►
From YouTube: IETF103-NMRG-20181107-1120
Description
NMRG meeting session at IETF103
2018/11/07 1120
https://datatracker.ietf.org/meeting/103/proceedings/
A
C
B
B
B
B
B
B
Okay,
very
good,
so
we
have
one
hour
meeting,
which
is
quite
short,
but
we
have
a
number
of
important
things
to
discuss.
So
the
agenda
is
quite
simple:
I'll
start
with
a
short
introduction,
highlighting
stages
of
the
research
group,
a
number
of
information
for
you
for
next
year
meetings
and
a
couple
of
other
stuff.
Then
we
will
have
a
session
on
some
summaries
of
the
different
progress
of
the
research
agenda
items,
namely,
intend
base
networking,
autonomic
networking
and
not
officially
use
official
intelligence
for
network
management.
B
This
meant
to
be
very
short
presentation
just
to
give
a
status
update.
We
may
have
one
or
two
questions
comments,
but
the
idea
is
really
to
take
most
of
the
time
of
this
session
to
have
a
first
discussion
on
the
future
of
energy
and
I.
Would
like
your
attention
for
for
this
discussion.
I
would
like
to
proceed
with
using
two
channels,
first
or
sure,
discussion
in
the
room
using
mic
and
what
we
have
here,
but
also,
if
at
pod
session,
so
I'd
give
you
the
link
to
the
effort.
B
But
I
would
like
everyone
willing
to
contribute
to
the
discussion
to
also
put
statement
or
whatever
ideas
on
this,
if
at
bat,
so
that
we
can
also
have
a
bit
more
input
than
just
people
going
to
the
mic
and
also
follow
people
that
will
be
removed.
It
will
be
easier
for
us
to
capture
the
different
ideas.
So
if
you
can
have
a
look
at
the
path,
it
will
be
very
useful
for
for
the
follow-up
of
the
discussion.
B
The
best
way
for
you
I
mean
thank
you,
a
very
quick
refresh
on
what
we
have
as
currently
defined
in
our
research
agenda.
This
is
not
limiting
any
activity
of
the
researcher,
but
this
is
some
activity
which
we
agreed
in
the
past
that
we
are
willing
to
work
on
that
we
have
some
plans
and
probably
some
documents,
some
work
in
progress,
so
in
turn-based,
networking,
AI
for
network
management
and
autonomic
networking
3.0.
B
This
enrolled
myself
had
published
a
bit
more
information
on
those
three
items
in
a
short
paper
in
a
knife
Rupali
standard
magazine
last
year.
So
if
you
want
to
have
a
bit
more
ideas
about
the
contents,
the
the
milestone
and
where
we
would
like
to
go
with
these
documents,
I
wish
these
themes.
You
can
go
into
this
into
this
document.
You
will
have
an
update
anyway
with
the
presentation
after
that,
in
terms
of
documents
related
to
the
research
group.
B
So
we
don't
have
current
documents
in
any
kind
of
review
or
research
group
document,
but
we
are
still
have
around
six
or
seven
depending
how
we
count
active
documents.
Individual
documents
related
to
energy,
I
will
say
all
of
them
relating
to
the
current
research
agenda,
which
is
good,
and
we
have
also
some
new
or
renewed
proposals
that
have
been
proposed
or
steel
towards
the
research
group,
and
we
still
have
to
discuss
about
those
documents,
mainly
our
own
network,
slicing,
new
approaches
to
I
mean
future
networks
and
that
kind
of
aspect.
B
This
is
just
a
proposal
for
next
year
meetings,
I
think
most
of
the
meeting
we
will
have
will
be
co-located
with
IETF
meetings.
So
this
is
a
tentative
list
of
next
row.
Meeting
there
is
one
meeting
location
in
discussion
with
I
am
2019
which
will
be
in
Washington
in
April
we
used
to.
According
to
the
the
poll
we've
made
a
couple
of
years
ago.
B
This
is
t
no
no
discussion
mainly
for
planning
aspect
in
terms
of
what
kind
of
session
we
would
like
to
have,
which
will
be
a
general
meeting
or
topic
specific
meeting.
Of
course,
what
we
would
like
to
learn
also
is
a
part
of
the
evolution
of
energy.
Is
to
renew
the
question.
The
poll
through
to
the
research
group
participants
about
the
clear
interest
to
co-locate
meetings
with
other
events,
essentially
maybe
other
scientific
conferences
or
maybe
also
industry.
B
Events
on
topics
that
are
related
or
different
to
network
management,
so
I
am
gnomes,
are
clearly
on
the
scope
of
network
management,
but
we
can
think
about
all
the
conference's
which
can
be
on,
for
instance,
a
I
machine
learning
specific
or
ughn.
So
there
are
very
different
types
of
conferences
we
could
think
about
co-locating,
but
there
is
a
need
to
add
assess
if
there
is
a
clear
interest,
actual
value
to
to
co-locate
such
events.
B
It
can
also
be
with
organizing
joint
meetings,
specific
topics
with
other
research
groups,
but
also
with
other
groups
in
order
somatization
group
development
organization
or
with
other
user
groups
like
operator
I,
know
right
type
of
meetings.
So
your
inputs
are
really
welcome.
I
will
send
an
email
to
them
in
English
to
start
a
poll
collecting
your
your
views
on
these
colocation
meetings.
B
D
Yeah,
thank
you,
okay.
