concept
of
native
ID
oriented,
except
that
in
this
draft
there
was
no
real
implementation
and
was
only
a
wished
feature.
So
the
interest
of
this
ID
base,
the
architecture-
is
that
you
don't
have
the
need
for
that
opinion
corner
for
comfort
when
and
how
we
propose
to
do.
It
is
by
applying
ICN
principles
and
we're
interested,
especially
in
the
hybridization
proposal,
which
is
an
implementation
of
ICN
within
IP
and
Luca.
Z
What
is
interesting
is
that
I
would
ACN
is
a
request
to
apply
communication
pattern
and
you
have
a
consumer
requesting
some
data,
and
this
data
is
not
fried
by
his
name.
Isn't
fire
and
this
request
is
propagated
in
a
hop
by
hop
fashion
by
the
Reuters
using
the
feebs,
and
it
leaves
some
state
every
time
it's
forwarded
by
rotor,
and
this
state
is
used
to
forward
the
response
back
to
the
originator
and,
at
the
same
time
it
leaves
some
temporary
cache
in
the
network
route
as
it
encounters.
Z
So
thank
you.
This
request
reply
pattern.
Mobility
of
the
consumers
is
native
because
it's
simply
a
matter
of
rationing,
pending
interest
or
next
request
for
data,
so
as
to
continue
the
communication
without
an
interruption
or
the
need
to
renegotiate
society
or
to
transfer
contexts.
Producer
mobility
is
not
supported.
Natively
and
the
draft
tries
to
describe
the
mechanisms
we
use
to
support
it
so
very
shortly.
Z
They
propose
it
for
link
failures.
We
proposed
it
for
mobility
and
our
scheme
is
doing
lightweight,
but
very
process.
What
is
important
in
our
scheme
is
that
both
forward
and
mobility
operations
actually
based
on
ideas
or
names
in
the
case
of
hybridization,
we
have
no
interaction
with
the
control
plane,
no
mapping
to
contact
or
date
and
no
new
note
to
deploy
a
provision
to
manage
mobility.
So
it
should
remove
most
of
the
challenges
in
operating
mobility.
Z
So
I
won't
describe
much
more,
but
the
key
aspects
to
remember
is
that
this
scheme
is
fully
on
Karla's
and
that
mobility
is
under
layer
3,
which
makes
it
very
convenient
for
hydrologic
network
supports
in
the
draft.
You
will
find
pointers
to
some
documents,
supporting
the
performance
of
the
scheme
for
scaling
or
for
performance
plus
some
optimization.
We
do
for
latency
sensitive
traffic
in
the
rest
of
the
presentation.
Z
I
will
illustrate
some
of
the
benefits
in
the
context
of
deployments,
which
is
the
object
of
the
second
draft
we
proposed,
and
we
focused
mainly
on
the
3gpp
architecture.
So
I
won't
go
into
detail
for
free
DDP,
but
just
illustrate
how
we
benefit
from
regulation.
In
this
context,
so
so
Jordan
me
it
should
be
wrapping
it
up
a
little
bit
short
sorry.
So
the
first
benefits
they
originate
from
the
use
of
Oracle's
to
play
pattern,
and
it's
simply
the
matter
of
ratio
in
interest
and
at
the
same
time,
it's
sufficient
for
you
to
load
balance.
Z
You
request
on
many
accesses
to
benefit
from
a
native
multihoming
ability
passed.
The
second
benefits
are
originating
from
the
use
of
intermediate
caches
in
pass
from
the
rooters,
and
this
allows
low,
latency
and
native
mythical
sting
of
the
request,
because
ruto
is
able
to
answer
on
behalf
of
a
consumer.
Z
The
second
types
of
benefits
originate
from
the
native
vocalist
properties
of
the
skin,
and
we
obtained
this
when
we
use,
for
example,
in
3gpp
haan
on
in
place
of
the
gtp
tunnels,
and
this
allows
on
Karla's
mobility
and
offloading
of
each
communication
at
the
edge
of
the
network.
So
I
want
to
tell
that
so,
of
course,
both
benefits
from
HIC
NM
for
the
use
of
ID
base.
The
mobility
can
be
combined.
