zhonghong ou what is simulation? computer simulation · simulation of data communications networks...
TRANSCRIPT
Sim
ula
tio
n o
f d
ata
co
mm
un
icati
on
s n
etw
ork
s
T-1
10
.61
30
Sys
tem
s E
ng
ine
eri
ng
in
T
-11
0.6
13
0 S
ys
tem
s E
ng
ine
eri
ng
in
D
ata
Co
mm
un
ica
tio
ns
So
ftw
are
������������
���������� ��
�������������������������������
Aa
lto
Un
ive
rsit
y
Zh
on
gh
on
g O
u9
/17
/2010
Ag
en
da
What
When
Zh
on
gh
on
g O
u2
Why
Where
How
Who
Wh
at
is s
imu
lati
on
?
•S
imula
tion is t
he im
itation o
f som
e r
eal th
ing,
sta
te o
f
affairs, or
pro
cess. (w
ikip
ed
ia)
•S
imula
tion is a
n im
itation o
f a r
eal w
ork
pro
cess o
r
syste
m o
ver
tim
e.
syste
m o
ver
tim
e.
3Z
ho
ng
ho
ng
OuH
um
an-in-t
he-loop s
imula
tion o
f oute
r space.
(wik
ipedia
)
Exam
ple
4Z
ho
ng
ho
ng
Ou
Hum
an-in-t
he-loop s
imula
tion o
f oute
r space.
(wik
ipedia
)
Key c
hara
cte
risti
cs
•A
cquis
itio
n o
f valid
sourc
e info
rmation a
bout
the
rele
vant
sele
ction o
f key c
hara
cte
ristics a
nd b
ehavio
rs.
•U
se o
f sim
plif
yin
gappro
xim
ations a
nd a
ssum
ptions
within
the s
imula
tion.
within
the s
imula
tion.
–th
e a
ssu
mp
tio
ns u
su
ally
ta
ke
th
e fo
rm o
f m
ath
em
atica
l or
log
ica
l re
latio
ns.
•F
idelit
y a
nd v
alid
ity o
f th
e s
imula
tion o
utc
om
es.
5Z
ho
ng
ho
ng
Ou
Co
mp
ute
r sim
ula
tio
n
•A
co
mp
ute
r p
rog
ram
, o
r n
etw
ork
of
co
mp
ute
rs,
tha
t a
tte
mp
ts
to s
imu
late
an
ab
str
act
mo
de
l o
f a
pa
rtic
ula
r syste
m.
•B
ee
n u
se
d in
:–
Na
tura
l syste
ms
•P
hysic
s (
co
mp
uta
tio
na
l p
hysic
s),
•A
str
op
hysic
s,
•A
str
op
hysic
s,
•C
he
mis
try,
•B
iolo
gy.
–H
um
an
syste
ms
•E
co
no
mic
s,
•P
sych
olo
gy,
•S
ocia
l scie
nce
,
•E
ng
ine
eri
ng
.
6Z
ho
ng
ho
ng
Ou
Cla
ssif
icati
on
(co
ars
e-g
rain
ed
)
•P
hysic
al sim
ula
tio
n r
efe
rs t
o s
imu
latio
n in
wh
ich
th
e r
ea
l th
ing
is
su
bstitu
ted
fo
r p
hysic
al o
bje
cts
, w
hic
h a
re o
fte
n c
ho
se
n
be
ca
use
th
ey a
re s
ma
ller
or
ch
ea
pe
r th
an
th
e a
ctu
al o
bje
ct
or
syste
m.
•In
tera
ctive
sim
ula
tio
n is a
sp
ecia
l kin
d o
f p
hysic
al sim
ula
tio
n,
oft
en
re
ferr
ed
to
as a
hu
ma
n in
th
e lo
op
sim
ula
tio
n,
in w
hic
h
oft
en
re
ferr
ed
to
as a
hu
ma
n in
th
e lo
op
sim
ula
tio
n,
in w
hic
h
ph
ysic
al sim
ula
tio
ns in
clu
de
hu
ma
n o
pe
rato
rs,
su
ch
as in
a
flig
ht sim
ula
tor
or
a d
rivin
g s
imu
lato
r.
–H
um
an
in
th
e lo
op
sim
ula
tio
ns c
an
in
clu
de
a c
om
pu
ter
sim
ula
tio
n
as a
so
-ca
lled
syn
the
tic e
nvir
on
me
nt;
–H
ard
wa
re in
th
e lo
op
sim
ula
tio
n;
–S
em
i-p
hysic
al sim
ula
tio
n;
–H
alf-o
bje
ct
sim
ula
tio
n.
7Z
ho
ng
ho
ng
Ou
Cla
ssif
icati
on
(fi
ne-g
rain
ed
)
•D
ete
rmin
acy:
–S
toch
astic
(ra
nd
om
)•
There
exis
ts s
om
e in
dete
rmin
acy in
its
futu
re e
volu
tion d
escribed b
y p
robabili
ty
dis
trib
utions;
•E
ven if th
e in
itia
l conditio
n (
or
sta
rtin
g p
oin
t) is
know
n, th
ere
are
many p
ossib
ilities the
pro
cess m
ight go to, but som
e p
ath
s m
ay b
e m
ore
pro
bable
and o
thers
less s
o.
•E
.g. sto
ck m
ark
et,
exchange r
ate
.
–D
ete
rmin
istic
•G
iven a
part
icula
r in
put,
it w
ill a
lways p
roduce the s
am
e o
utp
ut;
•G
iven a
part
icula
r in
put,
it w
ill a
lways p
roduce the s
am
e o
utp
ut;
•E
.g. m
ath
em
atical f
unction.
•S
tea
din
ess:
–S
tatic (
ste
ad
y)
•R
ela
ting to a
giv
en insta
nt of tim
e, tim
e n
ot consid
ere
d;
•U
sed for
estim
ation o
f q
uantities fro
m a
giv
en d
istr
ibution;
–D
yn
am
ic•
Develo
pm
ent of th
e s
yste
m b
ased o
n tim
e;
•O
utp
ut changes in
a s
yste
m in r
esponse to (
usually
changin
g)
input sig
nals
.
8Z
ho
ng
ho
ng
Ou
Cla
ssif
icati
on
(fi
ne-g
rain
ed
, co
nt.
)
•C
on
tin
uity:
–C
ontinuous
•In
math
em
atics,
a c
ontinuous f
unction is a
function f
or
whic
h,
intu
itiv
ely
, sm
all
changes in t
he
input
result in s
mall
changes in t
he o
utp
ut.
•C
onsid
ering a
function h
(t)
whic
h d
escribes t
he h
eig
ht
of a g
row
ing f
low
er
with r
espect
to t.
•“A
nalo
g”
syste
m.
–D
iscre
te•
The o
bje
cts
stu
die
d in d
iscre
te s
imula
tion–
such a
s inte
gers
, gra
phs,
and s
tate
ments
in logic
–do n
ot
vary
sm
ooth
ly,
but
have d
istinct, s
epara
ted v
alu
es.
•“D
igital” s
yste
m.
