INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
Impac o Admission Con ol Me hods o he
T a ic Managemen
Filip CHAMRAZ, I an BARONAK
Ins i u e o Telecommunica ions, Facul y o Elec ical Enginee ing and In o ma ion Technology,
Slo ak Uni e si y o Technology, Ilko ico a 3, 812 19 B a isla a, Slo akia
ilip.c[email p o ec ed], i [email p o ec ed]
DOI: 10.15598/aeee. 13i4.1437
Abs ac . The pape deals wi h Admission con ol
me hods (AC) in IMS ne wo ks (IP mul imedia sub-
sys em) as one o he elemen s ha help ensu e QoS
(Quali y o se ice). In he pape we a e ying o
choose he bes AC me hod o selec ed IMS ne wo k
node o allow access o he g ea es numbe o use s.
O he la ge numbe o me hods ha we e es ed and
conside ed good we chose wo. The pape compa es
Gaussian app oxima ion me hod and one o he mea-
su emen based me hod, speci ically „Measu ed Sum“.
Bo h me hods es ima e e ec i e bandwid h o allow ac-
cess o he g ea es numbe o use s/de ices and allow
hem access o p epaid se ices o mul imedia con en .
Keywo ds
AC me hods, IMS, IP, a ic con ol, QoS.
1. The P esen Gene a ion o
Ne wo ks
IP mul imedia subsys em (IMS) was c ea ed in 1999
as s anda d 3GPP. This s anda d was he i s y o
he c ea ion o con e gen ne wo k and c ea ion o a
single pla o m o p o ide mul imedia se ices. I was
o iginally designed o ensu e IP connec i i y in UMTS.
I b ough change om ci cui -swi ched echnology,
which was used in olde gene a ions o he sys ems, o
packe -swi ched echnology. IMS gua an ees QoS and
b ings a numbe o bene i s, echnical and economical,
o he se ice p o ide and cus ome oo. The bigges
ad an age is coope a ion wi h a p e ious gene a ion o
ne wo ks by buil -in ga es and s ong s anda diza ion.
I is used o all ypes o se ices, adio, ixed and cable.
IMS es ing ope a ion s a ed in 2006 in Japan and
Ko ea and 2007 in he Uni ed S a es. Today IMS is al-
eady ully de eloped in Slo akia ( o example elecom-
munica ions ope a o s O2, Telekom, O ange). IMS is
used o p o ide a wide ange o se ices, o example.
VoIP, IPTV, ideo communica ion, ans e o da a se -
ices and o he s [1].
2. The T a ic Managemen
Admission con ol me hods a e used by c ea ing a new
connec ion o decide ha a new connec ion will be ac-
cep ed o ejec ed. AC me hods a e based on p ob-
abili y heo y and ma hema ical s a is ics. Ha e he
ask keep he balance be ween he use o ne wo k e-
sou ces and p e iously ag eed on connec ion pa ame-
e s. I is he i s ac o be ca ied ou in he alloca ion
o ne wo k esou ces o a pa icula connec ion. AC
me hods a e he i s p o ec ion agains edundancy in
he ne wo k. New connec ion is allowed only i he e
is gua an eed QoS, o he wise he connec ion is e used.
QoS mus also be obse ed o he exis ing connec ions
in he ne wo k. I i is no me he new connec ion will
be allowed [2], [3], [4], [5], [6], [7].
AC me hod sol es he p oblem when he Nconnec-
ions in mul iplex wi h a o al capaci y C, he p oba-
bili y ha he sum o he immedia e bi a e i( )o
all connec ions in mul iplex exceeds he o al capaci y
C, is less han a gi en alue ε. This p obabili y can be
exp essed as
P"n
X
i=1
i( )≥C#< ε. (1)
AC me hods should sa is y h ee main condi ions:
•E ec i ely alloca e bandwid h o u ilize maxi-
mally o elecommunica ions ne wo k.
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 280
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
•Manage a elecommunica ions ne wo k o mee all
equi emen s o QoS.
•Does no alloca e he en i e bandwid h so ha no
o e load on he ne wo k node is [3].
