►
From YouTube: T2TRG Interim Meeting, 2019-11-15 (Part 2 of 2)
Description
T2TRG Interim Meeting, 2019-11-15 (Part 2 of 2)
A
B
C
D
E
B
MatÃas
Kovich
and
I'm
now
working
on
some
new
topics,
so
the
general
background
is
industrial
IOT,
where
we
have
kind
of
different
settings
in
the
wireless
that
we
see
in
the
home.
Bom
technology
that
are
not
working
with
is
so
called
time.
Sensitive
networking,
that's
a
group
of
standards,
that's
defined
by
the
I
Triple
E,
it's
somehow
related
to
debt
net,
that
net
is
kind
of
little
mini
Signet
working
at
layer,
3
and
there's
also
deterministic
networking
at
layer
2.
B
That's
this
TSM
group
of
standards
and
another
background
is
that
your
PC
foundation
is
having
a
new
activity
called
the
heat
level
communications
and,
basically,
so
far
they
have
been
working
on
the
management
levels.
So
if
you
have
your
controller,
how
to
get
the
gnostics,
how
to
configure
it
and
so
on,
this
could
be
done
vendor-independent
through
OPC
UA,
but
we
loaded.
Usually,
you
had
the
typical
few
classes
like
coffee,
net,
EtherCAT
and
so
on.
So
there
you
had
still
a
whole
bunch
of
different
systems
that
are
not
interoperable.
B
Have
basically,
this
vendor
lock-in
connected
to
it
here
is
that
OPC
UA
can
also
become
a
cross
vendor
standard.
That
goes
down
to
the
field
level,
so
talking
to
devices
connecting
controllers,
industrial
controllers,
between
different
vendors
and
so
on,
and,
of
course,
for
industrial
networks.
You
want
to
have
this
even
deterministic
duty,
and
so
they
based
the
real-time
performance
and
so
on
on
fuse,
and
so
we
now
have
two
new
ecosystems
TSM
on
top
of
that
OPC
UA
and
this
kind
of
gives
you
been
going,
inland
stack
for
industrial
IOT
and
overall.
B
B
Yeah
this
was
the
background
just
here
small
chart.
Most
of
the
communication
in
industry
is
still
wired
and
they
want
to
really
commit
there's
only
a
small
share
of
really
six
percent
Wireless,
it's
going,
but
still
the
rest
is
connected.
So
that's
why
TSN
with
Ethernet
or
basically
finit
is
an
important
target,
and
one
thing
is:
you
have
to
configure
this
connectivity
I
already
talked
about
this
excel
see
the
future
of
the
communication
activity
so
before
it
was
going
up
to
the
management.
B
Now
it
will
be
going
down
to
eight
devices
and
what
I
have
been
doing
in
this
project.
Rather
student
that
is
currently
visiting
Luke
askew
from
University
of
Bologna
how
to
describe
this
waves
in
descriptions
and
then
how
to
combine
this
with
the
network
configuration
to
give
you
an
idea
how
OPC
UA
looks
likes
on
the
left
is
a
feature
from
a
true
that
can
speak
OPC
UA.
You
have
multiple
notes
here.
They
are
addressable
and
with
rather
complex
you,
so
you
have
a
new
space
that
you
have
to
provide.
B
B
It's
like
an
object,
it
could
be
a
bearable
and
but
even
methods
are
type
like
this,
so
you
have
the
strong
typing
in
OPC
UA
and
another
thing
is
that
inspired
by
the
web,
but
it
made
everything
more
complex
again,
so
the
back
was
successful
because
it
was
so
simple.
They
thought
it's
better
if
you
have
more
options
and
make
it
topics
again,
so
they
are
bi-directional
references
which
was
actually
by
all
the
other
hyper
media
systems
died
out
there
and
the
web
succeeded
and
we
have
to
deal
with
bitmaps.
B
B
So
OPC
UA
hasn't
fight
anything
with
Gianna
in
their
own
documents.
They
always
use
OPC,
TCP
or
UDP,
and
but
that
then
only
has
a
hostname
and
port
to
identified
the
entry
point
to
an
OPC
ua
server
and
from
the
air,
and
basically
nothing
is
defined
one
thing
and
it
actually
comes
from
or
comes
human.