So
in
fact
this
is
just
a
status
update
because
there
is
there
isn't
any
update
on
the
draft
itself.
So
there
is
initial
discussions
on
this
at
the
IETF,
100
and
101,
and
also
some
some
discussion
on
the
NMR
G
interim
meeting
at
norms
2018.
So
the
discussions
led
to
three
drafts,
the
first
one.
It's
a
draft
on
terminology
or
the
structure.
The
work
is
was
structured
as
three
drafts,
the
first
one
on
terminology,
some
definitions
and
concepts,
and
this
is
the
the
draft
that
this
presentation
is
about.
D
So
there
is
this
raft
motion
and
emerging
network
intent,
concepts
which
I
believe
it
is
inspired
right
now,
I,
think
and
and
then
this
raft
is
about
how
to
handle
the
I
intent
and
how
to
validate
networks
that
are
behaving
like
as
intended.
So
there
is
some
updates
from
the
previous
verse
of
the
draft,
especially
on
editorial
stuff
and
tightening,
and
some
added
reference
so
intent.
D
The
fun
networking
is
kind
of
a
buzzword
right
now,
so
the
basic
idea
is
to
you
know
to
define
what
you
want
from
the
not
how
to
get
it,
and
this
is
obviously
is
something
that
is
decided,
but
at
least
and
I'm
avoiding
the
rhetorical
Alex
question.
This
is
not
Neil,
so,
for
example,
in
90s
there
is
a
lot
of
investigation
on
policy
based
management
and
how
to
define,
for
example,
high-level
policies
to
assure
the
system
and
how
to
live.
D
For
example,
the
specific
knobs
that
had
to
be
turned
to
the
policy
handlers
in
the
city
networking
management
system.
Also,
there
is
some
some
relation
with
selfs
models
in
provisioning,
so,
for
example,
how
to
define
the
services
and
how
to
lead
the
Matins
of
high
level
services
to
low
level
configurations,
how
to
locate
resources
and
so
on,
and
also
there
are
lots
of
models
on
the
information
hierarchies
and
abstractions
such
as
TM,
n,
F
caps
and
other
models
that
can
you
know.
D
Can
then
they've
tried
to
address
this
kind
of
a
description
of
the
the
system.
So
there
is
some
some
point
that
we
can
make
about
the
intent,
for
example,
how
it
differs
from,
for
example,
policies
or
it's
some
some
way
to
boot,
to
reboot
policies
or
reboot
selfs
models,
or
is
an
integration
of
different
of
these
different
concepts.
D
And
if
it's
different,
how
we
can
proceed,
how
we
can
try
to.
You
know
to
address
this
kind
of
problems.
What
are
the
implications,
for
example,
different
protocols,
so
differentiating
from
service
models
which
are
instances
of
service
that
are
provided
to
customers,
for
example,
in
the
RFC?
It's
3:09.
D
We
can
discuss
about
orchestration
and
mapping
to
underlying
resources,
and
there
is
this
concept
of
machine
to
machine
interactions
and
so
tree
provisioning,
on
the
other
hand,
policy
or
more
usually,
a
set
of
rules
that,
for
example,
uses
some
kind
of
model
such
as
iike
or
even
condition
action
or
some
variation
of
that
they
are
usually
imperative.
So
they
specify
what
to
do
under
some
circumstances.
D
For
the
time
the
day
or
something
like
that
and
are,
they
are
usually
machine
to
machine
interactions
and
at
some
point
in
the
network,
it's
necessary
to
have
some
Henry
in
order
to
to
to
transform
this
high-level
abstraction
to
low-level
configuration
on
the
network
in
the
end,
intent
is
high,
declarative
policy.
So
it's
we
saying
this
that
they
send
us
draft.
That
is
a
kind
of
a
policy,
so
it's
a
declarative
in
the
in
the
sense
that
it
defines
what
is
wanted
from
the
network
and
not
how
to
get
it.
D
So
there
are
some
some
discussion
items
on
intent.
For
example,
it's
necessary
to
define
intent
narrowly
or
broadly,
it's
necessary,
some
definition
to
work
on.
So,
for
example,
we
it's
necessary
to
or
to
put
the
things
in,
a
common
contest
for
network
management
or,
for
example,
do
some
kind
of
heated
washing
and
start
again
on
and
beyond
this
concept,
and
there
is,
for
example,
in
the
aneema
documents,
there
is
different
I
say
different
kinds
of
intents,
for
example
the
operation
of
intent,
ourselves
intent
or
flow
intent.
D
But
there
are
some
people
that
say
that
it's
just
one
intent
and
at
some
point
network
should
you
know
like
divide
these
different
kinds
of
intents
in
so
this
discussion
on
the
hierarchy
of
intent,
so
intent
is
for
just
for
a
device
or
for
the
network
or
for
services.
So
this
is
something
that
needs
to
be
also
to
be
define
it
in
better
in
a
better
way.
D
So,
besides
that,
probably
it's
necessary
to
have
you
know
a
reference
architecture
in
order
to
provide
in
to
enforce
intents
also,
there
is
some
discussion
on
the
functionaries
that
intent
can
be
applied,
for
example,
configuration
or
fault
management
or
performance,
or
something
like
this
and
I
mean
there's
a
lot
of
discussion
that
yesterday
on
a
NEMA
working
group
meeting,
there's
some
some
talk
about
intent
and
how
to
proceed,
and
some
nations
to
the
NMR
G
work.