Ne wo k node
IMS
ne wo k
Fig. 1: AC me hods.
2.1. Classi ica ion o T a ic
Managemen Me hods
AC me hods can be classi ied o se e al pa ame e s.
I depends on he iew, o pa ame e s wha hey wo k
wi h and unde which equi emen s a e e alua ed. The
i s way o di ide hese me hods is o di ide hem
on he basis o a ic pa ame e s, ob ained om p e-
de ined alues (Pa ame e Based Admission Con ol
Me hods - PBAC) o used online measu emen o ne -
wo k (Measu emen Based Admission Con ol Me h-
ods - MBAC). We can di ide hem h ough he use o
a bu e o pa ame e PLR (Packe Loss Ra io) o e -
ec i e bandwid h and mo e. We know se e al ens o
AC me hods ha a e in ended o be used in pa icula
ne wo ks o in some nodes o elecommunica ion ne -
wo ks. Wi h many o hem we wo ked well in ou In-
s i u e o elecommunica ions FEI STU in B a isla a.
The mos equen ly used me hods a e Gaussian ap-
p oxima ion me hod, me hod o e ec i e bandwid h,
di usion me hod, con olu ion me hod and o he s. This
pape compa es wo me hods, because i could no be
possible o compa e all known me hods and e en hose
mos equen ly implemen ed. Fo simula ion he wo
me hods we e chosen, Gaussian app oxima ion me hod
and “Measu ed sum“ algo i hm. These me hods we e
chosen because a e ying many o he s, hese came
ou be e han he o he s. And o cou se hese wo
me hods a e used in simila a ic models wha is he
eason why we we e in e es ing abou hose me hods
igh om e y beginning [3], [4], [7].
The u u e o AC me hods is expec ed in he use
o he me hods ha used he online measu emen o
ne wo k (MBAC), uzzy logic and neu al ne wo ks [10],
[12].
3. Gaussian App oxima ion
Me hod
In he e ms o dis ibu ion i is PBAC me hod.
The me hod o de e mining he equi ed bandwid h,
packe loss a e and memo y o e low is based on he
cen al limi heo em and Gaussian dis ibu ion. This
algo i hm app oxima es p obabili y dis ibu ion o ag-
g ega ed a ic p o ided ha he numbe o connec-
ions is N, i is close o in ini y and none o he connec-
ions is dominan . Wi h an inc easing o agg ega ed
a ic con e ges o Gaussian model. Each o connec-
ions is de e mined by he mean bi a e λiand i s
s anda d de ia ion σi.
λi=SP Ri,(2)
σ2
i=SP Ri(PPRi−SPRi).(3)
The esul ing mean alue o bi a e is
λ=
N
X
i=1
λi.(4)
And he esul ing alue o a iance is
σ2=
N
X
i=1
σi.(5)
I Xis he agg ega ed bi a e o Nconnec ions we
need o ind capaci y cg o which i holds
P{X > cg} ≤ ε, (6)
εis a p obabili y o packe loss
cg≈λ+α,σ,,(7)
α,=p−2ln(ε)−2ln(π).(8)
The new connec ion is accep ed only i cg< C, o h-
e wise is ejec ed. Es ima ion o he p obabili y link
o e low is
Po e low =P" N
X
i=1
i( )≥C!#=
=1
√2πe
−
(λ−C)2
2σ2.(9)
The p obabili y o packe loss is
Ploss =
PN
P
i=1
i( )−C+
λ=
=σ
λ√2πe
−
(λ−C)2
2σ2.(10)
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 281
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
i( )is a ac ual bi a e o connec ion i[bps]. The
es ima ion o equi ed bandwid h is calcula ed as
Cg=
N
X
i=1
λi+h
N
X
i=1
σi,(11)
whe e
h=p−2ln(ε)−2ln(π),(12)
hen he esul ing alue o equi ed bandwid h is
Cg=
N
X
i=1
λi+p−2ln(ε)−2ln(π)N
X
i=1
σi.(13)
This app oach has been al eady used in li e a u e
and he app oxima ion gi es e y good esul s only o
a la ge numbe o connec ions ha ha e simila s a-
iona y dis ibu ion, long pe iods o bu s and Gaus-
sian cha ac e . Fo a small numbe o connec ions and
sho pe iods o bu s i is no able o use hose ela-
ionships. The esul ing alue o equi ed bandwid h
would be oo conse a i e [7], [11], [15] and [16].