We
want
to
do
this
mapping
and
we
have
to
find
out
what
fits
together
and
the
last
time,
you'll
see
that
you
have
to
have
a
combination
of
server
address.
B
The
name
space
enter
node
ID,
and
this
is
something
that
we
can
can
put
in
there
later.
I
will
talk
a
bit
about
what
the
different
options
we
have
and
so
on,
and
otherwise
I
mean
from
from
a
high
level.
It
looks
like
quite
well
so
if
you
take
it,
OPC
UA
server,
plus
some
so-called
pops,
a
functionality.
That
would
be
something
that
corresponds
toward
the
Serbians
in
developing
India.
Tell
Celia
something
now,
if
I'm
called
functional
entity
that
pretty
much
corresponds
to
a
feeling
in
ourselves.
B
B
The
apparatus
system
I
already
use
that
is
n,
stand
up
speech,
so
they
are
not
of
these
ugly
radiations.
It's
all
built
around
the
I,
create
on
the
2.1
hue
standard.
So
that's
basically
the
standard
that
defines
Elance
and
switching
so
bridges,
endpoints
and
so
on,
and
basically
the
villain
header
is
used
to
encode
priority
and
some
additional
information
to
manage
this.
Getting
any
sleep
tracker.
B
So
how
does
it
work
so
the
blue
boxes
are
switches
here.
You
have
a
bunch
of
different
ink
nodes
basic
capability,
but
we
are
currently
using
is,
if
you
have
any
vices
that
are
not
TSN
aware,
you
can
still
reserve
bank
wave
between
the
switches,
and
this
is
only
Ethernet
cable.
You
have
one
application
here.
There
is
no
interference,
there
are
no
other
applications
talking
or
stealing
bandwidth
and
you'll
find
here
and
then
almost
twice
where
you
have
to
see.
C
A
B
So
I
mean
the
point
that
the
situation's
moment
is
you
don't
find
on
the
market
he's
an
enable
endpoints.
Yet
so
mostly
it's
prototypes
that
we
find
often
where
you
can
still
FPGA
based.
If
you
haven't
use
an
enable
a
device,
they
can
do
way
more.
They
again,
basically
sync
the
application
to
the
network
access
and
you
can
have
a
real
deadline
guarantee.
So
you
can
also
define
how
much
should
be
the
gt-r
that
your
message
arrives
early
or
late.
B
B
It's
different
queues
to
store
these
packets.
Hugh
belongs
to
a
certain
priority.
Let
us
define
the
VLAN
headers
and
therefore
they
are
eight
and
you
can
assign
then
specific
traffic.
It
was
specific
you
and
what
you
then
can
also
do
listed
is
fine
certain
kind
sense
when
the
gauge,
what
which
should
be
open
and
closed,
and
this
way
basically,
you
get
something
indicated
the
bottle
that
you
have
some
kind
of
timing,
dimensional
multiplex
to
access
the
in
Ethernet
medium
and
one
thing
is:
either
you
divide
it
by
traffic
classes.
B
So
that
means
that
multiple
friends
from
multiple
applications
get
one
reserved
window.
You
can
also
do
something
called
frame:
isolation
where
then,
basically,
a
single
application
or
a
single
frame
gets
its
own
window
and
you'll
be
delivered
in
detail.
That,
of
course,
is
then
also
getting
a
bit
more
complex.
If
you
have
multiple
switches,
the
basically
have
to
grow
with
this
time
window,
because
there's
always
some
kind
of
gt-r,
so
the
state
of
the
artists
that
he
proper
pack
them
together.
Overall,
you
get
the
better
performance.
B
No,
it
says
time
funding
sky,
so
at
the
moment
I
mean
different
things.
So
this
all
started
with
audio/video
bridging
where'd
was
basically
traffic
shaping
with
the
so-called
30-page
paper
and
nowadays,
a
time
away.
Shaper,
where
you
basically
use
the
precision
time
protocol
to
sync
all
your
nodes
and
then
have
the
time
and
multiplexing
to
reserve.
B
So
yeah,
so
then,
the
main
point
here
is
that
it's
actually
still
allowing
best-effort
traffic.
