Co
m
pu
ter Sci
ence a
nd Inf
or
mat
i
on
Tec
h
no
lo
gies
Vo
l.
1
, No
.
2
,
J
ul
y
2020
,
pp.
54
~
6
0
IS
S
N:
27
22
-
3221
,
DOI: 10
.11
591
/
csi
t.v
1i
2
.p
54
-
6
0
54
Journ
al h
om
e
page
:
http:
//
ia
esprime
.com/i
ndex.
php/csit
On the
perfo
rman
ce
of
switchin
g method
s in spa
ce divisi
on
mu
ltip
lexing bas
ed opti
cal net
wo
rks
Sridhar
I
yer
Depa
rtment
o
f
E
CE,
S.G
.
Ba
le
ku
ndri
Insti
tut
e
of Tec
hno
log
y
,
Shivaba
sava
n
aga
r
,
Bel
ag
avi
,
KA
,
I
ndia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
1
3
, 20
2
0
Re
vised
Ma
y
8
, 20
2
0
Accepte
d
Ma
y
19
, 20
2
0
In
the
cur
r
ent
work,
for
Space
D
ivi
sion
Mult
ipl
e
xing
base
d
Opti
c
al
Ne
tworks
(SD
M
-
b
-
OTN
s),
we
inve
stig
at
e
t
he
per
form
anc
e
of
var
ious
sw
it
ch
ing
m
et
hods
with
a
var
ia
t
ion
i
n
tra
ff
ic
evo
lut
io
n
over
diff
ere
nt
t
ime
fra
m
e
per
iod
s.
Initia
l
l
y
,
compari
son
of
the
exi
sting
m
e
thods
viz.,
ind
e
pende
nt
sw
it
ching
(InSw
),
fre
quency
sw
itc
hing
(FqS
w),
an
d
spac
e
sw
it
chi
n
g
(SpS
w)
demons
tra
te
s
that
(i
)
ove
r
longe
r
p
eri
ods
of
ti
m
e
fra
m
e,
FqS
w
provisions
low
n
et
wor
k
usage
,
and
(ii
)
SpS
w
offe
rs
l
ow
net
work
us
ag
e
for
shorter
per
i
ods
of
t
ime
fr
ame;
how
eve
r,
as
ti
m
e
fra
m
e
in
cre
ase
s
to
longer
per
iods,
SpS
w
start
s
to
ou
tpe
r
form
InSw
.
Next,
w
e
inv
est
iga
t
e
a
h
y
b
rid
sw
it
chi
ng
(H
y
Sw
)
m
et
hod
whi
ch
begi
ns
b
y
implementi
ng
In
Sw
and
th
en
shif
ts
to
the
use
of
SpS
w
aft
er
the
a
ct
iv
at
ion
o
f
spec
ific
num
ber
s
of
spa
ce
chann
el
s.
The
sim
ula
t
i
on
result
s
demo
nstrat
e
th
at
H
y
Sw
provision
s
subs
ta
nti
al
sav
ings
on
the
costs
inc
urr
ed
f
or
sw
i
tc
hing
,
and
with lower
spac
e
ch
annel
v
al
u
es
i
t al
so
offe
rs
a
ba
la
nc
e
in
th
e tra
d
e
-
off
which
occ
urs
be
twee
n
t
he
costs
associat
ed
for
activating
the
spa
ce
ch
annels
and
tha
t
inc
urre
d
for
sw
itching.
Ke
yw
or
d
s
:
Netw
ork
s
ha
ring
Re
so
urce al
loc
at
ion
Sp
ace
div
isi
on
m
ul
ti
plexing
Sw
it
chin
g
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Sr
id
har Iyer
,
Dep
a
rtm
ent o
f EC
E,
S.G. Bal
ek
undri
I
ns
ti
tute o
f
T
echnolo
gy,
Sh
iva
basa
va
na
gar, Bel
aga
vi, KA
, In
dia.
Em
a
il
: sridh
ari
ye
r1983@gm
a
il
.co
m
1.
INTROD
U
CTION
W
it
h
a
n
inc
re
m
ental
increas
e
in
In
te
r
net
tr
aff
ic
le
adi
ng
to
an
e
xpone
ntial
grow
t
h
of
de
m
and
s,
it
ha
s
been
nec
essar
y
to
inc
rease
the
co
rr
e
spo
nding
opti
cal
net
work
(
OT
N)
c
apacit
y.
H
owe
ver,
with
the
OTN
capaci
ty
increa
se,
sim
ultaneou
sly
,
the
in
dustry
al
so
str
ive
s
towa
rd
s
re
du
ci
ng
a
sso
ci
at
ed
netw
orkin
g
c
ost
s
[1
]
.
Fo
r
al
m
os
t
a
de
cade,
wa
vele
ng
t
h
div
isi
on
m
ul
ti
plexed
(
WD
M
)
OTNs
hav
e
ke
pt
pace
with
t
he
e
xponentia
ll
y
increasin
g
ca
pa
ci
ty
de
m
and
s
by
de
plo
yi
ng
sing
le
m
od
e
fi
ber
(S
MF
)
an
d
sh
ari
ng
num
e
rous
fr
e
qu
e
nci
es
(
or
wav
el
e
ng
t
hs
)
[
2]
ena
bled
by
wav
el
e
ng
t
h sel
ect
ive
switc
hes
(
W
S
Ss)
t
hat
fl
exibly
pe
rfor
m
fr
e
qu
e
ncy
s
witc
hin
g
(F
qSw
).
H
ow
e
ver,
once
a
de
plo
ye
d
SMF
is
ex
hau
ste
d
of
capaci
ty
,
an
e
xpone
ntial
incre
ase
in
fi
ber
s
a
m
ou
nt
will
be
require
d
to
m
eet
req
ue
ste
d
capaci
t
y
dem
and
w
hich
r
equ
i
res
no
vel
ne
twork
s
har
i
ng
strat
egies.
T
he
above
scenari
o
can
be
a
m
el
iorated
by
the
adoptio
n
of
s
pace
div
isi
on
m
ulti
plexing
(SDM)
te
c
hnology
w
hich
prov
i
des
m
or
e
capaci
ty
at
reduce
d
network
co
sts
sim
ultaneo
us
ly
with
bette
r
scal
a
bili
ty
and
inte
gr
at
io
n
[3
]
.
Th
e
SDM
base
d
OT
Ns
(
SD
M
-
b
-
OT
Ns
)
prov
i
de
f
reedom
deg
rees
in
bo
t
h,
s
pace
an
d
f
reque
ncy
w
her
ei
n,
e
ver
y
op
ti
cal
sign
al
is
ass
oci
at
ed
to
a
f
requ
ency
c
hannel
(
Fq
Ch
n)
a
nd
a
sp
ace
cha
nnel
(S
pC
hn)
w
hic
h
m
ay
be
de
note
d
as
M
,
.
.
.
.
.
2
,
1
m
Fq
and
N
,
.
.
.
.
.
2
,
1
n
Sp
,
res
pecti
ve
ly
[4
]
.