4. MBAC Me hods
The me hods a e based on on-line measu emen s o
a ic passing h ough he swi ch and he new con-
nec ion equi es only a minimum o in o ma ion. Bu
he addi ional in o ma ion imp o es he e iciency o
AC me hod. Ini ial es ima e o bandwid h is made o
he a ailable pa ame e s and u he adjus ed acco d-
ing o he measu emen esul s. On-line measu emen
mus be as enough. I applies ha he sho e mea-
su ing pe iod hen mo e connec ions can be se ed. AC
me hod based on he measu emen canno be used di-
ec ly by he cu en packe loss a es. The e o e use
a simple and mo e e icien way and measu emen o
bandwid h [8], [9], [13].
I Nconnec ions passing h ough he swi ch use he
bandwid h C, we y o es ima e he minimum band-
wid h C(N).C(N) is bandwid h ha hese connec ions
need o be able o gua an ee p ede e mined pa ame e s
o packe loss a e.
4.1. Algo i hm „Measu ed Sum“
I is one o MBAC me hods and i is used he a o e-
said p inciple. I is an imp o emen o „Simple sum“
algo i hm which was published in [7], [10] and [13]. Al-
go i hm allows connec ions un il
C + n+1 < µC, (14)
whe e C- he maximum capaci y o he line, C - is
he sum o nbi a es and connec ion n+1, n+1 - is
Fig. 2: Measu emen based AC me hods.
he bi a e connec ion, eques ing o pe mission and
µ- is use -de ined a ic usabili y ( alue is om ze o
o one).
Mos o en i is implemen ed in swi ches and ou e s
whe e we donno expec oo much load.
5. Simula ion
Simula ions and all he necessa y calcula ions o he
indi idual compa ed o all me hods we e de eloped in
Ma lab (R2010). All esul s o he indi idual simula-
ions a e shown h ough speci ic g aphs o hei be e
eadabili y and ollow much easie in e p e a ion. Be-
cause o all he necessa y calcula ions and hei esul s
should clea ly no choose a mo e app op ia e me hod
and he esul s should lose i s impo ance.
5.1. T a ic Model and Pa ame e s 1
Fo he simula ion (and all he necessa y calcula ions
had o be pe o med a each o he compa ed me hods)
we e de ined a ic pa ame e s. I was necessa y cal-
cula e wi h hese pa ame e s. And in e alua ing he
esul s, aking o conside some limi s, so ha we can
clea ly de e mine he app op ia eness o he me hod
[8], whe e we de ine: C- maximum capaci y o he line,
Po e low - maximum alue o p obabili y o line o e -
low, Ploss - maximum alue o p obabili y o packe
loss.
Pa ame e s we e de ined as ollows: C= 10 Mbps,
Po e low is om 10−7 o 10−5,Ploss is om 10−6 o
10−5.
In Fig. 3 you can see simula ed node o IMS ne wo k.
As a sou ce o a ic was used andomly gene a ed a -
ic ma ix on he size o n×T, we e n= 3000 ep e-
sen ed he numbe o used esou ces and T= 3000 ep-
esen ed he numbe o ime cycles, when he a ic was
simula ed. T a ic ma ix ep esen ed equi emen s
o ne wo k (o use s) o connec ion (use ’s access o
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 282
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
R1
AC me hod
10 Mbi /s
IMS
Fig. 3: Model o a ic o simula ions.
his subsc ip ion se ices). Indi idual ne wo k equi e-
men s ep esen ed speci ic mul imedia de ices (sma -
phone, elephone, elephone - VoIP). VoIP elephones
ep esen use s who use codecs G.711 and G.729E o
making a oice call (VoIP). I means ha he equi e-
men s o bandwid h anged om 12 kbps o codec
G.729E o 64 kbps o codec G.711. In Fig. 4 you can
see incoming equi emen s o he node R1.