•S
pecia
l case:
dis
cre
te e
vent
sim
ula
tion (
DE
S)
–M
anag
ing
events
in t
ime. T
he s
imula
tor
main
tain
s a
queue o
f events
sort
ed b
y t
he s
imula
ted tim
e they
should
occur,
reads the q
ueue a
nd trig
gers
new
events
as e
ach e
vent is
pro
cessed. It
is n
ot im
port
ant
to e
xecute
the s
imula
tion in r
eal tim
e, but ra
ther
to b
e a
ble
to a
ccess the d
ata
pro
duced b
y t
he
sim
ula
tion, to
dis
cover
log
ic d
efe
cts
in t
he d
esig
n e
tc. M
ost com
pute
r, lo
gic
-test and f
ault-t
ree
sim
ula
tions a
re o
f th
is type.
•L
oca
lity:
–D
istr
ibute
d•
Usin
g a
netw
ork
of
inte
rconnecte
d c
om
pute
rs t
o a
ccom
plis
h a
com
mon o
bje
ctive o
r ta
sk;
•S
imula
tions d
ispers
ed a
cro
ss m
ultip
le h
ost
com
pute
rs.
–Local
•U
sin
g a
sin
gle
com
pute
r to
conduct
the s
imula
tion.
9Z
ho
ng
ho
ng
Ou
Cla
ssif
icati
on
(fi
ne-g
rain
ed
, co
nt.
)
Sim
ula
tion
Dete
rmin
acy
Ste
adin
ess
Continuity
Localit
y
10
Zh
on
gh
on
g O
u
Sto
chastic
Dete
rmin
istic
Sta
tic
Dyn
am
ic
Continuous
Dis
cre
te
Dis
trib
ute
dLocal
Ag
en
da
What
When
Zh
on
gh
on
g O
u1
1
Who
Where
How
Why
Wh
en
to
use s
imu
lati
on
?
•M
od
elin
go
f n
atu
ral syste
ms o
r h
um
an
syste
ms in
ord
er
to g
ain
in
sig
ht in
to th
eir
fu
nctio
nin
g.
•Im
ita
tin
gte
ch
no
log
y fo
r p
erf
orm
an
ce
op
tim
iza
tio
n, sa
fety
e
ng
ine
eri
ng
, te
stin
g, tr
ain
ing
an
d e
du
ca
tio
n, e
xp
lori
ng
an
d g
ain
ing
ne
w in
sig
hts
in
to n
ew
te
ch
no
log
y.
•S
ho
win
gth
e e
ve
ntu
al re
al e
ffe
cts
of a
lte
rna
tive
co
nd
itio
ns a
nd
co
urs
es o
f a
ctio
n.
•E
stim
atin
gth
e p
erf
orm
an
ce o
f syste
ms to
o c
om
ple
x fo
r a
na
lytica
l so
lutio
ns. T
he
re
al syste
m c
an
no
t b
e e
ng
ag
ed
fo
r re
aso
ns, e
.g. it
ma
y n
ot b
e a
cce
ssib
le, it m
ay b
e d
an
ge
rou
s o
r u
na
cce
pta
ble
to
e
ng
ag
e, o
r it m
ay s
imp
ly n
ot e
xis
t.
12
Zh
on
gh
on
g O
u
Ag
en
da
What
When
Zh
on
gh
on
g O
u1
3
Who
Where
How
Why
An
aly
sis
vers
us s
imu
lati
on
•T
rad
itio
na
lly,
the
fo
rma
l m
od
elin
g o
f syste
ms h
as b
ee
n v
ia a
ma
the
ma
tica
l m
od
el, w
hic
h a
tte
mp
ts t
o f
ind
an
aly
tica
l
so
lutio
ns e
na
blin
g t
he
pre
dic
tio
n o
f th
e b
eh
avio
r o
f th
e
syste
m f
rom
a s
et
of
pa
ram
ete
rs a
nd
in
itia
l co
nd
itio
ns.
•C
om
pu
ter
sim
ula
tio
n is o
fte
n u
se
d a
s a
n a
dju
nct
to,
or
su
bstitu
tio
n fo
r, m
od
elin
g s
yste
ms f
or
wh
ich
sim
ple
clo
se
d
form
an
aly
tic s
olu
tio
ns
are
no
t p
ossib
le.
–C
om
mo
n f
ea
ture
: a
tte
mp
tin
g t
o g
en
era
te a
sa
mp
le o
f
rep
rese
nta
tive
sce
na
rio
s f
or
a m
od
el in
wh
ich
a c
om
ple
te
en
um
era
tio
n o
f a
ll p
ossib
le s
tate
s w
ou
ld b
e p
roh
ibitiv
e o
r
imp
ossib
le.
14
Zh
on
gh
on
g O
u
An
aly
sis
vers
us s
imu
lati
on
(co
nt.
)
•A
na
lysis
(tw
oste
ps):
–M
od
elin
g o
f th
e s
yste
m a
s a
tim
e-d
ep
en
de
nt
sto
ch
astic p
roce
ss;
–A
na
lytica
l so
lutio
n o
f th
e m
od
el.
•S
imu
latio
n (
fou
rste
ps):
–M
od
elin
g o
f th
e s
yste
m a
s a
dyn
am
ic s
toch
astic p
roce
ss;
–G
en
era
tin
g r
ea
liza
tio
ns o
f th
e p
roce
ss;
–C
olle
ctin
g d
ata
(m
ea
su
rem
en
t);
–S
tatistica
lly a
na
lyzin
g t
he
da
ta a
nd
dra
win
g c
on
clu
sio
ns.
•C
om
mo
n f
ea
ture
s:
–M
od
elin
g is c
om
mo
n•
Diff
eri
ng
with
re
sp
ect
to d
eta
ils;
•M
ath
em
atica
l a
na
lysis
usu
ally
utiliz
ing
re
str
ictive
assu
mp
tio
ns.
15
Zh
on
gh
on
g O
u
Math
em
ati
cal an
aly
sis
•P
ros:
–R
esu
lts o
bta
ine
d q
uic
kly
;
–R
esu
lts e
xa
ct;
–G
ivin
g in
sig
ht o
f th
e s
yste
m;
–A
llow
ing
op
tim
iza
tio
n.
•C
ons:
–R
eq
uir
ing
re
str
ictive
assu
mp
tio
ns;
–A
na
lysis
co
mp
lex;
–R
esu
lts (
mig
ht b
e)
limite
d to
eq
uili
bri
um
sta
te, o
r a
ve
rag
e
va
lue
s.
16
Zh
on
gh
on
g O
u
Sim
ula
tio
n
•P
ros:
–N
o c
on
str
ain
ts in
th
e m
od
el b
uild
ing
;
–E
na
blin
g c
om
ple
x s
yste
m;
–M
od
elin
g u
su
ally
str
aig
htfo
rwa
rd.
•C
ons:
•C
ons:
–T
ime
-a
nd
en
erg
y-c
on
su
min
g;
–R
esu
lts im
pre
cis
e (
pre
cis
ion
co
uld
be
im
pro
ve
d b
y m
ultip
le
ite
ratio
ns)
–G
ettin
g in
sig
ht m
ore
diff
icu
lt;
–O
ptim
iza
tio
n m
ore
diff
icu
lt (
ma
yb
e lim
ite
d to
th
e tri
al o
f a
fe
w
pa
ram
ete
r co
mb
ina
tio
ns).