So
if
you
have
these
nodes
and
your
IT
nodes
and
so
on,
they
can
snowshoe
at
the
same
table
so
same
infrastructure,
so
because
of
that
TSN
is
also
technology
that
allows
you
to
really
converge
your
network,
so
that
all
you
know,
IT
infrastructure
in
a
factory
for
instance,
can
be
used
to
save
tables.
Is
you
control
infrastructure.
B
B
Okay
and
to
manage
the
these
networks
at
layer,
two,
there
are
different
approaches.
The
one
I'm
showing
is
a
centralized
model,
it's
also
a
disputed
model
using
different
protocols
and
then,
of
course,
a
hybrid
model
that
combines
these
two
approaches.
What
has
to
be
done?
Basically,
that
you
have
some
entity
that
can
talk
to
your
applications
and
can
connect
basically
the
requirements,
the
quality
of
service
parameters
of
the
applications
and
they
talked
with
networking
configurator.
So
the
network
controller
that
then,
basically
translates
the
quality
of
service
requirements
to
a
certain
schedule.
B
So
these
gate
control
lists
that
I've
been
indicating
to
reserve
resources
on
these
switches
and
his
central
controller
is
then
party,
usually
using
net
cons
to
each
devices,
and
there
would
be
other
options
like
response
currently
working
on
poor
cons
and
but
for
some
reason,
I
found
that
net
conflicts
kind
of
the
de
facto
standard.
What
is
deployed
on
this
kind
of
network
equipment
rest
compared
only
seen
as
Sdn
controllers
northbound
to
configure
this
and,
in
short,
basically
what
we
also
need.
B
Here
the
mapping
is
also
possible,
so
we
had
proof
of
concept
implemented.
Luca
did
did
all
the
work
here,
and
so,
if
you
have
the
oval
stack
on
your
embedded
device,
that
could
be
the
serviette
and
we
have
been
running
network
server
that
has
multiple
data
stores.
This
could
be
identified
as
the
field
the
leaf
nodes
in
this
model.
So
usually
it's
any
model
that
30ft
thought
there
would
be.
The
properties
yang
also
supports
our
pcs.
B
Those
can
be
maps
to
actions,
and
you
can
also
have
notifications
in
this
model
and
they
are
50
banks
and
basically,
you
have
an
XPath,
because
it's
XML
based
Netcom
each
issue.
This
data
store
describing
nodes.
So
this
is
what
we
didn't
take,
basically
for
the
each
wave
part
to
build
an
net
conf
you
or
I,
to
address
individual
parameters.
B
This
is
how
many
messages
look
like.
As
I
mentioned,
it's
xml-based
there's
a
lot.
You
can
ignore,
see
a
node
that
identifies
what
is
the
method
that
you
want
to
apply
as
get
yes,
edit
conflict,
and
then
you
can
define
custom
RTC
is
to
do
something
on
on
the
data
and
there's
some
multiple
choice
is
actually
one
ways,
basically
nested
nodes
that
correspond
to
the
extract
to
describe
what
you
want
to
change
and
the
same
thing
is
known
and
then
have
the
actual
configuration.
B
This
then
leads
to
a
number
of
design
decisions
that
we
had
to
make
for
the
nine
days.
I
think
I
won't
go
into
detail
here.
It's
honest
if
you're
interested
the
main
takeaway
here,
so
you
have
another
option,
see
I
marked
some
of
the
not
so
good
choices
in
red.
We
did
the
same.
Basically
for
this,
you
a
if
you
have
these
existing
ecosystems
that
are
say
built
in
a
different
time.
B
There's
different
ideas,
different
mechanisms,
it's
getting
quite
hard
to
apply
a
lot
of
things
and
I
think
this
is
actually
an
exercise
that
has
to
be
done
that
we
look
more
into
these
existing
ecosystems
and
concentrate
more
on
the
complicated
cases.
To
really
understand.
Do
we
have
everything
covered.
Is
there
something
missing
and
there
are
always
choices
being
basically
drop?
A
few
things.
B
B
This
this
so
PD
starting
point
was
at
the
moment
they're
all
different
kind
of
vendor
specific
protocols,
and
there
are
some
convergence
going
on
and
the
main
convergence
that
we
can
see
at
the
moment
is
that
OPC
that
has
really
a
broad
base.
A
lot
of
vendors
are
in
there.