0 50 100 150 200
−20
−10
0
10
20
30
40
ime [ms]
bandwid h equi emen s [Kbps]
Fig. 4: Incoming equi emen s o bandwid h o a ic
model 1.
5.2. The Simula ion Resul s o
Gaussian App oxima ion
Me hod T a ic Model 1
The i s selec ed and simula ed was Gaussian app ox-
ima ion me hod. In Fig. 5 you can see a g aphical
simula ion esul o he esul an equi ed bandwid h
Cg. Fo access ask 9000 equi emen s. As you can see,
access is g an ed o 8031 use s/acceding de ices. I
means ha 89.23 % o all incoming equi emen s a e
enabled. O he a emp s o c ea e a new connec ion by
any o he de ice would al eady c oss maximum capac-
i y o line C= 10 Mbps. So ano he use will al eady
ha e access denied because he e is no a ailable band-
wid h.
In Fig. 6 you can see how changes he p obabili y
o line o e low. Fo 8031 use s, eaching he max.
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 5: Resul an equi ed bandwid h Cg.
capaci y o he line is he p obabili y o line o e low
9.473 ·10−7, which does no exceed he maximum al-
lowed p obabili y o line o e low Po e low (allowed
om 10−7 o 10−5).
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
0
0.5
1
1.5
2
2.5
3
3.5 x 10−6
numbe o use s
Po e low
Fig. 6: P obabili y o line o e low Po e low.
I is ob ious ha i he e a e mo e han 8031 use s in
ne wo k his p obabili y s ill does no exceed he max-
imum allowed alue. P oblem is he e is no a ailable
bandwid h o sa is y hose use s.
In Fig. 7 you can see how changes he p obabili y
o packe loss. Fo 8031 use s, eaching he maximum
capaci y o he line is he alue o p obabili y o packe
loss much g ea e , han selec ed maximum alue. Bu
o 6355 incoming equi emen s p obabili y o packe
loss is 4.6·10−6wha does no exceed he maximum
allowed alue. I means we a e no able o enable ac-
cess o 8031 use s which exceed maximum capaci y o
he line.Access is enable only o 6355 incoming equi e-
men s. I ep esen s ha 70.61 % o all incoming e-
qui emen s a e enabled. Fo 6355 equi emen s he u i-
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 283
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
0
1
2
3
4
5
6x 10−4
numbe o use s
Ploss
Fig. 7: P obabili y o packe loss Ploss.
liza ion o bandwid h (line) is 8.943 Mbps. You can see
he u iliza ion o line in Fig. 8. So we a e no able o
use whole bandwid h.
0 50 100 150 200
8.6
8.8
9
9.2
9.4
9.6
9.8
10
ime [ms]
U iliza ion o bandwid h [Mbps]
Fig. 8: U iliza ion o line.
5.3. The Simula ion Resul s o
„Measu ed“ Algo i hm T a ic
Model 1
In Fig. 9 you can see he simula ion esul o algo-
i hm „Measu ed sum“, which is measu emen based
AC me hod. You can see ha access is g an ed
o 751 incoming equi emen s. Once he e a e
751 use s/acceding de ices in ne wo k he e is no a ail-
able bandwid h o any o he use s. This small num-
be o use s ha access is g an ed may be caused by
he simplici y o he algo i hm ha uses his me hod.
Algo i hm assumed a ic usabili y µ= 0.7, which
means 70 % u iliza ion. The e o e he e ec o he
ad an ages o online measu emen s wi h his me hod
no show ully.
0 200 400 600 800 1000
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 9: The simula ion esul s o „Measu ed sum“ algo i hm.