In
SDM
-
b
-
OT
Ns,
t
he
dep
l
oyed
fi
be
r
capaci
ty
can
be
util
iz
ed
eff
ic
i
ently
as
dem
and
capaci
ty
of
a so
urce
a
nd
d
es
ti
nation
pair
m
ay
be
ap
pro
xim
at
ely
equ
al
to
the
ca
pacit
y
w
hich
th
e
fibe
r
can
pro
visio
n;
ho
weve
r,
t
o
f
ur
t
he
r
i
nc
rease
t
he
e
ff
ic
i
ency,
novel
sw
it
ching
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
On the
perf
orma
nce
of swit
chin
g meth
ods i
n
s
p
ac
e
divisi
on
mu
lt
iplex
in
g ba
s
ed… (
Sri
dh
ar
Iyer
)
55
and
sh
a
rin
g
te
chn
i
qu
e
s
will
be
re
quired
[
5,
6].
The
e
xisti
ng
OT
Ns
im
pl
e
m
ent
ind
e
penden
t
s
witc
hing
(In
S
w
)
wh
ic
h
pro
vid
e
s
high
flexi
bili
ty
bu
t
inc
rease
s
c
om
plexity
of
s
witc
hi
ng
nodes
[
7,
8].
T
o
reduce
com
plexity
,
stud
ie
s
in
[
9,
10]
hav
e
dem
on
strat
ed
im
ple
m
entat
ion
of
(i)
fr
e
qu
e
ncy
swit
chin
g
(FqS
w),
wh
ic
h
us
es
networ
k
nodes
that
s
witc
h
al
l
Sp
Ch
ns
placed
at
certai
n
f
reque
ncy
from
any
inp
ut
to
any
out
pu
t
di
recti
on
[11],
a
nd
(ii)
sp
at
ia
l
switc
hi
ng
(SpS
w),
whic
h
assig
ns
s
pe
ct
ral
superc
ha
nn
el
s
(
or
ba
ndwidth
of
a
ny
fi
ber
or
any
c
ore)
in
a
dynam
ic
m
anner
[11].
I
n
the
current
work,
c
on
si
der
i
ng
t
he
change
in
ti
m
e
f
ram
e
and
e
voluti
on
of
tra
ff
ic
ove
r
this
ti
m
e
fr
a
m
e
,
we
in
vestigat
e
the
im
pact
of
the
va
rio
us
sw
it
ching
m
et
ho
ds
on
the
pe
rform
ance
of
a
S
D
M
-
b
-
OTN
w
hich
ad
op
ts
m
ulti
ple
cor
e
fibe
rs
or
S
MF
bundle
s
as
the
fi
ber
s
ol
ution
i.e.,
the
ty
pe
of
tra
ns
m
issi
o
n
m
edia
in which
the
spa
ce su
per
c
ha
nnel
s ar
e
no
t c
ou
pled.
The
a
utho
rs
i
n
[9
]
ha
ve
i
nv
est
igate
d
de
plo
y
m
ent
of
reconfi
gura
ble
ad
d
-
dr
op
m
ulti
plexers
(R
OADMs)
for S
DM
-
b
-
OT
Ns
with
flexi
ble
f
reque
ncy
gri
d a
nd
e
nab
le
d
by
m
ulti
ple
core
fibe
rs;
howe
ver,
the
a
uthor
s
ha
ve
no
t
a
ddresse
d
Sp
S
w
in
t
he
s
tud
y.
In
[12,
13]
,
the
a
uthors
hav
e
in
vestig
at
ed
dif
fer
e
nt
resou
rce
assig
nm
ent
strat
egies
incl
udin
g
F
qSw; h
oweve
r,
in both
stud
ie
s,
the
aut
hors
ha
ve
not
a
ccounted
f
or
S
pSw.
The
aut
hors
in
[14]
hav
e
in
ves
ti
gated
var
i
ou
s
switc
hing
m
et
h
od
s
in
reg
a
rd
t
o
SD
M
-
b
-
O
TN
s;
howe
ve
r,
S
pSw
is
no
t
acco
un
te
d
for
within
t
he
stud
y.
I
n
[15
]
,
the
a
uthors
ha
ve
a
ddre
ss
ed
F
qSw
for
inv
est
igati
ng
SD
M
-
b
-
OT
Ns
w
hic
h
dem
on
strat
e
fl
exibili
ty
in
bot
h,
sp
ace
an
d
frequ
e
ncy
do
m
ai
ns
.
The
aut
hor
s
in
[16]
hav
e
no
t
c
onside
re
d
ei
ther
Fq
S
w
or
S
pSw
of
s
pace
s
up
e
r
channels
i
n
the
ir
pro
posal
of
a
rch
it
ect
ure
on
dem
and
co
nce
pt
w
hich
is
a
ppli
cable
to
SD
M
-
b
-
OT
Ns.
In
[
17
]
,
t
he
auth
ors
ha
ve
consi
de
red
SpSw
for
only
c
ertai
n
S
pSns
a
m
ou
nt
to
in
ve
sti
gate
SD
M
-
b
-
OT
N
pe
rfor
m
ance
in
data
ce
nters
w
hich
util
iz
es
va
rio
us
s
witc
hing
arc
hitec
tures.
Last
ly
,
the
a
uth
ors
i
n
[4,
8,
10
]
ha
ve
rev
ie
wed
diff
e
r
ent
SD
M
-
b
-
OT
N
node
a
rc
hitec
tures;
ho
wev
e
r,
unli
ke
c
urrent
stu
dy,
aut
hors
ha
ve
no
t acc
ounte
d for c
hange i
n
ti
m
e fr
am
e and
evo
l
ution o
f
t
r
aff
ic
.
I
n
the
c
urren
t
work,
init
ia
ll
y,
we
c
om
par
e
perform
ance
of
e
xisti
ng
s
witc
hing
so
l
utions
viz.,
I
nS
w
,
Fq
S
w,
a
nd
SpS
w
by
co
nductin
g
exte
ns
ive
si
m
ula
ti
on
ex
perim
ents
con
side
rin
g
reali
sti
c
network
to
polo
gi
es
an
d
par
am
et
ers.
T
he
res
ults
dem
on
strat
e
t
hat
(i)
ov
e
r
a
l
onger
ti
m
e
fr
am
e,
FqS
w
pro
visio
ns
l
ow
netw
ork
util
i
zat
ion
,
and
(ii)
SpSw
offer
s
lo
w
network
util
iz
at
io
n
for
s
h
or
te
r
ti
m
e
fr
am
es;
howev
e
r,
as
the
t
i
m
e
fr
am
e
increases
to
longer
pe
rio
ds,
S
pSw
sta
rts
t
o
outper
f
or
m
I
nSw.
Ne
xt,
f
or
postp
onin
g
t
he
costs
ass
ocia
te
d
with
act
iva
ti
ng
a
new
Sp
C
hn,
w
e
pro
po
s
e
a
ne
w
hy
br
i
d
s
witc
hing
(
HyS
w)
m
et
ho
d
w
hich
beg
i
ns
by
im
pl
e
m
enting
I
nSw
a
nd
then
s
hifts
t
o
use
of
F
qSw
a
fter
act
ivati
on
of
s
pecific
nu
m
ber
s
of
S
pC
hns.