He
has
now
specifying
something
that
can
work
at
the
field
level
and
to
have
deterministic
behavior
at
the
field
level
they
built
on
top
of
his
n.
So
to
get
basically
the
full
step
you
need
for
industrial
applications.
C
A
B
Actually
there
is
I
didn't
want
to
go
into
too
much
detail
this
step
in
between
which
is
the
IEC
I
Triple
E
68
of
two
working
group.
They
are
defining
an
industrial
proof
for
PSN
within
that
you
have
different
kind
of
classes,
and
but
this
is
basically
the
definition.
What
will
be
their
parameters,
the
selections
and
so
on?
What
an
industrialization
should
be
able
to
configure
at
a
network
level.
B
It's
a
really
configuration
so
basically,
usually
you
buy
it
you'll
see.
So
it's
it's
multi-layered
again.
So
I
guess
we
are
going
in
a
study,
different
correct
direction,
but
I
think
it's
what's
interesting
to
look
into
industrial
T,
so
usually
devices
you
can
buy
off
the
shelf
and
you
just
plug
them
in
and
then
they
get
minimal
configuration,
but
you
don't
have
to
do
so
much
controllers.
B
On
the
upper
hand,
you
usually
need
an
engineering
to
away
then
in
basically
install
a
software
or
model
the
software
that
should
be
running
there,
so
that
is
the
really
dynamic
and
the
main
is
then
going
on
in
the
HD
a
say
which
controller
should
control
at
other
controllers,
which
controller
takes
care
of
which
device
and
so
on.
So
you
get
a
lot
of
these
support
application
relationships
when
you
apply
these
kind
of
off-the-shelf
components
to
produce
either
a
car,
bread
or
whatsoever.
B
So
it's
something
that
overall
is
probably
done
once
the
design
time
of
a
factory
and
but
what
you
see
nowadays
is
that
it
becomes
more
and
more
dynamic,
so
exercise.
One
is
one
key
word
here.
Other
things
are
that
you
want
to
evolve
more
easily
and
reconfigure
devices
there.
For
instance,
you
have
some
predictive
or
preventive
maintenance.
B
A
B
I
mean
this
is
a
side
project,
so
it's
a
visiting
researcher
doing
this
currently
in
our
lab
the
working
on
a
paper
on
this
so
make
it
just
no
one
with
my
presentation
because
I
actually
cover
this.
So
this
is
the
over
setup,
how
it
would
look
like,
and
so
you
have
your
devices
in
the
field,
so
friends
in
some
sensors
and
actuator
and
something
that
is
an
actuator
but
also
provides
you
basic
information
about
the
position.
B
B
So
the
workflow
would
be
that
the
National
application
fetches
the
TDS
from
the
devices
as
usual,
but
those
are
then
enhanced.
With
these
capabilities
of
the
infrastructure,
then
image
these
capabilities
to
you
actual
quality
of
service
requirements.
So
basically
the
application
has
to
decide
how
often,
for
instance,
you
want
to
update
the
position
of
the
robot,
how
fast,
as
the
transport
system
should
be
controlled.
B
How
often
we
want
to
read
out
the
position
mean
from
the
center
and
so
on,
and
once
you
have
made
certain
choices
within
the
powers
of
the
quality
of
service
capabilities
depend
on
to
the
controller
at
one
tries
to
push
out
the
schedule,
so
heuristic
has
to
translate
and
actually
calculate
the
schedule.
That
is,
a
np-hard
problem
actually
come
from
the
requirements
down
to
the
state
control
list.
So
there
is
a
lot
of
Investigation,
Network,
calculus
and
so
on.
To
actually
do
that,
try
to
push
it
out
and
then
basically
report
back.
B
But
are
these
quality
of
service
parameters
super
insensitive,
irises
and
often
have
a
minimal
interval
to
actually
be
able
to
process
data
data
and
message
sizes
are
important
to
the
allocation
of
the
network.
You
want
to
know,
for
instance,
a
send
offset
so
the
time
between
the
sensor
reading
is
taken
when
it
hits
HP
network.
You
really
want
to
have
a
tight
timing
here
down
in
the
microsecond
range
there
might
be
some
receive
offset.
So
if
I
said
to
command
no,
the
network
delay.