5.4. T a ic Model and Pa ame e s 2
In he a ic model 2 he e a e he same pa ame e s
o ne wo k node as in model 1 ha you can see in
Fig. 3. The e a e same alues o C,Po e low and
Ploss. The di e ence be ween model 1 and 2 is in he
sou ce o a ic. As a sou ce o a ic was used an-
domly gene a ed a ic ma ix on he size o n×T, we e
n= 3000 ep esen ed he numbe o used esou ces and
T= 3000 ep esen ed he numbe o ime cycles, when
he a ic was simula ed. T a ic ma ix ep esen ed e-
qui emen s o ne wo k (o use s) o connec ion (use ’s
access o his subsc ip ion se ices). Indi idual ne -
wo k equi emen s ep esen ed speci ic mul imedia de-
ices (sma phone, elephone, elephone - VoIP). VoIP
elephones ep esen use s who use codecs G.711 and
G.729E o making oice call (VoIP). I means ha he
equi emen s o bandwid h anged om 12 kbps o
codec G.729E o 64 kbps o codec G.711. Sma phones
ep esen s use s who use da a downloads. Maximum
bandwid h o sma phones is 512 kbps. In Fig. 10
you can see incoming equi emen s o he node R1 o
a ic model 2.
5.5. The Simula ion Resul s o
Gaussian App oxima ion
Me hod T a ic Model 2
In Fig. 11 you can see a g aphical simula ion esul o
he esul an equi ed bandwid h Cg o Gaussian ap-
p oxima ion me hod. Fo access ask 100 equi emen s.
As you can see, access is g an ed o 88 use s/acceding
de ices. I means ha 88 % o all incoming equi e-
men s is enabled. O he a emp s o c ea e a new
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 284

INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
0 20 40 60 80 100
−100
0
100
200
300
400
500
ime [ms]
bandwid h equi emen s [Kbps]
Fig. 10: Incoming equi emen s o bandwid h o a ic
model 2.
connec ion by any o he de ice would al eady c oss
maximum capaci y o line C= 10 Mbps. So ano he
use will al eady ha e access denied because he e is no
a ailable bandwid h o sa is y his equi emen s.
0 20 40 60 80 100
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 11: Resul an equi ed bandwid h Cg.
In Fig. 12 you can see how changes he p obabili y
o line o e low. Fo 88 use s, eaching he maximum
capaci y o he line is he p obabili y o line o e low
1.031 ·10−6. This does no exceed he maximum al-
lowed p obabili y o line o e low Po e low (allowed
om 10−7 o 10−5) and do no come close o hese
alues. I is ob ious (in his model oo) ha i he e
a e mo e han 88 use s in ne wo k his p obabili y s ill
does no exceed he maximum allowed alue. P oblem
is he e is no a ailable bandwid h o sa is y hose use s.
In Fig. 13 you can see how changes he p obabili y
o packe loss. Fo 88 use s, eaching he maximum
capaci y o he line is he alue o p obabili y o packe
loss 9.848 ·10−6which does no exceed he maximum
0 20 40 60 80 100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5x 10−6
numbe o use s
Po e low
Fig. 12: P obabili y o line o e low Po e low.
allowed p obabili y o packe loss Ploss ( om 10−6 o
10−5) and do no come close o hese alues. So in
his model we a e able o g an ed access o all 88 use s
because he maximum allowed alue is no exceeded.
0 20 40 60 80 100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5 x 10−5
numbe o use s
Ploss
Fig. 13: P obabili y o packe loss Ploss.
In he compa ison wi h esul s o a ic model 1 you
can see ha we a e using whole a ailable bandwid h
and he e is no bandwid h o ano he ype o se ices.
5.6. The Simula ion Resul s o
„Measu ed“ Algo i hm T a ic
Model 2
In Fig. 14 you can see he simula ion esul o algo-
i hm „Measu ed sum“. You can see ha he maximum
line capaci y is exceeded o 43 acceding use s. This
small numbe o use s ha access is g an ed may be
caused by he simplici y o he algo i hm ha uses his
me hod. Algo i hm assumed a ic usabili y µ= 0.7,
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 285
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
which means 70 % u iliza ion. The e o e he e ec
o he ad an ages o online measu emen s wi h his
me hod no show ully.
0 20 40 60 80 100
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 14: The simula ion esul s o „Measu ed sum“ algo i hm.