W
e
com
par
e
the
perform
ance
of
existi
ng
s
witc
hing
s
olu
ti
on
s
with
HySw
and
r
esults
de
m
on
strat
e
that
HyS
w
postp
on
e
s
relat
ed
Sp
Ch
ns
act
ivati
on
c
os
ts
si
m
ultaneous
ly
pr
ovisi
onin
g
substa
ntial
savin
gs
on
c
os
t
s
incu
rr
e
d
f
or
switc
hing.
Re
s
t
of
the
pap
e
r
is
str
uctu
red
as
f
ollows:
In
sect
io
n
2,
w
e
detai
l
the
HySw
m
et
ho
d
a
nd
al
so
prese
nt
th
e
onli
ne
(
or
dy
nam
ic
)
resou
rce
a
ssig
nm
ent
al
gorith
m
s
app
li
cable
to
e
ver
y
co
nsi
der
e
d
switc
hi
ng
m
et
ho
d
.
Se
ct
ion
3
pr
ese
nt
s
the
si
m
ulati
on
set
up
an
d
t
he
va
rio
us
ob
ta
in
ed
sim
ulatio
n
re
su
lt
s
an
d
t
he
corres
pondin
g
disc
us
sio
ns
.
Finall
y,
sect
ion
4
c
oncl
ud
e
s the
stu
dy.
2.
HYBRID
SW
ITCHI
NG M
ET
HOD
Fr
om
the
exist
ing
st
ud
ie
s
it
is
cl
ear
that
on
on
e
ha
nd,
I
nSw
re
quire
s
le
ss
er
am
ou
nt
of
Sp
Ch
ns
a
nd
higher
com
plexity
of
netw
ork
s
witc
hing,
on
t
he
oth
e
r
ha
nd,
SpSw
re
duces
t
he
switc
hing
c
om
plexity
by
act
ivati
ng
high
er
am
ou
nts
of
Sp
Ch
ns
in
the
init
ia
l
tim
e
fr
a
m
e
per
io
ds.
T
he
pro
posed
HySw
m
et
hod
c
om
bin
es
the
best
at
trib
ut
es
o
f
I
nSw
a
nd
Fq
S
w.
We
pro
po
s
e
HySw
with
a
view
to
pos
tpone
t
he
relat
ed
Sp
C
hns
(i.e.
f
iber
s
or
c
ores)
act
ivati
on
co
sts
si
m
ultaneou
sly
pro
visio
ning
substanti
al
s
avin
gs
on
the
costs
i
ncurr
e
d
f
or
the
s
witc
hing.
The
HyS
w
m
e
thod
sta
rts
by
i
m
ple
m
enting
In
S
w
at
ea
r
ly
sta
ges
of
ti
m
e
f
ram
e
and
s
hifts
t
o
i
m
ple
m
entat
io
n
of
SpSw
upon
t
he
act
ivati
on
of
a
ce
rtai
n
Sp
N
am
ou
nt
of
S
pC
hn
s
.
He
nce,
H
ySw
sp
li
ts
S
D
M
b
-
OT
N
int
o
tw
o
la
ye
rs w
hich do n
ot
dep
e
nd
on one a
nothe
r
and a
re acti
vat
ed
at
var
ie
d
in
s
ta
nt of tim
e fr
am
e.
As
a
n
e
xam
ple
,
Fig
ur
e
1
sho
ws
the
ev
olu
ti
on
of
im
ple
m
e
ntati
on
of
HyS
w
for
2
Sp
N
.
In
su
c
h
a
scal
able
arc
hitec
ture,
I
nSw
is
i
m
ple
m
ented
f
or
1
N
(see
t
he
1
st
c
olu
m
n)
,
a
nd
f
or
2
N
(see
the
2
nd
c
olu
m
n)
.
Howe
ver,
i
f
ne
w
(see
the
3
rd
a
nd
4
th
c
oloum
n)
Sp
C
ns
are
ac
ti
vated
t
hen,
S
pSw
is
im
ple
m
ented
e
ns
uri
ng
that
the
nodes
wh
ic
h
we
re
de
plo
ye
d
t
o
im
ple
m
ent
I
nSw
init
ia
ll
y
rem
ai
n
act
ive
a
nd
op
e
rati
onal
.
O
ve
rall
,
the
m
ajor
idea
of
im
ple
m
enting
HyS
w
is
t
ha
t
init
ia
ll
y,
with
InSw
,
since
le
sse
r
SpC
hn
s
will
be
a
ct
ivate
d,
fine
r
value
of
gr
a
nula
rity
can
be
us
e
d
to
swi
tc
h
the
sig
nals;
at
la
te
r
sta
ges,
w
hen
the
act
ive
S
pC
hn
s
am
ount
i
ncr
eases
,
wit
h
Sp
S
w,
f
re
qu
e
nc
y
cl
ashes
can
be
pr
e
ve
nted
a
nd
in
tu
rn
the
nodes
com
pl
ex
it
y
can
be
m
inim
iz
ed.
He
nce,
there
occurs
a
postp
on
em
ent
in
S
pC
hn
s
bei
ng
ac
ti
vated
w
hich
will
resu
lt
in
reducti
on
of
t
he
ass
ociat
ed
costs.
Althou
gh,
S
pSw
will
nee
d
SpC
hn
s
t
o
be
act
i
vated
un
ti
m
ely
,
a
com
bin
at
ion
of
I
nSw
an
d
Sp
S
w,
as
i
n
H
ySw
,
will
al
le
viat
e
t
he
af
or
em
entio
ned
sc
ena
rio.
Also
,
f
or
S
pC
hns
to
be
act
ivat
ed,
ass
um
ing
that
netw
ork
op
erator
has
fibe
r
s
olu
ti
on
a
vaila
ble
at
his
e
nd,
opti
cal
am
plifie
rs
an
d
opti
cal
switc
hi
ng
m
at
rices
will
be
re
quire
d
to
be
dep
l
oyed
in
t
he
entire
netw
or
k
w
hich
will
i
ncu
r
inc
rease
d
instal
la
ti
on
cos
ts
at
early
t
i
m
e
fr
am
e.
Ho
we
ve
r,
as
the
m
ajo
r
be
ne
fits,
(i)
in
c
omparis
on
t
o
the
use
of
WSSs
by
In
S
w,
SpSw
use
s
op
ti
cal
switc
hing
m
at
rices
wh
ic
h
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol.
1
, N
o.
2
,
J
ul
y
20
20
:
54
–
6
0
56
are
m
uch
chea
per
a
nd
al
so
m
ini
m
iz
e
costs
a
sso
ci
at
ed
with
reg
e
ner
at
i
on
of
the
opti
ca
l
sig
nals,
an
d
(ii)
use
of
the
Sp
C
hns
ca
n
be
a
voide
d
f
or
l
onger
ti
m
e
fr
am
es
by
i
m
pl
e
m
enting
S
pSw
w
hich
will
postp
one
de
plo
y
m
ent
of
new fi
ber
so
l
ution
s
.
Figure
1.
Wo
r
ki
ng
of
HySw
m
et
hod wit
h
e
voluti
on
of ti
m
e fr
am
e con
side
ring th
reshold
val
ue
of
2
Sp
N
2.1.