B
How
long
does
it
still
take
in
the
device
itself
to
actually
provide
it
and
also
what
is
the
tock
synchronization?
What
is
the
mechanism,
via
several
standards
actually
out
there
and
on
the
side
on
the
requirements?
As
I
said,
the
application
knows,
which
is
what
is
maximum
interval,
so
our
first
mass
is
actually
one
two
to
fulfill
its
purpose.
What
is
the
overall
delay
that
is
allowed
that
it's
fresh,
for
instance?
What
is
the
receive
offset
service?
B
Yeah
so
my
hunch,
or
what
touch
why
play,
starts
to
slide,
and
these
are
basically
capabilities
of
the
things.
So,
if
I
use
this
I
know
what
is
the
half
a
table?
Also
actually
I
have
to
mention
dimension
is
that
the
robot,
for
instance,
is
able
to
quickly
process.
This
a
sensor
has
maybe
a
lower
one
and
then
what
is
possible
by
the
hot
way
and
so
on
is
described
here,
and
the
application
of
course,
then
has
to
also
choose
and
decide.
Based
on
this,
what
is
possible
of.
A
B
Indirectly,
because
the
the
components
are
wouldn't
be
able
to
actually
work
with
this,
it
still
needs
this
translation
step
to
really
work
on
networks,
article
gates,
control
terms
and
so
on,
but
yet
this
is
kind
of
representation
of
the
same
knowledge
requirements.
Okay,
so
this
is
used
to
essentially
produce
information.
B
A
B
B
There
are
some
at
forest
level
and
maybe
some
at
the
forum
level,
so
basically
because
the
maximum
message
service
over
HTTP
the
be
different
from
coop,
for
instance,
and
still
not
sure
which
we
are
I
believe
and
if
frame
sizes
relay
at
information
here,
maybe
there
should
there
should
be
more
abstraction,
but
if
you
really
want
to
calculate
in
the
end,
you
need
to
know
how
big
is
your
frame
really
on
on
layer?
Two
one
issue
is
that
we
don't
have
a
mechanism
to
actually
give
such
metadata
for
applications.
B
So
so
far
we
didn't
define
any
container
format
about
this.
However,
some
ongoing
work,
so
we
and
looked
I
think
it
was
you
project
or
something
which
he
I
has
something
that
they
use,
which
is
the
requirement
and
abilities
document
that
you
have
and
also
for
a
network
function.
Virtualization.
You
have
this
kind
of
descriptions.
What
are
the
group
I
officially
for
this
application
to
run
what
happen
is
we
fired
and
so
on.
A
B
To
conclude,
I
mean
if
n
is
quite
interesting
because
it
really
allows
convergence
there.
We
need
these
quality
of
service
parameters
who
have
requests
sent
to
that.
His
n
controller
may
be
site.
Info
TSN
is
also
investigated
for
wireless
solutions.
So
this
n
over
Wi-Fi,
but
also
5g,
is
a
topic
they're.
B
Probably
yeah
I
actually
have
to
catch
up
and
see
what
else
is
going
on
there,
which
is
pretty
obvious.
He
can
provide
the
semantic
information
here.
This
is
I
mean
a
first
step
and
I
was
quite
happy
to
see
that
this
topic,
which
I
wasn't
even
aware
of
when
we
started
this
project,
that
semantic
description
for
connectivity
is
actually
a
topic,
so
he
can
also
have
beer,
and
we
are
doing
the
first
steps
in
the
paper
that
we
are
planned
to
publish
and
another
issue
that
we
is
always
that.
B
Basically,
you
want
to
do
this
where
things
bindings,
the
devil,
is
in
the
details
and
we
need
to
make
way
more
efforts
there
to
figure
out
what
do
we
need
to
fulfill
this
vision
of
convergence
and
description
of
this
ole?
Some
other
point
is
that
we
could
discuss
maybe
offline
at
a
gear
or
whatsoever
and
I.
C
F
F
F
F
F
F
A
A
E
A
Crazy,
yes,
this
is
a
short
version
of
it.
Now
we're
each
in
this
room.
Okay
check
it
out.