17
Zh
on
gh
on
g O
u
Ag
en
da
What
When
Zh
on
gh
on
g O
u1
8
Who
Where
How
Why
Co
nd
ucto
r o
f sim
ula
tio
n
•Y
ou.
•O
ther
co-o
pera
tors
.
19
Zh
on
gh
on
g O
u
Ag
en
da
What
When
Zh
on
gh
on
g O
u2
0
Who
Where
How
Why
Pla
ce f
or
sim
ula
tio
n
•S
ingle
com
pute
r.
•C
luste
r of com
pute
rs.
21
Zh
on
gh
on
g O
u
Ag
en
da
What
When
Zh
on
gh
on
g O
u2
2
Who
Where
How
Why
Ho
w t
o b
uild
sim
ula
tio
n m
od
el
•C
alib
ration.
•V
erification.
•V
alid
ation
•V
alid
ation
23
Zh
on
gh
on
g O
u
Mo
del calib
rati
on
•C
an
be
ach
ieve
d b
y a
dju
stin
g a
ny a
va
ilab
le p
ara
me
ters
in
o
rde
r to
ad
just
ho
w t
he
mo
de
l o
pe
rate
s a
nd
sim
ula
tes t
he
p
roce
ss.
•F
or
exa
mp
le,
in t
he
sim
ula
tio
n o
f a
pe
er-
to-p
ee
r (P
2P
) n
etw
ork
, th
e t
yp
ica
l p
ara
me
ters
in
clu
de
jo
inin
g r
ate
, le
avin
g
ne
two
rk,
the
typ
ica
l p
ara
me
ters
in
clu
de
jo
inin
g r
ate
, le
avin
g
rate
, (c
hu
rn r
ate
), e
xch
an
gin
g r
ate
, p
ub
lish
ing
ra
te,
loo
ku
p
rate
etc
.
•T
he
se
pa
ram
ete
rs in
flu
en
ce
th
e b
eh
avio
rs o
f th
e P
2P
n
etw
ork
, fo
r in
sta
nce
, lo
oku
p s
ucce
ss r
atio
, a
ve
rag
e t
raffic
lo
ad
(b
yte
s),
an
d a
ve
rag
e n
um
be
r o
f m
essa
ge
s.
24
Zh
on
gh
on
g O
u
Mo
del veri
ficati
on
•C
an
be
ach
ieve
d b
y o
bta
inin
g o
utp
ut
da
ta f
rom
th
e m
od
el a
nd
co
mp
ari
ng
it
to w
ha
t is
exp
ecte
d f
rom
th
e in
pu
t d
ata
.
•F
or
exa
mp
le,
in t
he
sim
ula
tio
n o
f a
pe
er-
to-p
ee
r (P
2P
) n
etw
ork
, th
ere
a
re c
ert
ain
exp
ecta
tio
ns w
ith
re
ga
rd t
o t
he
lo
oku
p s
ucce
ss r
atio
, g
ive
n
the
ch
urn
ra
te,
exch
an
gin
g r
ate
etc
.
•In
so
me
ca
se
s t
he
syste
m o
r so
me
pa
rt o
f it c
an
be
an
aly
ze
d u
nd
er
sim
plif
ied
assu
mp
tio
ns.
•T
he
sim
ula
tio
n c
an
be
ru
n u
nd
er
the
sa
me
assu
mp
tio
ns;
at
lea
st
the
re
su
lts s
ho
uld
ma
tch
with
ea
ch
oth
er.
•In
re
al-
life
, th
is is u
su
ally
no
t so
re
alis
tic a
s it
is n
ot
ea
sy t
o g
et
the
th
eo
retica
l re
su
lts g
ive
n t
he
in
pu
t d
ata
.
25
Zh
on
gh
on
g O
u
Mo
del valid
ati
on
•C
an
be
ach
ieve
d b
y c
om
pa
rin
g t
he
re
su
lts w
ith
wh
at’s
exp
ecte
d b
ase
d o
n h
isto
rica
l d
ata
fro
m t
he
stu
dy a
rea
.
•It
is th
e b
est
an
d m
ost
relia
ble
me
tho
d.
•Id
ea
lly,
the
mo
de
l sh
ou
ld p
rod
uce
sim
ilar
resu
lts t
o w
ha
t h
as
•Id
ea
lly,
the
mo
de
l sh
ou
ld p
rod
uce
sim
ilar
resu
lts t
o w
ha
t h
as
ha
pp
en
ed
his
tori
ca
lly.
•If
mo
de
l o
utp
ut
va
lue
s a
re d
rastica
lly d
iffe
ren
t th
an
his
tori
ca
l va
lue
s, it p
rob
ab
ly m
ea
ns t
he
re’s
an
err
or
in t
he
mo
de
l.
•O
fte
n d
ifficu
lt to
ap
ply
, e
.g.
the
re is n
o r
ea
l syste
m,
or
me
asu
rem
en
ts a
re t
oo
exp
en
siv
e t
o c
on
du
ct.
26
Zh
on
gh
on
g O
u
Wh
at
if…
•N
o s
implif
ied m
ath
em
atical syste
m a
vaila
ble
, neither
no
his
torical m
easure
ments
exis
ting…
•E
xpert
intu
itio
n
–a
co
mm
on
an
d p
ractica
l m
eth
od
;–
a c
om
mo
n a
nd
pra
ctica
l m
eth
od
;
–“b
rain
sto
rmin
g”
with
pe
op
le w
ho
kn
ow
th
e s
yste
m in
ord
er
to
de
fin
e s
en
sib
le a
ssu
mp
tio
ns a
nd
in
pu
t d
ata
;
–a
n e
xp
ert
ca
n e
asily
re
co
gn
ize
“im
po
ssib
le”
resu
lts;
–n
ot so
re
liab
le.
27
Zh
on
gh
on
g O
u
Sele
cti
on
of
lan
gu
ag
es
Genera
l purp
ose languages
(C/C
++
, Java…
)
•M
ost users
have k
now
ledge o
f at le
ast
one la
nguages;
•A
vaila
ble
on m
ost com
pute
rs;
•C
ode e
asily
tra
nsport
ed;
•Low
cost o
f th
e p
rogra
ms;
•C
ode r
unnin
g faste
r;
Sim
ula
tio
nla
ng
ua
ge
s (
ge
ne
ral
pu
rpo
se
,G
AS
P, G
PS
S,
SIM
SC
RIP
T, S
imu
la)
•S
upport
ing m
any f
eatu
res n
eeded in the
pro
gra
mm
ing
of
a s
imula
tion m
odel;
•S
hort
er
develo
pm
ent
tim
e;
•Low
er
cost
•P
rogra
mm
ing w
ith
the a
id o
f th
e m
odelin
g
constr
ucts
of
the language
.•
Code r
unnin
g faste
r;
•F
lexib
le.