Onli
ne
res
ou
r
ce ass
i
gn
me
nt alg
orit
hms
I
n
I
n
this
s
ub
-
sect
ion
,
we
det
ai
l
the
onli
ne
r
eso
ur
ce
as
sig
nm
ent
al
gorith
m
s
wh
ic
h
are
app
li
cable
t
o
the
c
on
si
der
e
d
switc
hi
ng
m
e
thods.
Wh
e
n
I
nSw
is
t
o
be
i
m
ple
m
ented
th
en
i
niti
al
ly
,
k
-
sh
ort
est
path
(
k
-
SP
)
al
gorithm
is
use
d
t
o
r
ou
te
the
sign
al
s
f
ollo
wing
wh
ic
h
first
f
it
wavel
en
gth
a
ssign
m
ent
m
eth
od
is
us
e
d
to
a
ll
ocate
the
freq
ue
ncies
(or
wav
el
e
ngths
).
A
dem
a
nd
is
acce
pte
d
in
case
wh
e
n
the
desi
red
c
apacit
y
is
avail
able;
howe
ver,
act
iv
at
ion
of
a
ne
w
Sp
Ch
n
over
th
e
entire
S
DM
-
b
-
OT
N
li
nks
is
cond
ucted
when
the
desire
d
capaci
ty
is
not
f
ound
a
nd
the
al
gorith
m
is
then
re
sta
rted
f
or
acce
pt
ance
of
the
de
m
and
w
hich
is
inc
om
ing
.
It
m
us
t
be
no
te
d
that
a
de
m
and
w
hich
is
incom
ing
can
be
acce
pted
by
ut
il
iz
ing
a
sp
ace
su
pe
rc
hannel
wh
ic
h
exists
an
d
ha
s
the
sam
e
so
ur
c
e
an
d
destinat
ion
pair
by
virt
ue
of
t
he
act
iv
at
ion
of
a
ne
w
Sp
C
hn
w
hich
resu
lt
s
in
creat
ion
of
extra ca
pacit
y for all
s
pace s
uperc
ha
nn
el
t
hat ex
ist
.
Wh
e
n
F
qSw
is
to
be
im
ple
m
e
nted
the
n
init
ia
ll
y,
to
ser
ve
the
de
m
and
,
t
he
al
gorithm
sta
rts
by
c
hec
king
wh
et
her
the
re
a
lready
e
xists
a
sp
ace
superc
ha
nn
el
w
hich
ha
s
bee
n
a
ssig
ned
with
t
he
a
vaila
ble
desire
d
ca
pa
ci
ty
.
If
su
c
h
a
s
pace
su
pe
rc
hannel
e
xists
the
n,
dem
and
is
pro
visio
ned
with
e
xisti
ng
avail
a
ble
ca
pacit
y;
el
se,
al
gorithm
us
es
k
-
S
P
to
de
te
rm
ine
a
s
pac
e
superc
ha
nnel
w
hich
is
avail
able.
Nex
t
,
de
m
and
is
ser
ved
if
the
al
gorith
m
is
able
to
assi
gn
t
he
s
pace
super
cha
nn
el
w
hich
is
avail
able;
el
se,
act
ivati
on
of
a
ne
w
S
pChn
over
the
e
ntire
SD
M
-
b
-
OTN
li
nks
is
c
onduct
ed
a
nd
t
he
al
gorit
hm
is
then
restarte
d
f
or
t
he
acce
ptan
ce
of
the
dem
a
nd
w
hich
is
inc
om
ing
.
Wh
e
n
S
pSw
is
to
be
im
ple
m
ented
the
n
init
ia
ll
y,
al
go
rith
m
deter
m
ines
the
e
xistence
of
a
n
assig
ne
d
s
pectra
l
su
pe
rc
hannel
wh
ic
h
has
t
he
avail
able
desir
ed
capaci
ty
,
an
d
if
it
finds
su
c
h
a
sp
ect
ral
superc
ha
nn
el
,
de
m
and
is
serv
e
d
us
i
ng
th
e
f
ound
ca
pacit
y.
H
ow
e
ve
r,
if
su
c
h
a
s
pectral
superc
ha
nn
el
i
s
not
f
ound
t
he
n
al
go
rithm
us
es
k
-
SP
t
o
determ
i
ne
a
s
pectral
su
pe
rc
hannel
with
a
vaila
ble
desi
red
cap
ac
it
y.
Upo
n
the
su
ccess
fu
l
se
a
rch
i
ng,
al
gorithm
serves
dem
and
by
assigni
ng
the
de
te
rm
ined
s
pec
tral
supe
rch
a
nnel
;
el
se,
act
iv
at
ion
of
a
ne
w
Sp
Ch
n
ov
e
r
t
he
entire
S
DM
-
b
-
OT
N
li
nk
s
is
c
onduc
te
d
a
nd
the
al
gorithm
is
the
n
restarted
f
or
t
he
acce
pta
nce
of
th
e
dem
and
w
hich
is
i
nc
om
ing
.
F
inall
y
,
w
he
n
H
ySw
is
to
be
im
ple
m
ented
th
en,
the
al
go
rith
m
execu
te
s
InSw
ti
ll
the act
ivati
on
of
Sp
N
Sp
C
hs
n f
ollo
wing
wh
ic
h
th
e algori
thm
ex
ecutes S
pSw
f
or the
rem
ai
nin
g
ti
m
e fr
am
e.
3.
SIMULATI
O
N RESULTS
AND DIS
C
USSION
S
To
c
om
par
e
pe
rfor
m
ance
of
consi
der
e
d
s
wi
tc
hin
g
m
et
ho
ds,
sim
ulatio
n
e
xp
e
rim
ents
are
cond
ucted
su
c
h
that
the
c
onnecti
on
dem
ands
wh
ic
h
a
re
acce
pte
d
with
in
the
net
wor
k
are
ne
ver
to
r
n
dow
n.
Co
ns
i
de
rin
g
the
fact
that
a
ct
ivati
on
of
a
new
Sp
C
hn
ov
er
al
l
li
nk
s
res
ults
in
pro
visi
on
i
ng
e
xtra
ca
pacit
y
to
serv
e
a
new
dem
and
, s
im
ulati
on
s a
re c
onduct
ed
i
n
a m
ann
er
w
hich
le
a
ds t
o no bl
ock
i
ng
of
t
he dem
and
s. Als
o,
betw
een all
the
s
ource
a
nd
de
sti
nation
pa
irs,
unif
or
m
di
stribu
ti
on
is
f
ol
lowed,
an
d
spa
ce
a
nd
fr
e
qu
e
ncy
c
onve
rsion
is
not
per
m
it
te
d
in
t
he
si
m
ulati
on
s.
Fu
rt
her,
to
dete
rm
ine
the
cand
idate
paths
for
e
ver
y
s
ource
-
de
sti
nation
no
des
pairs,
we
us
e
k
-
S
P
with
a
fi
xed
va
lue
of
k
=
3
wh
ic
h
ha
s
bee
n
sho
wn
t
o
de
m
on
strat
e
the
best
pe
rfor
m
ance
in
the
SD
M
-
b
-
O
TNs
[
18
]
.