So
I
think
that
is
it
yeah.
B
Just
before
we
get
to
some
questions,
also
quick
update
at
last
I
ATF
after
the
IHF,
we
had
a
dedicated
workshop,
an
intern
base,
networking
and
in
GSM
area,
and
so
we
had
a
good
discussion.
Some
also
additional
plans
for
new
people,
and
you
knew
a
contribution
expected
and
also
in
the
meantime,
there
have
been
a
individual
contribution
and
new
draft
on
intern
classification
that
has
been
shared
with
the
group.
B
So
I
think
we
we
have
a
number
of
people
and
also
inputs
coming
in,
and
we
need
to
maybe
intensify
our
efforts
to
structure
it
and
make
it
a
bit
more
visible
inside
the
group
and
also
maybe
towards
anime
and
order
resources.
Other
groups
that
maybe
may
have
interest
to
see
this
work,
progressing
and
other
deals.
For
instance,
damn
Bogdanovich.
F
Just
interrupts
truck
very
little
about
closed-loop
bus
validation,
I
mean
all
of
intent
without
proper
validation
is
meaningless,
so
wouldn't
be
good
to
define
what
exactly
the
meaning
of
differences
Delta
between
operational
intended
state.
How
do
we
integrate
it
and
larger
graph
or
model
of
things
how
they
interact
with
each
other?
So
you
define
service
policy
and
some
other
layers.
F
They
all
interact
with
each
other
and
Dean
says,
are
sort
of
working
gene
community
on
how
to
bring
different
models
together
so
to
go
from
service
to
network
to
device
model,
but
as
you
stream,
your
telemetry
back.
But
what
does
it
really
mean?
And
this
work
clearly
belongs
here?
Coding
t5,
this
intersection
between
glass,
validation,
loop
and
intent.
D
So
even
if
this
is
it's
done
outside
and
we
can,
you
know
you
can
and
we
can
use
the
experience
from
the
from
an
engineering
part
if
you
a
I,
believe
that
we
need
a
document
in
D
in
the
effort,
F
and
also
I
think
that
it's
necessary
that
this
document
clearly
define
if
there
is
just
one
intent
of
there's,
this
division
on
different
kinds
of
intense,
so
I
need
that
this
must
be
address
in
this
document.
This,
my
okay.
B
Thank
you
if
there
is
any
I
mean
more
discussion,
recognizer
of
it
as
the
second
part
of
the
description
or
so
on
the
mailing
lists
or
in
the
corridors.
Thank
you,
I
think
Jefferson.
You
can
stay.
You
have
the
next.
One
is
also
yours.
I
know,
I
mean
argument.
I
wanted
you
just
to
make
one
shot
so,
okay,
so
thank.
D
You
so
me
again
good.
So
this
is
a
summary
of
autonomic
network,
which
is
one
of
the
topics
that
maybe
will
be
on
the
recharter
of
the
NMR
G.
So
this
is
a
recap
of
a
previous
presentation
at
an
emergency.
It
will
be
very
quick,
but
any
you
know
any
comment
we
can
give
em
the
mic
or
or
in
the
mailing
list,
so
first
defining
an
autonomic
system.
D
We
usually
talk
about
set
of
properties
which
make
system
autonomic
the
first
one,
which
is
the
the
system,
is
automatic
in
the
sense
it
can
conserve,
control
its
internal
functions
and
operations.
The
second
on
it
is
adaptive,
so
it
can
change
its
configuration
state
and
function
according
some
some
roles
and
intent,
for
example,
and
they
eat
somewhere.
So
it
can
monitor
its
operational
contest.
So
there
are
different
set
of
definitions
for
autonomic
for
internal
mix
system.
D
D
The
first
inventions
on
this
term
are
around
2005
and
2006,
so
there
are
several
intestines
for
that.
For
example,
there's
the
several
studies
of
using
artificial
intelligence
in
network
management
in
nineties,
then
around
2007
and
2009.
There
are
a
lot
of
momentum.
This
area,
for
example,
there
is
the
Amman
framework
of
Investigation.
Oh,
is
this
the
right
term
for
that
the
network
of
access
and
one
of
the
deliverables
are?
This
of
this
initiative
was
autonomic
management,
challenge
and
solution.
D
So
there
there
were
a
lot
of
you
know
like
investigation
initiatives
on
this
term
like
in
that
time,
and
then
there
is
an
a
decreasing
interest
in
around
2011
2013,
maybe
because
there
are
other
technologies
that
were
gaining
momentum
at
that
time,
for
example,
Sdn
and
NFV
in
the
contest
of
IETF
IRT
F.
There
are
several
efforts.
D
For
example,
there
were
three
autonomic
for
network
management
workshops
on
meeting
own
army
in
emerging
meetings
and
in
this
workshop,
stirrer
draft
on
gap,
analysis,
definitions,
network
operation,
wood,
strapping
detection
of
a
slave
violations
and
lessons
learning
other
stuff
like
that,
and
the
there
are
several
outcomes
of
this
energy
workshops.
For
example,
we
had
the
UK
and
both
of
the
use
cases
of
autonomic
networking
above
which
was
held
at
IAAF
90,
and
the
part
of
that
is
the
animal
working
group.