•R
equires a
lot of pro
gra
mm
ing w
ork
;
•S
usceptible
to e
rrors
28
Zh
on
gh
on
g O
u
Sim
ula
tio
n la
ng
ua
ge
s (
da
ta
co
mm
un
ica
tio
ns n
etw
ork
o
rie
nte
d,
OP
NE
T, Q
ua
lNe
t,
NS
2, O
MN
ET
++
)
•C
onta
inin
g n
etw
ork
build
ing b
locks;
•D
evelo
ped s
pecific
ally
for
the s
imula
tion
of data
com
munic
atio
ns n
etw
ork
s.
Sele
cti
on
of
sim
ula
tors
•ns-2
;
•ns-3
;
•G
loM
oS
im;
•O
PN
et;
•Q
ualN
et;
•O
MN
et+
+;
•N
AB
;
•J-S
im;
•S
imP
y;
•N
etH
aw
k E
AS
T;
•N
etH
aw
k E
AS
T;
•S
EN
SE
;
•S
idh;
•T
OS
SIM
;
•A
TE
MU
;
•A
rvora
;
•E
mS
tar;
•M
AT
LA
B.
29
Zh
on
gh
on
g O
u
ns-2
•T
he d
e f
acto
sta
ndard
for
netw
ork
sim
ula
tion.
Its b
ehavio
r is
hig
hly
tru
ste
d w
ithin
the n
etw
ork
ing
com
munity.
•It w
as d
evelo
ped a
t IS
I/U
SC
(Info
rmatio
n S
cie
nce I
nstitu
te,
Univ
ers
ity o
f S
outh
ern
Calif
orn
ia),
and
was s
upport
ed b
y t
he D
AR
PA
(Defe
nse A
dvanced R
esearc
h P
roje
cts
Agency)
and N
SF
(National
Scie
nce F
oundation).
•ns-2
is a
dis
cre
te-e
ve
nt
sim
ula
tor
org
aniz
ed a
ccord
ing t
o t
he O
SI
model and p
rim
arily
desig
ned t
o
sim
ula
te w
ired n
etw
ork
s.
•T
he s
upport
for
wirele
ss n
etw
ork
ing h
ad b
een b
rought
by s
evera
l exte
nsio
ns.
•T
he s
upport
for
wirele
ss n
etw
ork
ing h
ad b
een b
rought
by s
evera
l exte
nsio
ns.
•S
imula
tions a
re b
ased o
n a
com
bin
ation o
f C
++
and O
tcl(o
bje
ct
oriente
d e
xte
nsio
n o
f Tclcre
ate
d
by D
avid
Weth
era
llat
MIT
). I
n g
enera
l, C
++
is u
sed f
or
imple
menting p
roto
cols
and e
xte
ndin
g t
he
ns-2
lib
rary
. O
Tclis
used t
o c
reate
and c
ontr
ol th
e s
imula
tion e
nvironm
ent
itself,
inclu
din
g t
he
sele
ction o
f outp
ut
data
. S
imula
tion is r
un a
t th
e p
acket
level, a
llow
ing f
or
deta
iled r
esults.
•T
his
desig
n c
hoic
e w
as o
rigin
ally
made t
o a
void
unnecessary
recom
pila
tions if changes a
re m
ade
to the s
imula
tion s
et-
up. T
he d
esig
n o
f ns-2
tra
des o
ff s
imula
tion p
erf
orm
ance f
or
the s
avin
g o
f re
com
pila
tions,
whic
h is q
uestionable
if one is inte
reste
d in c
onducting s
cala
ble
netw
ork
sim
ula
tions. T
here
fore
, ns-2
is c
urr
ently u
nderg
oin
g a
majo
r re
desig
n,
one o
f th
e m
ain
develo
pm
ent
goals
of
its s
uccessor, n
s-3
, is
the im
pro
vem
ent
of
sim
ula
tion p
erf
orm
ance
.
30
Zh
on
gh
on
g O
u
ns-2
(co
nt.
)
•W
eaknesses:
–L
ack o
f m
od
ula
rity
;
–In
he
ren
t co
mp
lexity (
ns-2
wa
s c
an
did
ate
to
be
th
e b
asis
fo
r th
e Q
ua
lne
tsim
ula
tor
bu
t g
ot fin
ally
re
jecte
d);
–H
igh
co
nsu
mp
tio
n o
f co
mp
uta
tio
na
l re
so
urc
es. A
ha
rmfu
l co
nse
qu
en
ce
is th
at n
s-2
la
cks s
ca
lab
ility
, w
hic
h im
pe
de
s
the
sim
ula
tio
n o
f la
rge
ne
two
rks (
ns-2
is typ
ica
lly u
se
d fo
r sim
ula
tio
ns c
on
sis
tin
g o
f n
o m
ore
th
an
a fe
w h
un
dre
ds
no
de
s).
31
Zh
on
gh
on
g O
u
ns-3
•Lik
e its
pre
decessor, n
s-3
relie
s o
n C
++
for
the im
ple
menta
tio
n o
f th
e s
imula
tion m
odels
.
•H
ow
ever, n
s-3
no longer
uses o
Tclscripts
to c
ontr
ol th
e s
imula
tion,
thus a
bandonin
g t
he p
roble
ms
whic
h w
ere
intr
oduced b
y t
he c
om
bin
ation o
f C
++
and o
Tclin
ns-2
.
•In
ste
ad,
netw
ork
sim
ula
tions in n
s-3
can b
e im
ple
mente
d in p
ure
C+
+,
while
part
s o
f th
e s
imula
tion
optionally
can b
e r
ealiz
ed u
sin
g P
yth
on
as w
ell.
•M
ore
over,
ns-3
inte
gra
tes a
rchitectu
ral concepts
and c
ode f
rom
GT
NetS
, a s
imula
tor
with g
ood
scala
bili
ty c
hara
cte
ristics.
•T
hese d
esig
n d
ecis
ions w
ere
made a
t expense o
f com
patibili
ty.
In f
act, n
s-2
models
need t
o b
e
port
ed t
o n
s-3
in a
manualw
ay.
•B
esid
es p
erf
orm
ance im
pro
vem
ents
, th
e f
eatu
re s
et
of
the s
imula
tor
is a
lso a
bout
to b
e e
xte
nded.
For
exam
ple
, ns-3
is s
late
d t
o s
upport
the inte
gra
tion o
f re
al im
ple
menta
tion
s’ code b
y p
rovid
ing
sta
ndard
AP
Is,
such a
s B
erk
ele
y s
ockets
or
PO
SIX
thre
ads,
whic
h a
re t
ranspare
ntly m
apped t
o
the s
imula
tion.
32
Zh
on
gh
on
g O
u
Glo
Mo
Sim
•G
loM
oS
imw
as d
evelo
ped in 1
998 a
t U
CLA
(U
niv
ers
ity o
f C
alif
orn
ia,
Los A
ngele
s)
for
mobile
w
irele
ss n
etw
ork
s.
•It is w
ritten in P
ars
ec,
whic
h is a
n e
xte
nsio
n o
f C
for
para
llel pro
gra
mm
ing,
and b
enefits
fro
m
Pars
ec’s
abili
ty t
o r
un o
n s
hare
d-m
em
ory
sym
metr
ic p
rocessor
(SM
P)
com
pute
rs.