The
nu
m
ber
of
f
r
equ
e
ncy
slots
cons
idere
d
is
96
with
each
occupyi
ng
50
GH
z
of
sp
ect
r
um
wh
ic
h
am
ount
to
a
total
of
4.8
T
H
z
ba
ndwi
dth
[
19
]
.
F
urt
her,
t
he
co
ns
ide
re
d
de
m
and
s
ha
ve
a
fixe
d
capaci
ty
of
100
G
bps
wh
ic
h
are
assig
ne
d
over
a
fixe
d
50
GH
z
fr
e
quenc
y
slot.
Each
run
of
the
sim
ulati
on
i
s
de
te
rm
ined
as
the
m
ean
of
50
r
un
s
.
F
or
pe
rform
ance
evaluati
on
s
,
we
co
ns
i
de
r
tw
o
reali
sti
c
netw
ork
to
polo
gies:
sm
a
ll
distance
DT,
a
nd
la
r
ger
dista
nce
GE
A
NT
[
20
-
21
]
.
I
n
re
gard
to
ca
pa
ci
ty
,
fo
ll
owin
g
[
22]
w
hic
h
predict
the
com
pound
annual
grow
t
h
rates
(C
AG
Rs
)
rangin
g
betwee
n
20%
-
50%
de
pendin
g
on
th
e
ge
ographic
re
gions
,
sta
rting
val
ue
i
n
t
he
1
st
ye
a
r
i
s
set
t
o
5
Tb
ps
w
hich
is
t
hen
increase
d
e
ve
r
y
ye
ar
fo
ll
owi
ng
CA
GR
of
20%
or
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
On the
perf
orma
nce
of swit
chin
g meth
ods i
n
s
p
ac
e
divisi
on
mu
lt
iplex
in
g ba
s
ed… (
Sri
dh
ar
Iyer
)
57
40%.
As
the
pe
rfor
m
ance
m
e
tric
s
w
hich
underg
o
a
cha
ng
e
eve
ry
ye
ar
,
we
us
e
the
am
ount
of
(i)
a
gg
reg
at
e
act
ive
S
pChn
s
(
AgAcS
pC
hns)
,
a
nd
(ii)
netw
ork
usa
ge
(N
et
Usg)
(in
%)
w
hi
ch
is
e
valuat
ed
as
Ed
A
c
S
p
F
S
s
p
FS
H
N
e
t
U
s
g
d
1
j
j
j
[1,
4]
wh
e
re, i
n n
um
erator
,
ge
ner
at
e
d dem
and
s
am
ou
nt
is
denoted
by
d
,
j
H
de
no
te
s
ho
ps
a
m
ou
nt
w
hich
a
dem
and
d
tr
aver
se
s,
a
nd
j
FS
de
no
te
s
f
reque
nc
y
slots
am
ou
nt
w
hich
a
dem
and
d
util
iz
es.
Furthe
r,
in
de
no
m
inator,
F
S
s
p
de
no
te
s
f
re
qu
e
nc
y
slots
am
ou
nt
wh
ic
h
eve
ry
S
pCh
s
upports,
A
c
S
p
de
no
te
s
am
ou
nt of
Sp
C
hns
wh
ic
h
a
re acti
ve
, and
Ed
de
note
s
am
ou
nt of
net
work gra
ph’s
e
dg
e
s.
3.1.
Perfo
r
ma
nce
comp
ariso
n
of existin
g sw
itching m
eth
od
s
In
this
s
ub
-
sec
ti
on
,
we
pr
ese
nt
perform
ance
com
par
iso
n
r
esults
of
the
e
xisti
ng
s
witc
hi
ng
m
et
ho
ds.
The
re
su
lt
s
are
pr
ese
nted
i
n
F
igure
2
c
onsid
erin
g
the
CA
G
R
values
of
20%
and
40%.
From
the
resu
lt
s
it
can
be
see
n
that
tre
nd
s
of
res
ults
is
the
sam
e
fo
r
bo
t
h
the
c
onsidere
d
to
polo
gies.
Spec
ific
al
ly
,
f
or
bo
t
h
to
po
l
og
ie
s
,
on
ly
tw
o
Sp
C
hs
are
nee
de
d
by
InSw
an
d
FqSw
for
a
ti
m
e
fr
am
e
per
io
d
of
7
a
nd
5
ye
ars
co
rr
es
pondin
g
to
a
CAGR
value
of
20%
a
nd
40%,
res
pecti
vel
y.
He
nce,
it
ca
n
be
in
ferred
that
at
lo
wer
va
lues
of
netw
ork
loa
ds
bo
t
h,
I
nS
w
and
FqSw
inc
ur
l
ower a
sso
ci
at
e
d i
nv
est
m
ent co
s
ts. Furthe
r,
i
n r
egard t
o uti
li
za
ti
on
of
t
he net
work,
it
can
be
obse
r
ved that
bo
t
h,
I
nSw a
nd FqSw
achieve
h
i
gh
e
st perfo
rm
ance o
f
app
roxim
ately
4
0%
i
n
the
init
ia
l
per
i
od
s
of
tim
e
fr
am
e.
Hen
c
eforth,
f
or
F
qSw,
NetUs
g
val
ue
m
ini
m
iz
es
and
rem
ai
ns
below
25%.
It
m
us
t
be
no
te
d
that
highest
Net
Us
g
va
lues
of
InSw
an
d
F
qSw
co
rr
es
ponds
t
o
the
fact
t
hat
th
e
ver
y
first
S
pC
h
ha
s
exh
a
us
te
d
the
capaci
ty
and
a
seco
nd
Sp
C
h
a
ct
ivati
on
is
require
d
w
hic
h
m
i
ni
m
iz
es
us
age
of
netw
ork
.
F
urt
her,
after
t
he
a
fore
m
entioned
j
un
ct
ur
e
is
reac
he
d,
A
gA
c
Sp
C
hn
s
is
obser
ve
d
to
inc
ur
a
n
expo
nen
ti
al
in
crease
wh
e
reas,
NetU
sg
value
s
c
onti
nu
e
to
hav
e
lo
wer
val
ues.
H
e
nce,
it
can
be
i
nf
e
rr
e
d
t
hat
F
qSw
is
no
t
a
favoura
ble
switc
hing
s
ol
ution
f
or
lo
nger
per
i
od
s
of
tim
e
f
ram
e
even
th
ough
e
ff
ic
ie
nt
fr
e
qu
e
ncy
al
lot
m
ent
m
e
tho
ds
m
ay
be
us
e
d
wh
ic
h
howev
e
r
will
onl
y
sli
gh
tl
y
incr
ease
spe
ct
ral
-
e
ff
ic
ie
ncy.
It
ca
n
al
s
o
be
see
n
that
I
nSw
is
a
ble
t
o
pro
vision hi
gher
netw
ork usa
ge values
of a
ppr
oxim
a
te
ly
4
5%
f
or lo
ng
e
r
ti
m
e fr
am
e p
erio
ds
.