D
So
this
is
a
term
that
appeared
and
appears
in
a
lot
of
works
in
the
IDF
or
TF
in
the
case
of
the
animal
working
group,
the
first
charter
after
group,
it's
really
precise
on
the
terms
on
the
the
milestones
and
the
topics
that
should
be
at
by
the
working
group.
So,
for
example,
there
is
the
development
of
protocol
specifications
for
the
discovery
and
negotiation
of
autonomic
nodes
and
the
protocol
which
which
was
developed
before
that
is
grasped.
D
Then
there
is
also
development
of
a
bootstrapping
at
rest
infrastructure,
and
there
is
brisk
for
that
and
there
is
also
the
autonomic
control
plane,
which
is
the
ACP.
So
this
was
done
like
these,
like
using
just
a
precise
diff
in
the
final
set
of
work
items
in
order
to
you
know
avoid
boiling
the
ocean,
considering
that
autonomic
network
is
kind
of
you
know
like
a
big
thing,
and
at
that
point
the
Charter
said
that
unchartered
work
is
encouraged
as
an
individual
submission
or
an
emerging
submission.
D
So
there
is
this
connection
between
what
is
is
done
in
a
Nemo.
What
is
not
done
in
a
NEMA
and
an
emerging
submissions
and
in
November
2018
there
is
this.
There
is
this
point
to
recharter
anima
and
maybe
to
refocus,
hope
or
closed
source.
In
fact,
this
and
the
yesterday
there
there
were
some
discussions
on
that,
but
I
think
it's
not
defined
at
this
point.
What
is
what
will
be
the
next
NEMA
charter?
So
it's
open,
but
still
the
determine
it's
it's
it's
important.
It's
still
on
you
know
in
in
the
sense
it's
it's
appearing.
D
For
example,
in
the
host
speak
series
tomorrow
there
was
a
talk
on
out
on
autonomous
network,
so
you
can
say
that
there's
some
connection
on
that,
and
there
is
also
lots
of
work
about
cognitive
networks
and
intelligence,
driven
adults
and
also
intent
based
network
that
are
also
there's
some.
Some
connections
to
autonomic
networking
so
I
think
it's
it's
easy
to
say
that
this
autonomic
network
is
still
something
that
it's
interesting
or
has
interesting
networking
community
and
there
are
also
new
application
areas
for
automation,
not
working,
for
example,
programmable
infrastructures
like
Sdn
and
NFV.
D
There's
new
errors
of
application,
like
5g
IOT,
is
marked
X
and
also
and
I
think.
This
is
an
important
thing
that
it's
needed
to
have
a
better
link
with
operation
communities
like
ripe
and
the
nog
and
others
so
something
some
some.
Some
people
say
that
one
of
the
problems
of
photonic
networking
is
the
lack
of
you
use
deployments
and
the
guys
that
can
you
know
can
can
show
some
oaken
can
can.
D
D
E
E
How
will
you
manage
as
a
single
system
a
highly
distributed
system
where
each
one
of
them
should
act
autonomously
in
certain
ways,
but
as
a
as
a
unified
system
in
the
other
ways,
depending
on
from
your
perspective-
and
this
is
one
area
that
I
haven't
seen
much
research
done-
I
haven't
seen
it
addressed
within
the
ITF
and
there
is
a
step
in
between
needed
for
us
to
be
able
to
effectively
solve
some
of
the
challenges
that
you
are
proposing.
You
know
to
for
the
work
to
be
done
here.
G
Rick
Tyler
Airbus
just
to
answer
Dan
and
actually
to
answer
your
question
on
the
previous
slide.
The
deep
space
communications
community
has
a
lot
of
experience
with
autonomy
not
only
for
networking
but
also
for
sensor
platforms
and
things
like
that,
and
they
have
brought
some
work
to
the
IETF
on
how
they
manage
autonomous
devices.
C
G
Common
work
is
going
we've,
we
did
a
hot
RFC
at
the
last
IETF
about
it
and
we're
sort
of
going
around
the
groups
at
the
moment.
So
if
you
are
interested,
what
we're
doing
with
autonomous
management
for
individual
devices
and
it's
exactly
as
Dan
said,
you've
got
a
you've
got
to
have
the
individual
devices,
particularly
massively
distributed.
Networks
such
as
the
solar
system.
You've
got
of
the
devices,
you've
got
to
have
some
autonomy
and
then
the
network
autonomy
emerges
on
top
of
that.
So
if
people
are
interested
DTN
it's
tomorrow
morning,
there'll
be
another
presentation.
H
Thank
you
Rick.
You
know
from
Holly,
and
I
would
like
to
propose
another
work.
You
just
mission
in
previous
slide,
which
is
self
chop.
Actually,
your
last
idea,
I
present
it
in
a
neutral.
The
main
content
is
to
discuss
the
self
chop.
Things
I
think
people
have
some
interest
on
that.
Firstly,
we
can
come
up
with
a
taxonomy
of
the
self
chop,
whether
people
in
you
know,
communic,
you
know,
community
or
the
industry
can
have
our
consensus.
H
Okay
in
the
network
life
cycle,
they
are
maybe
three
or
four
five
aspect
of
the
self
talk
and
second
I
think
they
might
be
useful
that
we
do
some
analysis
of
the
self
chop
technologies
in
different
scenarios,
the
TC
network,
metro
network
back
or
neck
or
the
etc,
and
to
see
whether
or
some
emerging
new
technologies,
such
as
in
ten,
his
network
or
machine
learning
and
artificial
intelligence
helped
the
USF
job
to
be
applicable
in
real
network.