New
pro
tocols
and m
odule
s m
ust
be w
ritten in P
ars
ec t
oo.
•R
espects
OS
I m
odel. G
loM
oS
imis
desig
ned t
o b
e e
xte
nsib
le,
with a
ll pro
tocols
im
ple
mente
d a
s
module
s in the G
loM
oS
imlib
rary
.
•C
apable
of
support
ing p
ara
llel environm
ent. T
he n
etw
ork
is s
plit
into
diffe
rent
sub-n
etw
ork
s,
each
of th
em
bein
g s
imula
ted b
y d
istinct
pro
cessors
. T
he n
etw
ork
is p
art
itio
ned in s
uch a
way t
hat
the
num
ber
of
nodes s
imula
ted b
y e
ach p
art
itio
n is h
om
ogeneou
s.
num
ber
of
nodes s
imula
ted b
y e
ach p
art
itio
n is h
om
ogeneou
s.
•G
loM
oS
imuses a
n o
bje
ct-
oriente
d a
ppro
ach.
How
eve
r, t
he d
esig
ners
realiz
ed t
hat
a p
ure
ly
obje
ct-
oriente
d a
ppro
ach w
ould
not
be s
cala
ble
. In
ste
ad,
Glo
MoS
impart
itio
ns t
he n
odes,
and
each o
bje
ct
is r
esponsib
le f
or
runnin
g o
ne layer
in t
he p
roto
col sta
ck o
f every
node f
or
its g
iven
part
itio
n. T
his
help
s t
o e
ase t
he o
verh
ead o
f a larg
e n
etw
ork
.
•W
eaknesses:
–W
hile
eff
ective for
sim
ula
ting
IP
netw
ork
s, it is n
ot capable
of sim
ula
ting
any o
ther
type o
f netw
ork
.
–Lack o
f a g
ood a
nd in-d
epth
docum
enta
tion. G
loM
oS
imsto
pped r
ele
asin
g u
pdate
s in
2000. In
ste
ad, it w
as
chosen a
s th
e c
ore
of th
e c
om
merc
ial Q
ualN
etsim
ula
tor,
and is n
ow
update
d a
s Q
ualN
et.
33
Zh
on
gh
on
g O
u
Glo
Mo
Sim
(co
nt.
)
34
Zh
on
gh
on
g O
u
Fig
. G
loM
oS
im a
rchitectu
re
OP
Net
•O
ptim
ize
d N
etw
ork
En
gin
ee
rin
g T
oo
ls (
OP
Ne
t) is a
dis
cre
te-e
ve
nt
ne
two
rk s
imu
lato
r firs
t p
rop
ose
d b
y M
IT in
19
86
, w
ritt
en
in
C+
+.
It
is a
we
ll-e
sta
blis
he
d a
nd
wid
ely
use
d c
om
me
rcia
l su
ite
fo
r n
etw
ork
sim
ula
tio
n.
•It
use
s a
hie
rarc
hic
al m
od
el to
de
fin
e e
ach
asp
ect
of
the
syste
m.
–T
he
to
p le
ve
l co
nsis
ts o
f th
e n
etw
ork
mo
de
l, w
he
re t
op
olo
gy is d
esig
ne
d;
–T
he
ne
xt
leve
l is
th
e n
od
e le
ve
l, w
he
re d
ata
flo
w m
od
els
are
de
fin
ed
;
–A
th
ird
le
ve
l is
th
e p
roce
ss e
dito
r, w
hic
h h
an
dle
s c
on
tro
l flo
w m
od
els
. –
A th
ird
le
ve
l is
th
e p
roce
ss e
dito
r, w
hic
h h
an
dle
s c
on
tro
l flo
w m
od
els
.
–F
ina
lly, a
pa
ram
ete
r e
dito
r is
in
clu
de
d to
su
pp
ort
th
e t
hre
e h
igh
er
leve
ls.
•T
he
hie
rarc
hic
al m
od
el re
su
lts in
an
eve
nt
qu
eu
e f
or
the
dis
cre
te
eve
nt
sim
ula
tio
n e
ng
ine
an
d a
se
t o
f e
ntitie
s r
ep
rese
ntin
g t
he
n
od
es t
ha
t w
ill b
e h
an
dlin
g t
he
eve
nts
. E
ach
en
tity
in
th
e s
yste
m
co
nsis
ts o
f a
fin
ite
sta
te m
ach
ine
wh
ich
pro
ce
sse
s t
he
eve
nts
d
uri
ng
sim
ula
tio
n.
35
Zh
on
gh
on
g O
u
OP
Net
(co
nt.
)
•P
ros:
–C
ap
ab
le o
f e
xe
cu
tin
g a
nd
mo
nito
rin
g s
eve
ral sce
na
rio
s in
a
co
ncu
rre
nt
ma
nn
er;
–S
up
po
rtin
g t
he
use
of
mo
de
ling
diffe
ren
t se
nso
r-sp
ecific
h
ard
wa
re,
su
ch
as p
hysic
al-
link t
ran
sce
ive
rs a
nd
an
ten
na
s;
–S
up
po
rtin
g c
usto
m p
acke
t fo
rma
ts;
–G
rap
hic
alin
terf
ace
to
de
ve
lop
mo
de
ls,
als
o c
an
be
use
d t
o
mo
de
l, g
rap
h,
an
d a
nim
ate
th
e r
esu
ltin
g o
utp
ut.
•C
on
s:
–S
uffe
rin
g f
rom
th
e s
am
e o
bje
ct-
ori
en
ted
sca
lab
ility
pro
ble
ms a
s
ns-2
.–
No
t su
pp
ort
ing
as m
an
y p
roto
co
ls a
s n
s-2
be
ca
use
of
its
co
mm
erc
ial fe
atu
re.
36
Zh
on
gh
on
g O
u
Qu
alN
et
•A
co
mm
erc
ial a
d h
oc n
etw
ork
sim
ula
tor
ba
se
d o
n t
he
G
loM
oS
imco
re.
•It
exte
nd
s t
he
Glo
Mo
Sim
offe
rin
g b
y b
rin
gin
g s
up
po
rt,
a
de
ce
nt d
ocu
me
nta
tio
n,
a c
om
ple
te s
et
of
use
r-fr
ien
dly
to
ols
fo
r b
uild
ing
sce
na
rio
s a
nd
an
aly
zin
g s
imu
latio
n
ou
tpu
t.
•Q
ua
lNe
ta
lso
exte
nd
s t
he
se
t o
f m
od
els
an
d p
roto
co
ls
su
pp
ort
ed
by t
he
in
itia
l G
loM
oS
imd
istr
ibu
tio
n.
•A
s it is
bu
ilt o
n t
op
of
Glo
Mo
Sim
, Q
ua
lNe
tis
wri
tte
n in
P
ars
ec.
37
Zh
on
gh
on
g O
u
Qu
alN
et
(in
terf
ace)
38
Zh
on
gh
on
g O
u
OM
NeT
++
•In
co
ntr
ast to
ns-2
an
d n
s-3
, O
MN
eT
++
is n
ot a
ne
two
rk s
imu
lato
r b
y d
efin
itio
n, b
ut a
g
en
era
l p
urp
ose
dis
cre
te e
ve
nt-
ba
se
d s
imu
latio
n fra
me
wo
rk, b
ase
d o
n o
bje
ct-
ori
en
ted
de
sig
n.