This
HP
printer
uses
jabber
okay,
so
it's
using
a
chat
protocol
who
would
have
guessed
up
who
gets
to
say
what
the
printer
gets
to
use
only
personally
gets
to
say
that
with
their
design,
not
Cisco
ball
in
it,
not
the
w3c.
A
A
Conditions,
and-
and
this
is
one
device
right-
this
recurs
throughout
the
market
and,
however
many
types
of
devices
there
are
right.
The
administrator
has
to
go
and
do
that
order
to
figure
out
what
sort
of
access
the
device
needs.
Now,
why
would
anybody
care
for
access
device
needs
on
the
network?
The
answer
is
these
devices
cannot
possibly
defend
themselves,
they
don't
they
try
and
they
should
try.
They
should
keep
trying.
A
Sumer,
you
name
it
it's
there,
my
cousin
and
I'm,
sorry
for
taking
20
for
everybody.
20
visits
us
my
cousin
just
got
a
smart
oven
that
woke
her
up
at
5:30
in
the
morning
to
tell
her
that
need
to
be
cleaned.
I
asked
her
why
the
gods
meter
would
allow
that
networked
when
just
to
tell
you
that
it
needs
to
be
clean
when
it's
actually
the
threat
to
the
rest
of
her
environment
and
there
is
no
protection.
So
this
is
where
a
vacuum
usage.
A
So
let's
talk
about
assumptions
and
assertions
from
an
assumption
side.
We
start,
we
say
a
think-
has
a
single
small
convergence?
That's
my
definition,
part
of
my
definition
of
the
day.
It's
not
general
person
not
trying
to
solve
things,
are
tight
and
constrain.
They
have
a
very
loose
eacute,
very
little
memory
very
little
battery.
What's
not
on
the
slide
and
what
is
most
important.
You
have
a
very
little
programmer
time.
A
So
in
there
talk
about
based
on
that,
we
have
a
couple
of
assertions
we
commit
because
the
thing
has
a
sigmoid
small
number
of
uses.
Therefore,
we
can
enumerate
those
cases
and
the
people
who
are
best
in
position
to
enumerate
those
items,
because
there
are
the
manufacturer
themselves
and
the
mechanism
needs
to
protect
devices
that
are
made.
So
this
is
we've
had
access
lists
in
the
network.
A
A
A
Twice
emits
a
bun
URL
in
some
way
shape
or
form,
or
it
is
otherwise
learned.
Then
the
network
to
pick
that
up
send
it
over
to
this
one
man.
It
could
be
an
adjunct
here
to
play
server.
The
triple-eight
server
sees
a
URL
know.
What's
going
to
do
with
it,
you
do
what
it
does
with.
You
else.
Looks
it
up.
Those
are
this
file
server.
It's
just
great.
A
A
Have
to
Jason
and
what
we've
done
is
we've
taken.
The
access
list
model
we're
configuring
the
network's
who
are
considering
network
access,
so
we
take
that
access
model,
abstract
out
the
IP
addresses
and
in
some
cases,
some
cases
alone
we
need
to
fill
in.
We
provide
a
template
for
the
administrator
to
fill
in
some
information
as
to
what
sort
of
access.
A
Instantiated
form
into
the
network
once
it
approval
steps
happen
because
the
different
editor
it
needs
to
clean
up
or
castle,
and
then
the
device
is
the
microcytic.
So
you
provide
the
appropriate
network
access
Lighting's,
which
can
talk
to
the
light
bulbs
well
and
a
thermostat
in
solitude
at
age.
That.
A
A
A
Just
throbbin
RC
gotta
have
the
tooling
so
just
our
policy
developers
the
way
this
is
being
read,
there's
a
little
weird
yeah
this
you
have
some
fun
mood
maker,
org,
and
so
you
can
do
some
policy
offering
you
can
go
to
my
maker
double-click
on
it
today
and
you
can
try
this
out.
Maybe
you
take
the
results
of
the
mud,
pile
and
then
visualize
it.
This
is
cool.
This
is.
A
You
just
visualize
it
all
right,
and
then
we
have
some
cooling
that
implements.
So
this
is
an
example
of
the
template
here.
One
of
the
things
that
mud
does,
because
it
abstracts
out
you
way
that
at
the
access
is
managed.
You
have
your
choice
as
to
how
you're
going
to
deliver
the
accesses
to
our
network.