Thank.
D
B
I
Instead
industry
standard,
a
specification
group
again
se
ETS
I
called
a
year
and
I
experience
your
networking
intelligence
actually
last
year,
I
was
here
presenting
the
progress
of
that.
So
until
now
we
already
I
mean
studied
and
published
the
five
work
items,
and
we
have
the
first
version
white
paper.
We
have
the
use
cases
for
network
intelligence
requirement
architecture
and
we
already
started
it
through
parks.
One
is
from
China
Telecom
Mikado
from
Telecom
Italia
mobile.
I
J
Iii,
don't
see
them,
okay,
I
know,
I
know,
I
said
well,
I
am
going
to
present.
Now
is
a
brief.
I
would
be.
Very
briefly
is
the
update
after
my
representation
of
how
to
introduce
the
AI
into
this
case
would
be.
Some
evolution
is
how
to
extend
no
base
intelligence
network
management
that
we
are
considering
right
now
beyond
software
and
logical
boundaries,
please
Nate
slide
this.
J
J
B
B
J
B
J
C
G
J
J
E
J
J
I
B
B
Can
you
try
to
speak
Pedro,
no,
okay,
so,
on
the
topic
of
the
use
of
AI
for
network
management,
the
idea
was
to
consider
a
number
of
requirements
or
aspect
specific
to
that.
The
management
operation
can
be
quite
limited
in
terms
of
the
ability
of
the
administrator
or
the
management
solution
or
objects
to
have
coordinated
actions.
B
Human
operator
are
not
omnipotent,
so
they
cannot
do
and
see
everything,
but
still
there
is
also
the
management
solutions
that
are
usually
by
Percilla
or
confined
to
a
particular
domain.
So
it's
difficult
to
have
a
system-wide
or
network
wide
view
and
coordination,
and
so
actions
are
quite
separate
and
spirit
nature.
B
There
is
the
aspect
of
autonomic
network
devices
which
we
have
just
talked
about
previously,
and
but
still
they
have
to
have
some
level
of
interaction
with
I
will
say,
a
menagerie
which
can
be
a
human
or
another
system,
and
so
they
need
to
provide
the
ends
for
the
management
plan
of
the
networks
and
the
final.
The
last
part
is
the
aspect
of
objects
which
can
all
be
intelligent
to
some
extent
in
order
to
resolve
some
management
problems
in
a
distributed
way.
If
objects
can
concentrate
in
to
solve
their
own
part
of
the
program.
B
Some
of
the
requirements,
standardized
protocols
are
essentially
two
for
the
object
to
interact
freely.
Object
can
come
from
different.
Vendors
must
be
able
to
interpolate
which
outcomes
can
be
expected
from
the
results
and
bigger
opportunities
to
cooperate.
Emoji
newest
interfaces
based
on
ontology
or
semantic
approach
to
facilitate
the
object
integration
so
using
adapters.
B
Here
the
implementation
of
distributed
intelligence,
rezoning
and
in
familiar
program
resolution
can
enhance
the
behavior
of
the
network
so
to
find
sense,
empower
the
data,
information,
knowledge,
wisdom
pyramid
how
to
I
mean
at
different
stages
of
information.
We
can
reach
in
these
stages
also
to
step
out
the
single
point
of
failure
for
the
centralized
management
approaches
on
the
area
of
Tetrarch
optimization
to
make
it
becomes
a
cooperative
task
where
all
objects,
being
physical,
virtual
or
even
human
operator
can
cooperate
to
evolve
the
network
towards
its
objectives.
B
So
it's
really
a
collaboration
between
the
different
entities
involved
in
the
network,
behavioral
network
management,
but
also
analysis
and
recognition
of
different
traffic
patterns
which
can
enhance
the
distributed,
intelligent
and
then
spider
distributed.
Intelligence
ability
of
evolution
is
increased,
more
object
but
less
complexity,
but
also
a
property
of
self-healing
that
can
become
a
widespread
reality.
B
Okay,
so
sorry,
for
being
a
only
reading,
the
slides,
I
think
Pedro
have
much
more
knowledge
about
how
you
would
like
this
to
be
presented
just
to
make
a
final.
Maybe
concluding
remark
on
this
use
of
different
AI
techniques
for
network
management
is
an
area
we
think
our
is
in
scope
of
our
research
agenda.
B
One
of
the
challenge
in
this
area
is
that
many
many
activities
going
on
in
artificial
intelligence
or
machine
learning
techniques,
especially
different
types
of
Martian
learning
techniques
which
are
not
already
suited
to
network
management
tasks
on
network
characteristics,
so
there
is
I
think
our
first
level
of
discussion
to
have
which
is
in
that
space.
What
clearly?
What's
precisely
we
would
like
to
investigate,
which
is
specific
to
network
and
network
management?
B
For
me,
it
will
be
of
low
value
to
just
allow
any
kind
of
AI
or
machine
learning
techniques
to
be
presented
in
this
research
group.
This
is
not
the
idea.
The
idea
is
really
to
understand
the
relationship
between
those
existing
or
new
techniques,
new
approaches
for
the
evolution
and
the
future
of
network
management.
I
will
close
the
discussion
here
for
that,
because
I
think
we
have
time
now
to
address
the
discussion.