•It
is m
ostly a
pp
lied
to
th
e d
om
ain
of n
etw
ork
sim
ula
tio
n, g
ive
n t
he
fa
ct th
at w
ith
its
IN
ET
pa
cka
ge
it p
rovid
es a
co
mp
reh
en
siv
e c
olle
ctio
n o
f In
tern
et p
roto
co
l m
od
els
.–
The IN
ET
Fra
mew
ork
conta
ins m
odels
for
severa
l w
ired a
nd w
irele
ss n
etw
ork
ing p
roto
cols
, in
clu
din
g U
DP, T
CP, S
CT
P, IP
, IP
v6,
Eth
ern
et, P
PP, 802.1
1,
MP
LS
, O
SP
F,
and m
any o
thers
.
•In
ad
ditio
n, o
the
r m
od
el p
acka
ge
s s
uch
as th
e O
MN
eT
++
Mo
bili
ty F
ram
ew
ork
an
d
Ca
sta
lia fa
cili
tate
th
e s
imu
latio
n o
f m
ob
ile a
d h
oc n
etw
ork
s o
r w
ire
less s
en
so
r C
asta
lia fa
cili
tate
th
e s
imu
latio
n o
f m
ob
ile a
d h
oc n
etw
ork
s o
r w
ire
less s
en
so
r n
etw
ork
s.
•O
MN
eT
++
sim
ula
tio
ns c
on
sis
t o
f so
-ca
lled
sim
ple
mo
du
les (C
++
) w
hic
h r
ea
lize
th
e
ato
mic
be
ha
vio
r o
f a
mo
de
l, e
.g. a
pa
rtic
ula
r p
roto
co
l. M
ultip
le s
imp
le m
od
ule
s c
an
be
lin
ke
d to
ge
the
r a
nd
fo
rm a
co
mp
ou
nd
mo
du
le.
•O
MN
eT
++
utiliz
es N
ED
(N
Etw
ork
De
scri
ptio
n)
lan
gu
ag
e to
co
mb
ine
th
e s
imp
le
mo
du
les in
to c
om
po
un
d m
od
ule
s a
nd
de
fin
e th
e n
etw
ork
to
po
log
ies. N
ED
is
tra
nsp
are
ntly r
en
de
red
in
to C
++
co
de
wh
en
th
e s
imu
latio
n is c
om
pile
d a
s a
wh
ole
.
39
Zh
on
gh
on
g O
u
NA
B
•N
etw
ork
in
A B
ox (
NA
B)
is a
dis
cre
te e
ve
nt
sim
ula
tor
de
ve
lop
ed
at
EP
FL (
La
usa
nn
e,
Sw
itze
rla
nd
).
•N
AB
is d
ed
ica
ted
to
MA
NE
Ts s
imu
latio
n.
•N
AB
is fo
cu
sin
g o
n s
ca
lab
ility
an
d v
isu
aliz
atio
n a
nd
fe
atu
res a
ve
ry r
ea
listic m
ob
ility
mo
de
l (a
co
nstr
ain
ed
fe
atu
res a
ve
ry r
ea
listic m
ob
ility
mo
de
l (a
co
nstr
ain
ed
w
ayp
oin
t b
ase
d o
n c
ity m
ap
s).
•N
AB
’s d
esig
n is n
od
e-o
rie
nte
d (
an
d o
bje
ct-
ori
en
ted
); t
ha
t is
ea
ch
no
de
is r
ep
rese
nte
d b
y a
n o
bje
ct.
It
is w
ritt
en
in
O
Ca
ml. I
t is
op
en
so
urc
e.
40
Zh
on
gh
on
g O
u
J-S
im•
J-S
im, d
eve
lop
ed
at U
niv
ers
ity o
f Illin
ois
at U
rba
na
-Ch
am
pa
ign
, is
a g
en
era
l p
urp
ose
Ja
va
-ba
se
dsim
ula
tor
mo
de
led
aft
er
ns-2
.
•U
nlik
e n
s-2
, J-S
imu
se
s th
e c
on
ce
pt o
f co
mp
on
en
ts, re
pla
cin
g th
e n
otio
n th
at e
ach
n
od
e s
ho
uld
be
re
pre
se
nte
d a
s a
n o
bje
ct.
•J-S
imu
se
s th
ree
to
p le
ve
l co
mp
on
en
ts, e
ach
co
mp
on
en
t is
bro
ke
n in
to d
iffe
ren
t p
art
s
an
d m
od
ele
d d
iffe
ren
tly w
ith
in t
he
sim
ula
tor:
–
the targ
et
node (
whic
h p
roduces s
tim
uli)
;
–th
e s
ensor
node (
that
reacts
to t
he s
tim
uli)
;
–th
e s
ink n
ode (
the u
ltim
ate
destination f
or
stim
uli
report
ing).
–th
e s
ink n
ode (
the u
ltim
ate
destination f
or
stim
uli
report
ing).
•P
ros:
–C
om
ponent-
base
d a
rchitectu
re s
cale
s b
etter
than t
he o
bje
ct
oriente
d m
odel used b
y n
s-2
and o
ther
sim
ula
tors
.
–A
pplic
ations m
ay b
e s
imula
ted,
and t
here
is s
upport
for
the c
onnection o
f re
al hard
ware
sensors
to t
he s
imula
tor.
•C
on
s:
–R
ela
tively
com
plic
ate
d t
o u
se;
–F
aces its
share
of in
effic
iencie
s.
Java,
in g
enera
l, is a
rguably
less e
ffic
ient
than m
any o
ther
languages.
41
Zh
on
gh
on
g O
u
JiS
T•
Ja
va
in
Sim
ula
tio
n T
ime
(JiS
T)
allo
ws t
he
im
ple
me
nta
tio
n o
f n
etw
ork
sim
ula
tio
ns in
sta
nd
ard
Ja
va
. It
is m
ostly u
se
d in
co
nju
nctio
n w
ith
SW
AN
S, a
sim
ula
tor
for
mo
bile
a
d h
oc n
etw
ork
s b
uilt
on
to
p o
f JiS
T.
•N
etw
ork
sim
ula
tio
ns in
JiS
Ta
re m
ad
e u
p o
f e
ntitie
sw
hic
h r
ep
rese
nt th
e n
etw
ork
e
lem
en
ts, fo
r e
xa
mp
le n
od
es, w
ith
sim
ula
tio
n e
ve
nts
be
ing
fo
rme
d b
y m
eth
od
in
vo
ca
tio
ns a
mo
ng
th
ose
en
titie
s.
•T
he
en
titie
s a
dva
nce
th
e s
imu
latio
n tim
e in
de
pe
nd
en
tly b
y n
otify
ing
th
e s
imu
latio
n
co
re.
•W
hile
th
e c
od
e in
sid
e a
n e
ntity
is e
xe
cu
ted
lik
e a
ny a
rbitra
ry J
ava
pro
gra
m, o
nly
th
e
inte
ractio
ns
be
twe
en
th
e in
div
idu
al e
ntitie
s a
re c
arr
ied
ou
t in
sim
ula
tio
n tim
e.