It
can
be
through
normal
ACLs.
It
can
be
open
it
through
source
tags
that
cisco
offices
here.
A
The
important
points,
the
first
of
which
is
mud,
configures
the
network,
we're
not
configuring
devices.
There
are
no
new
control
interfaces
on
the
device
back.
If
you
can
find
a
mic
URL
to
your
device
out
of
man
without
you
safe
on
a
label
or
something
like
that,
you
never
have
to
touch
the
device
at
all.
You
can.
A
The
device
therefore
requires
don't
semantic
understanding
of
what
about
violence.
Manufacturer
does,
but
the
device
does
not
right
so
very
little
processing
right
on
the
device.
All
it
has
to
do
is
admit
the
money
well
how's
the
system,
architecture,
if
you
can
get
them
on
your
own
without
the
device.
That's
a
good
example
like
before.
Oh
man,
right
where
you
have
a
table
of
keys
for
you
to
reach
to
buy,
you
have
a
device.
A
A
The
other
thing
is,
you
can
use
what
you
want.
You
could
just
mark
the
rest
that
includes
to
decide
that
all
you
want
to
do
is
take
this
for
classification
purposes.
Rule
under
classify
want
to
know
it's
on
your
network
just
want
to
know
policies,
never
mind
of
the
main
factory.
Okay,
you
could
also
not
even
bother
imposing
of
access
things
Walt.
You
could
just
as
easy
information
for
auditing
purposes.
John
expect
some
people
will
do
perhaps
take
them
by.
A
So
have
them
access
control
mechanisms
in
your
network
you're
not
going
to
get
access
control
just
because
you're
using
pal.
Fortunately,
most
modern,
routers
and
switches
have
this.
So
that's
pretty
generally
applicable,
that's
sort
of
cool.
Isn't
it
we've
done
a
lot
of
work.
We
focused
on
healthy
access
to
primary.
A
A
A
A
A
A
A
A
A
A
A
Configuration
for
me
long
term
is
that
I
don't
want
the
device
to
the
service
I
provided
mechanism
such
that
the
device
that
is
certain
now
but
I
really
like
to
a
grip
those
out
but
I,
think
well.
My
manufacturers,
one
dealing
with,
will
never
show
up
in
this
room
right.
It
took
me
two
years
to
get
them
tomorrow
with
they
know,
that's
how
difficult
was
just
just
use
lldp,
which
they
were
already
using.
A
A
A
A
A
A
A
A
A
A
G
A
A
A
Devices
no
I'm,
looking
at
the
same
steps
which
I
described
for
eyes
first
you're,
going
to
eventually
so
get
these
candidates
instead
of
instead
we're
actually
going
to
use
the
co-op
protocol
here
and
we're
going
to
use
resource,
then
the
create
the
candidates.
Nothing
strange
there
same
way
and
then,
when
you
dispute
it,
then
we
actually
going
to
use
the
core
protocol
and
the
resource
directory
and
and
resource
links
that
showed
you,
then
what
works.
So
everything
is
involved
here
is
things
that
you
already
have.
A
If
you
have
a
co-op
infrastructure,
you
don't
need
to
add
anything
x-ray.
If
you
want
this,
the
first
thing
you
did
so
basically
there
is
gonna,
be
I
talked
about
you
had
this
server
in
the
network
request
and
he
looks
very
onerous
very
come
by
the
same
thing
here.
We
call
it
a
tea
stun
server,
but
basically
it's
just
a
crop
server
which
hosts
a
resource,
a
dedicated
resource,
which
we
call
reflexive
resource
and
the
only
thing
what's
gonna
happen
is
when
he's
in
a
co-op
request.
A
A
To
be
resource-
and
it
is
going
to
return
it
to
this
resource-
will
actually
get
this
value
when
it
receives
the
request,
then,
if
you
have
mouth
you
may
have
multiple
devices
communicating
with
it,
so
you
may
get
get
requests
from
another
one
and
then
it's
going
to
assign
get
assign
the
value
from
where
that
request
comes
and
return
it.
This
will.
This
is
a
little
different,
the
normally
resources,
because
this
will
get
and
then
it's
going
to
send
it
back
and
then
of
obviously
now
you
have.
You
have
sent
this
request
through.