G
A
retainer
again
just
on
the
last
presentation,
I
think
it
was
really
good.
It
contained
lots
of
really
good
points,
but
I
think
all
of
us
in
this
room
with
an
interest
and
all
the
other
SDO
is
outside
the
RTF
IETF.
We
need
to
actually
answer
some
of
these
questions.
I
I,
it's
great
that
we
developing
a
good
texana
me
in
which
we
can
base
documents.
It's
great
that
we're
highlighting
what
the
problem
space
is
and
I
understand.
This
work
needs
to
be
done,
but
we
have
to
get
on
and
do
this
now
it's
it's.
B
I
My
point
is
very
simple:
in
addition
to
this
gentleman's
comments
actually
I
like
the
the
lastest
slides,
but,
however
I
think
there
is
one
missing
point,
because
this
trend
of
AI
or
intelligence
is
based
on
the
data,
the
Big
Data
technology.
However,
in
network
of
weather
area,
so
we
cannot
collect
the
enough
data
and
I'm.
Seeing
that
you
know
HF,
there
are
some
fun
groups
are
working
on
this
elementary
or
related
work.
I
B
Okay,
are
we
go
through
this
slide
Jeff,
because
we
don't
have
a
lot
of
time
and
I
really
would
like
to
give
the
floor
to
the
room
to
address
some
of
the
discussion
we
would
like
to
happen.
It
will
be
an
ongoing
process,
but
for
the
future
of
energy
we
are
still
in
the
selection
process
of
an
additional
culture.
There
have
been
some
issues
for
that,
but
hopefully
soon
released
an
announcement
for
this.
B
A
couple
of
slide
just
to
highlight
how
I
see
this
evolution
of
the
energy,
so
some
important
question
and
principles.
We
need
to
go
through
again.
If
you
want
to
approach
this
the
right
way.
What
is
our
shared
goal?
We
are
a
research
group
not
with
not
only
research
individuals,
so
we
need
to
understand
and
maybe
define
more
precisely.
What
is
the
collective
outcome
we
would
like
to
achieve
with
the
research
group?
Okay,
as
I
said
just
presenting
work
and
writing
draft
is
not
a
goal.
B
Okay,
so
also
who
is
the
public
and
which
are
the
communities
that
are
involved
or
that
should
be
involved
or
that
we
would
like
to
target
it's
important
to
understand.
If
we
target
service
provider
end-users,
the
research
community,
the
industry
community,
the
standardization
community
so
to
understand
our
goal,
our
objectives
where
we
want
to
deliver
so
that,
because
what
we
will
deliver
can
be
quite
different
depending
on
who
is
the
public,
so
we
have
to
define
our
scope
with
respect
to
this
question
and
also
who
will
be
benefit
from
the
outcome
of
the
research
group.
B
Also,
what
is
our
approach
our
way
of
working?
So
energy
is
a
quite
old
research
group.
We
have
been
generation
of
different
researchers
coming
here,
but
we
can
also
think
about
how
we
want
to
work.
I
mean
a
research
group
is
not
a
working
group.
We
don't
have
the
same
constraints.
We
are
pretty
free
to
make
our
meetings
or
work.
The
way
we
want
I
mean
a
draft
is
only
a
tool
for
IETF.
We
can
use
it
because
we
are
associated
with
isodd,
but
we
are
not.
B
This
is
not
prescriptive,
so
we
can
work
our
own
way
and
we
have
to
maybe
discuss
this
for
for
the
future
of
energy,
also
very
important.
What
is
the
link
of
this
research
group
is
IETF
I?
Think
it's
important
that
we
can
benefit
from
the
colocation
of
this
of
these
groups
to
benefit,
and
maybe
impact
I
have
some
influence
towards
IETF.
I
will
just
go
through
the
slide
and
animation
of
the
decision.
B
B
Also,
as
it
was
say
several
times,
we
can
use
the
knowledge
and
the
inputs
from
the
different
groups,
but
also
we
can
provide
knowledge
to
other
groups.
So
this
is
very
important
to
enrich
our
understandings,
broaden
our
views
in
order
to
address
the
right
problems.
The
right
way,
a
bit
of
the
process.
We
can
be
more
or
less
disruptive
or
incremental,
with
the
evolution
of
the
energy
I
know,
I
use
the
term
chattering,
but
we
don't
have
to
formally
write
a
new
charter.
This
is
one
of
the
possible
outcome.
It
can
be
just
redefined.
B
New
research
themes,
a
new
research
item,
so
we
have
our
the
gradient
can
be
whatever
we
decide.
There
is
no
urgency.
We
need
to
take
time
to
make
sure
we
make
the
right
decision
and
we
can
still
work
while
we
are
progressing
to
be
defined
at
the
scope
of
energy,
so
we
can
take
up
to
a
next
year.
Meetings.
B
So,
in
terms
of
organization,
do
we
want
to
concentrate
on
a
set
of
hot
topics,
and
we
make
like
specialized
meeting
on
that
or
we
want
to
have
more
research
challenges
and
cover
them
in
a
more
sustained
way,
but
also
very
important,
I
think
to
capture
the
discussion
we
will
have
for
the
coming
months.
I
mean
one
one
tool
could
be
the
wiki,
maybe
not
the
best
interactive
way,
but
at
least
I
would
like
that
we
don't
lose.