•T
he
in
tera
ctio
ns b
etw
ee
n e
ntitie
s c
orr
esp
on
d to
syn
ch
ron
iza
tio
n p
oin
ts a
nd
fa
cili
tate
th
e p
ara
llel e
xe
cu
tio
n o
f co
de
at d
iffe
ren
t e
ntitie
s, re
su
ltin
g in
a p
ote
ntia
l p
erf
orm
an
ce
g
ain
.
•T
he
offic
ial d
eve
lop
me
nt o
f JiS
Th
as s
talle
d, a
s it is
no
lo
ng
er
ma
inta
ine
d b
y its
o
rig
ina
l a
uth
or,
Rim
on
Ba
rr.
42
Zh
on
gh
on
g O
u
Sim
Py
•S
imP
yis
a p
roce
ss-o
rie
nte
dd
iscre
te-e
ve
nt
sim
ula
tor;
•U
nlik
e t
he
oth
er
sim
ula
tors
, n
o p
ub
lic a
va
ilab
le n
etw
ork
mo
de
ls
exis
t fo
r S
imP
y. I
nste
ad
, it is a
ba
re s
imu
latio
n A
PI
wri
tte
n in
P
yth
on
.
•In
Sim
Py,
th
e b
asic
sim
ula
tio
n e
ntitie
s a
re p
roce
sse
s.
Th
ey a
re
exe
cu
ted
in
pa
ralle
l a
nd
ma
y e
xch
an
ge
Pyth
on
ob
jects
am
on
g
ea
ch
oth
er.
e
ach
oth
er.
•M
ost p
roce
sse
s in
clu
de
an
in
fin
ite
lo
op
in
wh
ich
th
e m
ain
actio
ns o
f th
e p
roce
ss a
re p
erf
orm
ed
.
•B
esid
es a
bstr
actio
ns f
or
pro
ce
sse
s a
nd
th
e r
ela
ted
exch
an
ge
of
ob
jects
, S
imP
yp
rovid
es in
str
uctio
ns f
or
the
syn
ch
ron
iza
tio
no
f sim
ula
tio
n p
roce
sse
s a
nd
co
mm
an
ds f
or
the
mo
nito
rin
g o
f sim
ula
tio
n d
ata
.
43
Zh
on
gh
on
g O
u
NetH
aw
kE
AS
T•
En
vir
on
me
nt fo
r A
uto
ma
ted
Syste
ms T
estin
g (
EA
ST
) is
a te
st
au
tom
atio
n a
nd
tra
ffic
ge
ne
ratio
n to
ol th
at a
llow
s u
se
rs to
e
mu
late
/sim
ula
te o
ne
or
mo
re n
etw
ork
ele
me
nts
in
th
e
tele
co
mm
un
ica
tio
ns n
etw
ork
.
•E
AS
T c
an
be
use
d fo
r fe
atu
rete
stin
g a
s w
ell
as lo
ad
testin
g.
•E
AS
T p
rovid
es a
n e
asy to
use
, in
tuitiv
e G
UIfr
on
t-e
nd
th
at is
•
EA
ST
pro
vid
es a
n e
asy to
use
, in
tuitiv
e G
UIfr
on
t-e
nd
th
at is
co
nsis
ten
t a
cro
ss a
ll p
roto
co
ls.
•T
he
EA
ST
GU
I is
JA
VA
ba
se
d a
nd
he
nce
pla
tfo
rm in
de
pe
nd
en
t.
•N
etH
aw
kw
as o
rig
ina
lly b
ase
d in
Ou
lu, a
nd
wa
s a
cq
uir
ed
by
EX
FO
(C
an
ad
ian
co
mp
an
y)
on
Ma
rch
12
, 2
01
0.
44
Zh
on
gh
on
g O
u
NetH
aw
kE
AS
T (
co
nt.
)
45
Zh
on
gh
on
g O
u
Matl
ab
•M
AT
LA
B (
ma
trix
la
bo
rato
ry)
is a
nu
me
rica
l co
mp
utin
g
en
vir
on
me
nt a
nd
fo
urt
h-g
en
era
tio
n p
rog
ram
min
g la
ng
ua
ge
.
•It w
as o
rig
ina
lly d
esig
ne
d to
so
lve
lin
ea
r a
lge
bra
typ
e p
rob
lem
s
usin
g m
atr
ice
s.
•D
eve
lop
ed
by M
ath
Wo
rks, M
AT
LA
B a
llow
s:
–m
atr
ix m
an
ipu
latio
ns;
–m
atr
ix m
an
ipu
latio
ns;
–p
lottin
g o
f fu
nctio
ns a
nd
da
ta;
–im
ple
me
nta
tio
n o
f a
lgo
rith
ms;
–cre
atio
n o
f u
se
r in
terf
ace
s,
an
d in
terf
acin
g w
ith
pro
gra
ms w
ritte
n in
o
the
r la
ng
ua
ge
s,
inclu
din
g C
/C+
+, Ja
va
, S
QL
, a
nd
Fo
rtra
n e
tc.
•C
om
me
rcia
l pro
du
ct, b
ut w
ide
ly u
se
d in
in
du
str
y a
nd
aca
de
mia
.–
Ma
ny a
lgo
rith
ms a
nd
to
olb
oxe
s f
ree
ly a
va
ilab
le
46
Zh
on
gh
on
g O
u
Gra
nu
lari
ty a
nd
mo
bilit
y
47
Zh
on
gh
on
g O
u
Para
llelism
an
d in
terf
ace
48
Zh
on
gh
on
g O
u
Po
pu
lari
ty a
nd
lic
en
se
49
Zh
on
gh
on
g O
u
Po
pu
lari
ty
50
Zh
on
gh
on
g O
u
Sim
ula
tion r
esult o
f 2000-2
005 p
roceedin
gs o
f th
e M
obiH
oc
confe
rence
Pit
falls o
f sim
ula
tio
n•
Sim
ula
tio
n s
etu
p–
Sim
ula
tio
n typ
e:
•D
yn
am
ic v
s.
ste
ad
y-s
tate
(e
xe
cu
te o
ne
typ
e o
f sim
ula
tio
n a
nd
re
po
rt r
esu
lts
on
th
e o
the
r ty
pe
of sim
ula
tio
n).
–M
od
el va
lida
tio
n &
ve
rifica
tio
n:
•E
xe
cu
te s
imu
latio
ns w
ith
a m
od
el th
at h
as n
ot b
ee
n v
alid
ate
d in
th
e s
pe
cific
e
nvir
on
me
nt.
–P
RN
G (
Pse
ud
o R
an
do
m N
um
be
r G
en
era
tor)
va
lida
tio
n &
ve
rifica
tio
n:
–P
RN
G (
Pse
ud
o R
an
do
m N
um
be
r G
en
era
tor)
va
lida
tio
n &
ve
rifica
tio
n:
•S
om
eo
ne
[9
] e
stim
ate
s th
at th
e N
S-2
PR
NG
is o
nly
va
lid f
or
se
ve
ral
tho
usa
nd
nu
mb
ers
.