A
A
A
A
Two
ways
abilities:
he
can
create
the
resource
himself,
which
we
call
candidate
resource.
The
only
value
of
this
resource
is
this
IP
address
report
that
he
received
from
from
when
they
look
for
it.
It's
not
a
sensor
value
or
anything
else.
Just
this
IP
address
important
he's
creating
those
and
then
he's
creating
a
link
for
that.
A
resource
lead,
and
here
we
have
a
seaside.
A
Define
a
new
relation
which
is
called
candidates
and
the
anchor,
so
this
is
actually
linked.
This
Uriah
is
linked
to
add
to
a
real
resource.
I
talked
about
the
route
sense
or
resource
link
is,
but
it
could
be
whatever
resources
you
exposed.
So
basically,
you
can
say
that
this
is
a
candidate
for
these
resorts,
so
that's
the
first
way
and
then
you're
going
to
register
both
of
these
to
their
resource
director,
and
this
is
for
the
resource
directory.
You
register
these
resource
links
to
resources
and
then
other
endpoints
and
fetch
them.
A
That's
the
first
way
of
naming
it.
The
other
way
is
that
you
take
your
register
sensor
story,
your
sensory
subject-
and
you
put
IP,
address
this
candidate
information
directly
into
our
that,
so
you
replace
whatever
this
local
IP
address
and
port
that
that
the
resource
link
contains
with
this
candidate.
A
There
are
pros
and
cons
for
both
of
these
methods.
I
will
not
go
into
those
here
now.
It
was
a
tiny
issues,
but
but
you
can
do
both
ways
and
then
whoever
wants
to
contact
you?
Basically,
the
one
that's
going
to
use
this
candidate,
so
he
fetches
these
from
resource
directory
normal
procedures.
Now,
in
the
first
alternative,
he
will
fetch
both
a
resource
for
the
candidate
and
there's
a
resource.
And
again
you
have
this.
A
So
you
know
that
this
is
the
candidates
for
four
for
each
resource
and
he
will
replace
when
he's
going,
to
send
a
gate
for
three
sensorvalue.
He
would
put
the
candidate
into
this
game
Qri.
Then
he
will
send
the
get
instead.
Normally
he
would
try
to
sit
directly
here
this
address
here
and
hopefully
it
will
reach
rich
so
and.
A
A
A
But
then
he
also
registers
this
candidate
yeah.
But-
and
here
is
he
here-
he
actually
includes
this
candidate.
The
relation
is
candidate.
The
anchor
and
here
this
IP
address
port
is
the
same
as
here.
So
this
is
how
a
system
together,
because
you
need
to
know
because
there
may
be
multiple
devices
subscribing
you
don't
really
do
anything
with
a
candidate
know
what
to
use
it
for
right.
So
that's
actually
so
this
was
the
first
day.
A
A
Good
thing
about
sending
a
request
attack
trigger
a
response
is
that
you
will
always
get
a
stack,
so
you
will
see
for
a
subject
and
not
finding
s
for
now.
So
that's
always
have
choice.
You
you
will
have
to
eat.
If
you
don't
care,
if
you
trust
it's
not
going
to
be
done,
you
can
just
send
something
simple,
which
is
not
going
to
cost
so
much
process.
A
A
A
Using
his
address
so
then
he's
using
again
normal
resource
director
procedure
he
wants
to,
he
wants
to
have
a
temperature
sensor
range,
gets
that
one
back.
Then
he
also
one
he's
asking
candidates
which
again
using
damper
which
is
associated
with
this
one.
He
gets
this
one
back.
This
is
Ted,
never
take
it
takes
this.
One
puts
intricate
and
senses
here
main
reach
and
may
not
this
dead
over.
Obviously,
you
have
different
max
can
be
you
know.
Some
are
more
restricted,
another
way
where
they
gonna.
Let.
A
And
then
he
sends
the
response
back.
We
will
train
twenty
kids
or
whatever
the
temperature
values.
This
is
basically
the
use
case,
normal
coke
usage.
This
is
not
about
you
know,
setting
up
by
real
client
media
streamer.
Anything
between
these
using
the
resource
directory
trying
to
find
the
resource
leaves
for
the
date
that
is
interested
in
temperature,
for
example,
or
whatever,
and
then
contact.