B
What
will
be
our
discussion
so
we
have
the
pad
already
today
we
have
the
meaningless,
but
we
have
also
may
be
a
way
to
centralize
a
bit
these
information
so
that
everyone
can
also
improve
and
refer
to
that.
I
will
not
go
through
that,
but
this
is
just
highlight
some
ideas.
I
have
on
my
site
about
potential
topics
for
discussion
for
a
bit
provocative
some
time
about
things
we
can
investigate
in
this
in
this
group.
But
now
the
floor
is
also
for
for
the
room
and
the
people
online.
Also,
they.
G
No
wait
a
second
in
so
originally
jumping
a
line
do
not
jump
in
the
line,
but
jumping
on
the
end
of
Dan's
question.
I.
Don't
think
that
is
a
valid
question.
This
is
a
research
group.
It
should
always
be
looking
at
both
the
near
time
near
future
and
its
aspirationally
into
the
distance
trying
to
see
you
know
what
is
coming
out
of
the
really
cutting
edge
academia
way
before
it
hits
even
the
edges
of
what
industry
is
looking
at,
but
also
looking
at
the
demands
that
are
coming
from
industry.
G
You
know,
we've
done
the
SNMP
thing,
we've
sort
of
done
the
net
comfrey
thing,
and
now
we're
looking
at
the
next
standardized
management
protocols
for
various
areas.
The
IRT
F
working
groups
I
feel
work
best
when
they
are
seeing
what
all
the
IETF
groups
are
doing
and
saying
that's
nice.
But
are
you
aware
they're
doing
that
over
there,
and
this
plays
into
a
bigger
picture
on
what's
happening
in
the
research
community
and
that
sort
of
steering
and
guidance
and
coordination
I
think
is
a
real
benefit
for
an
IRT
F
group
and
the
animal
manage.
E
You
mentioned
the
important
linkage
between
the
ITF
and
the
IRT
F
and
I
RT
F,
making
the
ITF
aware
of
certain
things.
The
problem
there
that
sometimes
we
tend
to
jump
over
certain
fields
and
not
pay
attention.
There
is
a
mediums
that
is
not
being
covered
and
then,
although
the
research
community
was
aware
of
those
research,
it
wasn't
sexy
enough
in
order
to
do
it
and
it
just
gets,
and
then
it
comes
to
the
ITF.
E
G
Taylor
again,
risks
responding
to
that
I
completely
agree,
but
I
think
it's
the
IRT
F
working
groups
responsibility
to
balance
those
two
things,
because
yeah
research
is
looking
at
the
next
thing.
Always
the
next
thing-
and
that
is
varies
the
cart
before
the
horse,
sometimes
but
I-
think
the
IRT
F
can
straddle
that
can
have
a
foot
in.
G
F
So
coming
from
ATF
side
of
things,
if
you
jumped
in
you
think
before
solving
all
think,
eventually
lose
credibility.
So
it's
it's
cool
to
jump
in.
You
think
you
know
we're
going
to
be
artificial
intelligence,
offer
our
block
chains
and
solve
world
hunger,
but
there's
some
stuff
to
be
solved.
And
if
you
look
at
this
year,
specifically,
it's
not
efficient
money
present.
There's
a
lot
of
intent,
driven
interesting
stuff.
F
People
are
working
on
so
there's
something
to
learn
from
the
industry
and
trying
to
defy
problem
they're
trying
to
solve,
but
there's
a
lot
of
problem
that
are
unsolved.
That
still
require
a
lot
of
hard
thinking.
You
know
under
stable
work
so
rather
than
jumping
20
years
from
now,
let's
again
try
to
reconvene
what
ITF
is
doing
and
help
them
to,
if
not
coordinated,
at
least
working
on
top,
because
research
group
RTF
is
looking
a
bit
ahead,
but
there
should
be
tangible
outcomes
and
eventual
handover
to
engineering
to
build
the
stuff.
Otherwise,.
B
I
you
Jeff
and
I'm
speaking
as
an
individual
I
think
this
is
one
of
the
question
we
should
address,
not
maybe
at
only
research
group
level,
but
maybe
RTF
or
relationship
between
area
ths,
the
type
of
research
and
with
the
anticipation
we
need
to
have
I,
think
they're,
very
diverse
view
about
this.
I
understand
your
view,
but,
for
instance,
I
mean
for
the
last
year's
in
IOT
F.
The
message
was
quite
clear:
that
IOT
F
is
not
meant
to
do
kind
of
present
at
ization.
It's
meant
to
do
research,
Advanced,
Research,
the
excellence
of
research.
B
So
you
see
it's
very
tight,
the
different
direction
of
what
you
are
mentioning
that
our
current
problem,
that
maybe
our
gf
cannot
see
pate
and
provide
I
mean
inputs
to
the
working
group.
So
it's
we
need
to
reconcile
those
different
viewing
and
choose
for
our
research
group.
What
are
the
tough?
Maybe
we
can
combine
I
mean
have
two
different
levels:
I,
don't
know
if
you
have
the
means
to
do
that,
but
sure
right.
E
Speaking
as
individual
and
NomCom
2018,
we
asked
many
candidates
about
ITF,
IRT
of
linkage
and
I
RTF.