–V
ari
ab
le d
efin
itio
n:
•T
he
re a
re 6
74
va
ria
ble
s d
efin
ed
in
th
e n
s-d
efa
ult.t
cl f
ile o
f N
S-2
.27
.
–S
ce
na
rio
de
ve
lop
me
nt:
•L
ack o
f in
de
pe
nd
en
ce
be
twe
en
pa
ram
ete
rs;
•L
ack o
f ri
go
rou
s s
ce
na
rio
s, n
o b
en
ch
ma
rk s
ce
na
rio
s.
51
Zh
on
gh
on
g O
u
Sim
ula
tio
n e
xecu
tio
n
•S
ett
ing the P
RN
G s
eed:
–N
ot se
ttin
g p
rop
erl
y, N
S-2
use
s a
de
fau
lt s
ee
d o
f 1
23
45
fo
r e
ach
sim
ula
tio
n r
un
;
•S
cenario initia
lization:
–M
ost sim
ula
tio
ns s
tart
with
em
pty
ca
ch
es, q
ue
ue
s, a
nd
ta
ble
s, d
ete
rmin
ing
an
d r
ea
ch
ing
th
e s
tea
dy-s
tate
leve
l o
f ta
ble
s, d
ete
rmin
ing
an
d r
ea
ch
ing
th
e s
tea
dy-s
tate
leve
l o
f a
ctivity is p
art
of th
e in
itia
liza
tio
n.
•M
etr
ic c
olle
ction:
–O
utp
ut n
ee
ds to
be
in
lin
e w
ith
th
e r
eq
uir
ed
gra
nu
lari
ty.
52
Zh
on
gh
on
g O
u
Ou
tpu
t an
aly
sis
•S
ing
le s
et o
f d
ata
:–
Ta
kin
gth
e f
irst
se
t o
f re
su
lts fro
m a
sim
ula
tio
n a
nd
acce
ptin
g t
he
re
su
lts a
s “
tru
th”.
•S
tatistica
l a
na
lysis
:–
No
t u
sin
g t
he
co
rre
ct
sta
tistica
l fo
rmu
las w
ith
th
e d
iffe
ren
t fo
rms o
f o
utp
ut.
•C
on
fid
en
ce
inte
rva
ls:
•C
on
fid
en
ce
inte
rva
ls:
–A
cu
lmin
atio
n o
f se
ve
ral o
f th
e p
revio
us a
na
lysis
issu
es.
–C
on
fid
en
ce
in
terv
als
acco
un
t fo
r th
e r
an
do
mn
ess a
nd
va
rie
d o
utp
ut
fro
m a
sto
ch
astic s
imu
latio
n.
•P
ub
lish
ing
:–
Th
e la
ck o
f co
nsis
ten
cy in
pu
blis
hin
g s
imu
latio
n b
ase
d s
tud
y r
esu
lts
dir
ectly im
pa
cts
th
e t
rustw
ort
hin
ess o
f th
ese
stu
die
s.
–A
ne
w r
ese
arc
he
r ca
nn
ot
rep
ea
t th
e s
tud
ies t
o s
tart
his
or
he
r o
wn
fo
llow
-on
re
se
arc
h.
53
Zh
on
gh
on
g O
u
Co
nclu
sio
n
•E
ach o
f th
e p
itfa
lls d
iscussed t
akes a
way f
rom
the
goals
of m
akin
g t
he r
esearc
h:
–re
pe
ata
ble
,
–u
nb
iase
d,
–ri
go
rou
s,
–sta
tistica
lly s
ou
nd
.
•It
is s
till
a long w
ay t
o g
o!!
!
54
Zh
on
gh
on
g O
u
Refe
ren
ces
•1
. P
asiL
assila
. S
-38
.31
48
Sim
ula
tio
n o
f D
ata
Ne
two
rks.
•2
. W
ikip
edia
. h
ttp
://e
n.w
ikip
ed
ia.o
rg/w
iki/S
imu
latio
n.
•3
. W
ikip
edia
. h
ttp
://e
n.w
ikip
ed
ia.o
rg/w
iki/C
om
pu
ter_
sim
ula
tion
.
•4
. W
ein
ga
rtn
er
E, vo
m L
eh
n H
& W
eh
rle
K (
20
09
) A
Pe
rfo
rma
nce
Co
mp
ari
so
n o
f R
ece
nt
Ne
two
rk S
imu
lato
rs. IE
EE
In
tern
atio
na
l Co
nfe
ren
ce
on
Co
mm
un
ica
tio
ns (
ICC
'09
):1
-5.
•5
. D
alim
irO
rfa
nu
s, Jo
ha
nn
es L
essm
an
n, P
ete
r Ja
na
cik
, L
aza
r L
ach
ev
(20
08
) P
erf
orm
an
ce
of w
ire
less n
etw
ork
sim
ula
tors
: a
ca
se
stu
dy.
Pro
ce
ed
ing
s o
f th
e 3
nd
AC
M
wo
rksh
op
on
Pe
rfo
rma
nce
mo
nito
rin
g a
nd
me
asu
rem
en
t o
f h
ete
rog
en
eo
us w
ire
less a
nd
w
ire
d n
etw
ork
s (
PM
2H
W2
N '0
8):
59
-66
.w
ire
d n
etw
ork
s (
PM
2H
W2
N '0
8):
59
-66
.
•6
. D
. C
urr
en
. A
su
rve
y o
f sim
ula
tio
n in
se
nso
r n
etw
ork
s. S
tud
en
t p
roje
ct,
w
ww
.cs.b
ing
ha
mto
n.e
du
/~ka
ng
/tea
chin
g/c
s58
0s/d
avid
.pd
f,20
07
.
•7
. S
tua
rt K
urk
ow
ski, T
racy C
am
p &
Mic
ha
el C
ola
gro
sso
(20
05
) M
AN
ET
sim
ula
tio
n s
tud
ies:
the
in
cre
dib
les. A
CM
SIG
MO
BIL
E M
ob
ile C
om
pu
tin
g a
nd
Co
mm
un
ica
tio
ns R
evie
w 9
(4):
5
0-6
1.
•8
. W
ikip
edia
. h
ttp
://e
n.w
ikip
ed
ia.o
rg/w
iki/M
AT
LA
B.
•9
. K
. P
aw
liko
wski, J
. Je
on
g, a
nd
R. L
ee
. L
ett
ers
to
th
e e
dito
r. IE
EE
Co
mm
un
ica
tio
ns
Ma
ga
zin
e, p
ag
es 1
32
–1
39
, 2
00
2.
55
Zh
on
gh
on
g O
u
Co
nta
ct
Info
rmati
on
•C
ours
e w
eb p
age:
•h
ttp
s://n
op
pa.tkk.fi/n
op
pa/k
urs
si/t-
11
0.6
13
0/
•C
onta
ct em
ail:
•zh
on
gh
on
g.o
u@
tkk.fi
•O
ffic
e h
our:
•O
ffic
e h
our:
•F
ri 1
1-1
2 r
oo
m A
10
9
•Q
ustions &
Suggestions?
56
Zh
on
gh
on
g O
u