(a)
(b)
Figure
2. A
gA
cSpC
hn
s
(
le
ft
) a
nd N
et
Us
g
(
ri
gh
t
)
re
su
lt
s
for C
AG
R
value
s
of 20% a
nd
40% co
ns
ide
rin
g
the n
et
wor
k
t
opol
og
ie
s
(
a
) D
T, a
nd (b) GE
AN
T
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol.
1
, N
o.
2
,
J
ul
y
20
20
:
54
–
6
0
58
The
S
pSw
m
et
hod
is
obse
rv
e
d
t
o
act
ivate
hi
gh
e
r
Sp
C
hns
a
m
ou
nt
in
th
e
init
ia
l
tim
e
fr
a
m
e
per
i
ods
since
it
n
ee
ds
to
c
onne
ct
al
l
t
he
source
a
nd d
est
inati
on p
ai
rs
i
n
the
S
DM
-
b
-
OT
N
t
hro
ugh
i
nd
i
vidual S
MF,
a
nd
hen
ce
,
usa
ge
of
net
work
has
a
lowe
r
val
ues
un
ti
l
su
c
h
a
c
onne
ct
io
n
is
c
om
plete
d.
As
t
he
tim
e
fr
am
e
per
i
od
s
increase,
SpS
w
is
see
n
to
ou
t
perform
InSw
eve
n
th
ou
gh
S
pSw
ha
s
m
uch
lo
wer
fl
exibili
ty
of
s
witc
hing.
The
af
or
em
entione
d
occ
urs
s
ince
S
pSw
does
no
t
inc
ur
a
ny
s
ubsta
ntial
capaci
ty
losses
owin
g
to
f
re
quenc
y
cl
ashi
ng,
a
nd
t
he
sam
e
can
be
obser
ved
w
he
n
CA
GR
val
ue
is
40
%
unde
r
w
hich
,
S
pSw
has
a
n
ap
pro
xim
ate
NetUs
g value
of 60% a
nd
75% for
the
DT
a
nd the
GE
A
NT
n
et
w
ork
to
po
l
og
y,
r
es
pecti
ve
ly
.
3.2.
Perfo
r
ma
nce
comp
ariso
n
of existin
g sw
itching m
eth
od
s
w
ith h
ybri
d s
w
itching
me
thod
In
this
sub
-
sect
ion
,
we
pr
es
ent
the
pe
rfor
m
ance
com
par
iso
n
resu
lt
s
of
t
he e
xisti
ng
s
witc
hin
g
m
et
ho
ds
with
the
pro
posed
HyS
w
m
eth
od.
For
the
H
ySw
m
et
ho
d,
I
nSw
is
i
m
ple
m
ented
un
ti
l
cert
ai
n
Sp
N
am
ou
nts
of
Sp
Ch
ns
ha
ve
be
en
act
ivate
d
f
ollow
i
ng
w
hic
h,
t
he
S
pSw
m
et
hod
is
im
ple
m
ented.
I
n
t
he
si
m
ulati
on
s,
w
e
var
y
the
val
ue
of
Sp
N
as
10
,
8
,
6
,
4
,
2
N
Sp
.
T
he
sim
ulatio
n
ex
pe
rim
ent
resu
lt
s
a
re
pr
e
s
ented
i
n
Fig
ur
e
s
3
an
d
4
f
or
the
DT
a
nd
th
e
GE
AN
T
net
work
to
po
l
og
i
es
co
ns
ide
rin
g
the
CA
GR
va
lues
of
20
%
a
nd
40%,
res
pe
ct
ively
.
Sp
eci
fical
ly
, fr
om
Figu
re
3 i
t
can
be
obse
rv
e
d t
hat
the
HyS
w m
et
ho
d f
or
a
lowe
r val
ue
of
2
N
Sp
le
ads t
o t
he
po
st
ponem
ent
of
la
r
ge
r
am
ou
nt
of
S
pChns
by
approxim
at
e
l
y
9
to
10
ye
ar
s
and
sim
ul
ta
neo
us
ly
ens
ur
es
t
ha
t
the
com
plexity
o
f
the n
et
wor
k
no
des
is l
ow. Fu
r
ther,
for
lo
nger
p
eri
od
s
of tim
e fr
am
e, in
co
m
par
ison
to
t
he
Sp
S
w
m
et
ho
d,
t
his
lo
w
2
N
Sp
value
is
al
so
seen
to
m
ini
m
i
ze
the
act
ive
Sp
Ch
ns
am
ou
nt
since
it
prov
isi
on
s
a
la
ye
re
d
netw
ork
w
hich
c
om
pr
ise
s
of
a
sin
gle
dim
ension
a
nd
gran
ula
rity
of
t
he
f
requen
cy
.
Also,
in
the
la
te
r
pe
rio
ds
of
the ti
m
e fr
am
e o
wi
ng to
t
he
l
ow
value
s
of
Ne
tUsg
,
it
ca
n be
ob
s
er
ved th
at
t
he use
of the
H
ySw m
et
ho
d
re
su
lt
s
in
an
i
ncr
ease
d
act
ive
dim
en
sio
ns
am
ount.
The
a
forem
ent
ion
e
d
is
a
re
sul
t
of
the
fact
that
w
he
n
the
S
pS
w
m
et
ho
d
is
im
pl
e
m
ented
at
the
la
te
r
per
i
ods
of
the
ti
m
e
fr
am
e
,
it
le
ads
t
o
a
c
ha
ng
e
-
ov
e
r
sta
ge
w
her
ei
n,
t
he
HySw
m
et
ho
d res
ults
in the acti
vatio
n of m
or
e Sp
C
hn
s
am
ou
nt i
n com
par
ison t
o
the S
pSw m
et
ho
d.
(a)
(b)
Figure
3. A
gA
cSpC
hn
s
(
le
ft
) a
nd N
et
Us
g
(
ri
gh
t
)
re
su
lt
s
for a
CAG
R
v
al
ue of
20% c
onsid
erin
g
the
n
et
w
ork
t
opologies (a
) DT,
and
(b) G
EANT
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
On the
perf
orma
nce
of swit
chin
g meth
ods i
n
s
p
ac
e
divisi
on
mu
lt
iplex
in
g ba
s
ed… (
Sri
dh
ar
Iyer
)
59
66
(a)
(b)
Figure
4. A
gA
cSpC
hn
s
(
le
ft
) a
nd N
et
Us
g
(
ri
gh
t
)
re
su
lt
s
for a
CAG
R
v
al
ue of
40% c
onsid
erin
g
the
n
et
w
ork
t
opologies (a
) DT,
and
(b) G
EANT
The
af
or
em
entione
d
high
val
ues
of
NetUs
g
pro
vision
e
d
by
S
pSw
for
hea
vily
loade
d
tra
ff
ic
dem
and
s
can
be
see
n
m
or
e
a
pp
a
re
ntly
in
Fi
gure
4.
Al
so
,
f
or
la
te
r
per
iod
s
of
ti
m
e
fr
a
m
e,
irres
pecti
ve
of
Sp
N
value
,
it
i
s
ob
s
er
ved
t
hat
HySw
m
erg
es
to
a
sin
gle
NetUsg
value
w
hich
is
m
uch
higher
tha
n
that
re
ached
by
I
nSw
.