Raising
awareness
of
some
work
in
the
academia
that
is
useful
to
the
problems
is
very
helpful
to
the
ITF
so
that
the
ITF
will
not
be
surprised
with
the
emergence
of
a
technology
that
suddenly
becomes
a
private
standard
within
the
industry,
and
we
had
you
know
essentially
no
no
input
to
help
define
it,
and
you
end
up
with
a
very
fractioned
ecosystem
that
is
causing
the
problems
to
the
underlying
industry.
To
deliver
the
needed
services
to
them.
K
And
pollution
to
the
to
stress
on
to
comments
that
then
made
I
myself,
I
am
involved
with,
for
example,
the
whole
streaming
telemetry
and
I
feel
the
paying
a
lot
right.
So
precisely
what
Dan
was
saying,
so
it
looks
like
that.
A
private
you
know
initiative
was
fostering
on
one
side
and
I
feel
that,
for
example,
I
just
was
totally
unprepared
to
build
a
counter
plan
or
a
plan
for
something
that
is
open
and
shared
by
the
community,
and
things
like
that.
G
Rick
Taylor
again
I
totally
agree
with
then
I
thought
that
was
a
really
nice
summary,
but
I
think
the
balance
of
that
going
back
to
your
question
of
is
the
IR
TF
about
keeping
the
IETF
aware
of
what's
going
on
in
research,
or
is
it
about
fostering
research
I
think
we're?
Well,
you
are
all
pretty
smart
guys.
Why
can't
you
do
both.
B
So
now
I
comment
a
bit
from
my
recent
experiences
as
as
a
chair,
because
I
was
also
a
researcher
I
contributed
to
an
emergency
order.
Research
group
I
think
my
experience
is
that
the
topic
is
very
important.
I
mean
the
topics
we
are
investigating,
of
course
very
important,
but
there
is
no
miracle.
We
need
the
resources
to
address
this,
and
most
of
the
time
is
not
that
only
the
topic
is
enough.
We
need
to
have
the
right
resources
at
the
right
moment.
I
think
just
an
example
about
autonomic
networking.
B
This
was
research
for
the
past
10
15
years
sometimes,
but
there
have
been
a
kind
of
several
people
coming
from
different
backgrounds
at
the
same
time,
making
some
meetings
and
presentation
in
energy.
So
there
was
this
kind
of
coordination
of
effort,
and
then
it
created
some
both
and
eventually
becomes
a
working
group
and
I.
Think
if
you
don't
have
this
coordination
of
effort,
I
mean
people
already
willing
to
contribute
on
a
sustained
way
more
than
one
people
and
one
on
one
person.
Sorry,
then
we
can
achieve
in.
B
G
Again
now
this
is
with
my
IETF
working
group.
Chair
hat
on
you
can
ask
us
to
ask
our
working
groups
and
say:
do
you
know
what
the
the
IRT
F
n
emoji
or
any
other
research
group
they're
doing
some
cool
stuff
that
you
should
probably
be
aware
of
go
jump
on
their
mailing
list?
Go
go.
You
know,
everyone
has
a
spare
to
use
d.
Hopefully
you
know
the
the
meeting
on
Tuesday
morning
go.
Go
see
what
they're
doing
see.
If
you
can
help
I
mean
you
can
ask
us
IETF
working
groups
to
to
chip
in.
B
B
E
Unlocked
on
which
some
ideas
and
topics
PostNet
confit
on
the
era
that
should
have
been
done
a
long
time
ago,
because
this
is
a
10th
year
old
concept.
There
is
now
being
standardized
and
we
are
realizing
that,
as
we
are
building
the
working
systems
that
those
tools,
those
those
architectures-
are
really
not
addressing.
What
we
need
to
move
forward
that
you
know
what
we
will
need
for
autonomic
networks
to
build.
We
cannot
use
those
tools
today
and
we
mean.
I
I'm
will
so,
in
addition
to
rake
comments,
actually
I
think
there
are
two
points,
a
key
that
our
I
mean
this
research
group.
We
need
to
think
about.
So
why
is
about
a
continuing
T
so
as
as
we
already
done
some
work
previously,
so
we
still
need
to
continue
this
direction.
For
example,
the
existing
I
mean
our
draft
in
this
research
group,
such
as
our
intent
of
the
other
related
work.
The
second
day
is
about
a
creative.
I
So
if
you
look
at
all
the
research
groups
in
RTF,
so
some
of
the
research
group
are
closed
because
either
they
are
lack
of
creative
or
it's
already
implementing.
So
we
don't
need
to
do
research
anymore,
so
on
I.
Think
the
what
Lauren
has
done,
especially
for
those
workshops,
is
very
good
to
trigger
the
creative
creation
for
energy
to
make
this
user
group
for
another,
maybe
five
years
to
work
on
those
hot
topics.
So
on
this
point,
I
would
suggest
that
yeah
we
put
this
into
the
consideration.
Thank
you.
Okay.
Thank.
B
B
B
I
mean
it's
it's
nice
to
have
a
state
where
we
discuss
and
explore
new
new
approaches,
but
we
also
need
to
concrete
it,
make
it
clear
what
is
our
it
is
our
plan
and
then
implement
the
plan,
even
if
it's
a
research
group,
so
I
try
to
stir
it
this
this
transformation.
But
it's
a
research
group
again,
it's
a
collective
approach,
so
your
inputs,
your
discussion,
your
view
are
very
important.
Today
we
collect
it
some
of
them
it
will
continue
on
the
minimalists
or
on
the
wiki.