Z
What
we
propose,
we
have
discussion
in
the
draft
about
partial
insertion,
so
it
is
where
we
can
have
ID
even
a
subset
of
nodes
or
use
as
a
v6
in
between
each
ACN
to
extend
the
reach
of
the
proposal
will
require
chastening
endpoints,
but
we
try
to
make
the
insertion
really
minimal
through
easy
to
use
implementation
and
non
HSN.
Traffic
can
also
benefit
from
the
mobility
pattern
based
on
IDs,
so
to
complete.
I
would
say
that
we
have
proposed
a
new
paradigm
based
the
purely
on
identifier.
Z
We
bid
on
Abaddon,
it
brings
us
native
consumer
mobility
and
we
have
made
a
proposal
for
lightweight
producer
mobility
based
on
that
brain
updates.
So
we
analyzed
several
deployment
raiders
and
the
different
benefits
we
gained
and
as
a
computer
I
would
say
it's
a
general-purpose
architecture
with
nice
on
Collison
netsupport
abilities
and
it
offers
the
possibility
to
insert
the
choice
en
and
benefit
ready
for
this
new
transport
features.
Z
B
E
E
That
will
go
back
to
testing
there.
The
the
M
chord
project
that
he
mentioned
in
his
previous
presentation
today
was
an
import
project
called
c-3po,
which
is
an
LTE
4G
open-source
core,
essentially
trying
to
do
in
this
POC
is
to
establish
the
inner
working
of
an
SR
v6
underlay
for
and
I,
and
involve
the
PC
core
in
the
3gpp
standards.
E
E
So
we
ran
two
scenarios
through
the
EPC
core.
One
was
policy
based
encapsulation
of
the
ipv4
gtp
traffic
ran
into
jumbo
frame
problems,
as
you
would
expect,
and
then
we
did
the
interworking
with
ipv4
gtp
tunnels,
with
the
tid
stored
in
the
IP
at
the
SR
v6
pile
of
headers
to
get
the
interworking.
So
we
didn't
rebuild
the
header
at
the
other
end
and
both
of
those
were
successful.
E
The
software
is
available
here,
that's
the
generic
software.
This
is
where
we
would
expect
these
to
be
after
they
get
generally
released.
If
you
wish
to
go
in
and
try
this
yourself,
like
I
said
they,
both
both
scenarios
were
successful
in
testing.
Be
aware
that
the
jumbo
frames
will
cause
you
problems
if
you're
just
trying
to
do
straight
encapsulation,
but
the
ipv4
gtp
policy-based
interworking
was
successful
as
well.
E
C-3Po
code
release
we're
expecting
one
here
in
the
next
couple
days.
Actually
it's
not
718,
that's
yeah,
the
718.
Sorry,
our
next
step,
sort
of
go
back
and
and
redo
the
baseline
testing
and
try
to
push
this
towards
line
rate
just
to
see
how
the
software
performs
and
the
contributors
to
the
solution
are
listed
there.
If
you
need
any
information
or
detailed
help,
feel
free
to
reach
out
to
us.
Okay,.
U
See
we
talked
about,
it
is
another
demo
that
you're
putting
together
with
using
the
OIE
open
source
software,
so
that
one
will
be
complimentary
to
what
Mark
was
talking
about
and
their
plans
to
actually
integrate
the
two
to
two
demos
together
and
live,
even
in
the
planning
stage
of
that.
So
the
the
objective
is
again
for
the
Oei
model
to
for
the
Oei
open
source
to
actually
be
used
as
a
base
for
for
demoing
SR
v6
gateways,
as
showed
in
the
in
the
previous
presentation.
U
The
the
problem
is
that
the
Oei
basically
right
now
only
supports
the
combined
s
gwgw.
So
there's
not
really
an
annoying
interface
to
actually
do
the
work,
so
we're
basically
focusing
our
effort
to
showing
it
on
the
n3.
And
then
we
see
whether
we
can
actually
separate
the
sgw
from
the
PGW
inside
your
way
I,
and
and
that
then
we'll
follow
the
fo.
Our
v6
mobility
for
you.
U
B
Thank
You
Ashley,
thank
you,
Mark
Thanks,
so
with
dark,
I.
Think
I
think
this
is
good
work
in
general,
anything
about
implementation
deployments.
We
want
that
feedback.
I
think
this
great
stuff.
But
that
will
conclude
this
meeting
thanks.
Everyone,
yeah
I,
think
blue
sheet.
Saturation
is
one
more
Brosius
who
has
a
place
so.