He
nce
,
from
al
l
the
ob
t
ai
ned
re
su
lt
s
it
can
be
in
ferre
d
that
the
re
is
a
tr
ade
-
off
bet
wee
n
t
he
costs
ass
oc
ia
te
d
f
or
act
iv
at
ing
the
S
pChns
a
nd
for
switc
hi
ng.
F
ur
t
her,
t
he
use
of
t
he
HyS
w
m
et
ho
d
with
lo
wer
Sp
N
values
is
able
t
o
offe
r
a
ba
la
nce
bet
wee
n
the
af
or
em
en
ti
on
ed
cost
s.
4.
CONCL
US
I
O
N
In
the
cu
rr
e
nt
work,
we
i
nve
sti
gated
perfor
m
ance
of
t
he
s
witc
hing
s
olu
ti
on
s
wh
ic
h
a
re
app
li
cable
to
SD
M
-
b
-
OT
Ns.
The
resu
lt
s
re
vealed
t
hat
(i)
com
par
ed
to
ot
her
s
witc
hi
ng
s
olu
ti
ons,
FqSw
dem
on
strat
es
lowe
r
netw
ork uti
li
zat
ion
ov
e
r
a l
onger
pe
rio
d of
ti
m
e fr
am
e, an
d (ii
)
S
pSw
ou
t
pe
rfor
m
s I
nSw
f
or
l
onge
r
ti
m
e
fr
am
e
per
i
od
s
si
nce
i
t
is
a
ble
to
a
void
both
,
fr
a
gm
entat
ion
of
s
pectr
um
and
losses
of
capac
it
y
ow
in
g
to
f
r
equ
e
ncy
cl
ashin
g.
F
ur
t
he
r,
res
ults
al
so
dem
on
strat
ed
t
hat
f
or
s
horter
per
i
od
s
of
tim
e
fr
am
es,
Sp
S
w
offer
s
l
ow
net
work
util
iz
at
ion
as
it
act
ivate
s
la
rg
e
am
ou
nts
of
S
pChns.
C
on
si
de
rin
g
t
he
a
f
or
e
m
entioned,
we
pr
opos
e
d
t
he
HyS
w
strat
egy
w
hich
reduces
c
os
ts
associat
ed
with
act
ivati
ng
a
new
Sp
C
hn
as
it
beg
ins
by
im
plem
enting
InSw
a
nd
then
sh
i
fts
to
t
he
us
e
of
FqS
w
a
fter
act
ivati
on
of
spe
ci
fic
nu
m
ber
s
of
SpC
hn
s
.
T
he
res
ul
ts
dem
on
strat
e
d
t
hat
HySw
m
ai
ntain
s
t
he
s
pectra
l
-
eff
ic
ie
ncy
ov
er
a
lo
ng
e
r
ti
m
e
fr
am
e
and
pro
visio
ns
substanti
al
sa
vings
on
the
i
ncurr
e
d
s
w
it
chin
g
costs
.
A
s
a
n
im
po
rtant
obser
vatio
n,
it
is
f
ound
tha
t
in
the
la
te
r
pe
rio
ds
of
ti
m
e
fr
am
e,
us
e
of
HyS
w
re
su
lt
s
in
a
n
i
ncrea
sed
act
ive
di
m
ension
s
am
ou
nt
since
,
us
e
of
S
pSw
at
la
te
r
per
i
od
s
of
ti
m
e
f
ram
e
le
ads
to
a
cha
nge
-
over
sta
ge
wh
e
rein
,
HyS
w
res
ul
ts
in
the
act
ivati
on
of
m
or
e
Sp
C
hn
s
a
m
ou
nt
in
com
par
iso
n
t
o
Sp
S
w.
It
is
al
s
o
obser
ve
d
tha
t
the
us
e
of
Hy
Sw
with
l
ow
e
r
Sp
N
val
ues
is
a
ble
to
offer
a
bala
nce
in
tra
de
-
off
wh
ic
h occu
rs b
et
ween
c
os
ts a
sso
ci
at
ed f
or
a
ct
ivati
ng
t
he
S
pChns a
nd
for swit
chin
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol.
1
, N
o.
2
,
J
ul
y
20
20
:
54
–
6
0
60
REFERE
NCE
S
[1]
E.
Agrel
l
,
et
a
l.
,
“
N.
Gisin.
Road
m
ap
of
op
ti
c
al
c
om
m
unic
at
ions”
,
Journal
o
f
Opt
i
cs,
IOP
Press
,
v
ol.
18,
no.
1
,
pp
.
1
-
40,
2016
.
[2]
S.P.
Singh,
e
t
a
l.
,
“
Stud
y
on
Mit
i
gat
ion
of
Tra
ns
m
ission
Im
pai
rme
nts
and
Iss
ues
a
nd
Challeng
es
w
it
h
PLIA
-
RW
A
i
n
Optic
a
l
W
DM
Networks”,
Journ
al
of
Optic
a
l
Co
mm
unic
ati
on,
De
Gr
uyt
er
,
vol
.
3
3,
no
.
2
,
pp
.
83
-
1
01,
2012
.
[3]
T.
J.
Xi
a,
et
al.
,
“
Introduc
t
ion
of
s
pec
tr
al
l
y
and
spa
ti
all
y
fl
exi
b
le
op
t
ic
a
l
ne
tworks”,
I
EE
E
Comm
uni
c
ati
ons
Magazine
,
vol.
53
,
no
.
2
,
pp
.
24
–
33
,
2015
.
[4]
D.
Klonid
is,
e
t
al
.
,
“
Spect
r
all
y
and
Spa
ti
a
lly
Flex
ibl
e
Opt
ical
Network
Pla
nning
and
Ope
rat
ions”
,
IEEE
Comm
unic
ati
ons Magazine
,
vol
.
53,
no
.
2
,
pp
.
69
-
78,
2015
.
[5]
P.J
.
W
inz
er
“
Spatial
m
ultiple
x
in
g
in
f
ibe
r
optics:
Th
e
10×
sca
li
n
g
of
m
et
ro/
cor
e
ca
pa
ci
t
ie
s”
,
Be
l
l
Labs
Techni
ca
l
Journal
,
vo
l. 19, pp. 22
–
30,
2014
.
[6]
G.
M.
Sar
idi
s,
et
al.
,
“
Surve
y
and
eva
lu
at
ion
of
spa
ce
d
ivi
sion
m
ulti
ple
xing:
From
technologies
to
op
ti
c
al
n
et
works
”,
IEE
E
Comm
unications
Surve
ys
a
nd
Tutor
ial
s
,
vo
l
.
17
,
no
.
4
,
pp
.
2
136
–
2156,
2015
.
[7]
P.J.
W
inz
e
r,
et
a
l.
,
“
From
sca
li
ng
dispariti
es
to
in
te
gra
te
d
par
al
l
el
i
sm
:
A
decat
hlon
for
a
decade”,
I
EE
E
Journal
o
f
Lightwav
e
and
T
ec
hnolog
y
,
pp.
3
5,
no
.
5
,
pp
.
109
9
–
1
115,
2017
.
[8]
D.M.
Marom
,
et
al.
,
“
Sw
it
chi
ng
soluti
ons
for
W
DM
SD
M
opti
ca
l
n
et
works
”,
IE
EE
Comm
unic
ati
ons
Magazine
,
vo
l
.
53,
no
.
2
,
pp
.
60
–
68,
2015
.
[9]
F.J.
Moreno
-
M
uro,
et
al
.
,
“
Eva
luation
of
Cor
e
-
Conti
nui
t
y
-
Co
nstrai
ned
ROA
DM
s
for
Flex
-
Grid/MCF
Opti
ca
l
Networks”
,
I
EEE/OSA Journal of
Opti
cal
Com
municat
ions a
nd
Net
work
ing
,
vol
.
9
,
no
.
11
,
pp
.
1
041
-
1050,
2017
.
[10]
D.M.
Marom
,
et
al.
,
“
Surve
y
of
Photonic
Sw
it
c
hing
Archi
tectur
es
and
T
ec
hnolo
gie
s
in
Support
of
Spatially
a
nd
Spect
ra
lly
Fl
exible
Opt
ic
a
l
Ne
tworking”
,
IE
EE
/O
SA
Journal
of
O
pti
cal
Comm
unic
ati
ons
and
Netw
orking
,
vo
l.
9,
no
.
9,
pp
.
1
-
26
,
201
7.
[11]
N.
K.
Font
ai
ne
,
et
a
l.
,
“
H
et
ero
g
e
neous
spac
e
-
d
ivis
ion
m
ult
ipl
ex
in
g
and
joi
nt
wave
le
ngth
sw
it
chi
ng
demons
tra
ti
on
”
,
in
IE
EE
/OSA
Optic
al
F
ibe
r Co
mm
unic
ati
ons C
onf.
(
OFC)
,
pp.
1
-
3,
2015
.
[12]
P.
S.
Khodashe
nas,
e
t
a
l.
,
“
Co
m
par
ison
of
Spect
r
al
and
Spa
tial
Super
cha
nn
el
Allocati
on
Sch
emes
for
SD
M
Networks”,
I
EEE/OSA Journal of
Lightwave
Te
chnol
ogy
,
vo
l. 3
4,
no
.
11
,
pp
.
27
10
–
2716,
2017
.
[13]
D.
Sirac
usa
,
e
t
al.
,
“
Resourc
e
a
ll
oc
at
ion
po
li
c
ies
in
SD
M
opti
ca
l
ne
tworks”,
in
IEE
E
Inte
rnat
i
onal
Confe
ren
c
e
Optic
al
Net
work
Design
and
Mo
del
ing
(
OND
M
)
,
pp.
168
–
173,
20
15.
[14]
B.
Shari
at
i
,
et
al.
,
“
Im
pac
t
of
spat
ia
l
and
spe
ct
r
al
gra
nula
r
ity
on
th
e
per
form
an
ce
o
f
SD
M
net
works
base
d
on
spat
ial
superc
hannel
sw
it
chi
ng
”,
IE
EE
/
OSA
Journal
o
f L
ight
wave Tec
h
nology
,
vol
.
35
,
no.
13
,
pp
.
2559
–
2568,
2017
.
[15]
C.
Rott
ond
i,
et
al
.
,
“
Routi
ng
,
m
od
ula
ti
on
form
at
,
b
aud
ra
te
and
spe
c
trum
al
lo
ca
t
ion
i
n
opti
c
al
m
et
ro
r
i
ngs
with
fl
exi
bl
e
grid
and
few
-
m
ode
tr
ansm
ission”,
IE
EE J
ournal
o
f
Lightwave
Te
c
hnology
,
vol
.
35
,
no.
1,
pp.
61
–
70
,
2017
.
[16]
A.
Muham
m
ad,
et
al
.
,
“
Resourc
e
al
lo
cation
for
spac
e
-
d
ivi
sion
m
ult
iplexing:
optical
whi
te
box
ve
rsus
opti
ca
l
bl
ack
box
net
working
”
,
IE
EE J
ournal
o
f
Lightwave
Te
c
hnology
,
v
ol
.
33
,
no.
23,
pp.
4928
–
4941,
2015
.
[17]
M.
Fiorani
,
et
a
l
.
,
“
Spatial
div
isi
on
m
ult
ipl
exi
ng
for
high
ca
p
acit
y
opt
ical
in
te
rc
o
nnec
ts
in
m
odul
ar
data
c
ent
ers”
,
IEE
E
/OSA
Jo
urnal
of
Optic
a
l
C
omm
unic
ati
ons
and
Net
w
ork
ing
,
vol. 9, no. 2, pp. A143
–
A153,
20
17.
[18]
J.
Pere
ll
ó
,
et
a
l.
,
“
Flex
-
grid
/SDM
bac
kbone
net
work
d
esign
with
interc
or
e
XT
-
l
imite
d
trans
m
ission
rea
ch
”,
IEE
E
/OSA
Jo
urnal
of
Optic
a
l
C
omm
unic
ati
ons
and
Net
w
ork
ing
,
vol. 8, no. 8, pp. 540
–
552,
2016.
.
[19]
S.
I
y
er
,
et
a
l.
,
“
Spect
ra
l
and
Pow
er
-
Eff
icien
c
y
I
nvesti
gation
in
Single
and
Mult
i
-
Li
n
e
-
R
at
e
Op
ti
c
al
W
ave
l
ength
Division
Multi
pl
exe
d
(W
DM
)
Networks”,
Phot
on
ic
s
Net
work
Co
mm
unic
ati
ons,
S
pringer
,
vol.
33,
no.
1,
pp
.
39
-
5
1
,
2017.
[20]
S.
I
y
er
,
“
On
R
outi
ng,
Modula
t
ion
Form
at
,
Sp
ac
e
and
Spe
ct
ru
m
Alloc
a
ti
on
with
Prot
ection
i
n
Spac
e
Div
ison
Multi
ple
x
ing
ba
sed
E
la
sti
c
Opti
ca
l
Ne
tworks”,
J
ournal
of
In
formation
and
Teleco
mm
unic
ati
on,
T
ayl
or
&
Franci
s
,
vol.
4
,
no
.
1
,
pp
.
105
-
117,
2020
.
[21]
S.
I
y
er,
“
An
Online
Rout
ing
Algorit
hm
for
Space
Divis
ion
Multi
ple
x
ing
Based
Elastic
Optic
a
l
Networ
ks”,
Inte
rnational
Jo
urnal
of
Comm
u
nic
ati
on
Ne
tworks
and
Distribut
ed
Syst
ems,
Ind
e
rs
ci
enc
e
,
vo
l.
24
,
no
.
2
,
pp
.
167
-
185,
2020
.
[22]
Cisco
Visua
l
Networki
ng
Ind
ex:
Forec
ast
an
d
Methodo
log
y
,
2016
–
2021,
W
hit
e
p
ape
r
,
2
01
7.
h
tt
ps:/
/www
.
ci
sco
.
com/c
/e
n/
us/soluti
ons/coll
at
er
al
/serv
ice
-
pr
ovide
r/vi
su
al
ne
t
working
-
inde
x
-
v
ni/
complete
-
whi
te
-
pap
er
-
c11
-
481360.
pdf
Evaluation Warning : The document was created with Spire.PDF for Python.