6. Compa ison o Gaussian
App oxima ion Me hod and
„Measu ed Sum“ Algo i hm
6.1. Compa ison o T a ic Model 1
In Fig. 15 you can see simula ion esul s o a -
ic model 1 o „Measu ed sum“ algo i hm and Gaus-
sian app oxima ion me hod and Di usion me hod pub-
lished in [13] and [14]. F om his di ec compa ison o
Gaussian app oxima ion me hod and „Measu ed sum“
algo i hm i is ob ious ha Gaussian app oxima ion
me hod is much be e o he ou a ic node R1.
As was men ioned „Measu ed sum“ algo i hm (yellow
line) allows access o 751 use s, Gaussian app oxima-
ion me hod (g een line) allows access o 6355 use s
and di usion me hod (blue line) o 2360 use s.
F om simula ion esul s and he g aphs p esen ed
o a ic model 1 i is ob ious ha he p e e able AC
me hod in ou a ic node is Gaussian app oxima ion
me hod.
O cou se „Measu ed sum“ algo i hm is easie o im-
plemen in he node because he e a e less compu ing
equi emen s, cos s (i is easie o implemen = eco-
nomical). „Measu ed sum“ algo i hm is capable o do
g ea wo k in he nodes whe e we do no expec oo
much load. In ou node we assumed 70 % u iliza ion.
Gaussian app oxima ion me hod wo ks g ea only i
he e is high alue o equi emen s in he node. In
ou case whe e he e a e 9000 equi emen s i wo ks
g ea . We we e able o enable access o 8031 use s.
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 15: Compa ison o simula ed me hods o a ic model 1.
Bu we mus be e y ca e ul and wa ch he p obabili y
o packe loss. Because by 8031 enabled accesses he e
is oo high alue o p obabili y o packe loss. Accep -
able alue o p obabili y o packe loss is 4.6·10−6and
i means “only” 6355 enabled accesses. This numbe
ep esen s 70.61 % o all incoming equi emen s. I
also ep esen s 89.43 % u iliza ion o line (use o band-
wid h). The es o bandwid h is ee o use o ano he
ype o se ice. The e is 1.057 Mbps a ailable band-
wid h.
When we compa e he esul s o simula ion Gaussian
app oxima ion me hod allows access o 8.46 imes mo e
use s ha „Measu ed sum“ algo i hm.
6.2. Compa ison o T a ic Model 2
In Fig. 15 we can see simula ion esul s o a ic
model 2 o „Measu ed sum“ algo i hm and Gaussian
app oxima ion me hod F om his di ec compa ison o
Gaussian app oxima ion me hod and „Measu ed sum“
algo i hm o a ic model 2 i is ob ious ha Gaus-
sian app oxima ion me hod is s ill much be e o ou
a ic node R1. As was men ioned „Measu ed sum“
algo i hm (blue line) allows access o 43 use s, Gaus-
sian app oxima ion me hod (g een line) allows access
o 88 use s. F om simula ion esul s and he g aphs
p esen ed o a ic model 2 i is ob ious ha he
p e e able AC me hod in ou a ic node is s ill Gaus-
sian app oxima ion me hod.
O cou se „Measu ed sum“ algo i hm is easie o im-
plemen in he node because he e a e less compu ing
equi emen s, cos s (i is easie o implemen = eco-
nomical). „Measu ed sum“ algo i hm is capable o do
g ea wo k in nodes whe e we do no expec oo much
load. In ou node we assumed 70 % u iliza ion.
Gaussian app oxima ion me hod wo ks g ea only i
he e is high alue o equi emen s in node. In ou case
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 286
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
0 10 20 30 40 50 60 70 80 90
0
1
2
3
4
5
6
7
8
9
10
11
numbe o use s
channel capaci y [Mbps]
Fig. 16: Compa ison o simula ed me hods o a ic model 2.
when he e a e 100 equi emen s i s ill wo ks g ea .
We we e able o enable access o 88 use s. In compa i-
son wi h esul s o a ic model 1 you can see ha he
alues o p obabili y o packe loss a e in p e-de ined
ange. So we a e able o enable 88 % o all incoming
equi emen s. In his case he e is no mo e a ailable
bandwid h o o he ype o se ices. When we com-
pa e he esul s o simula ion o a ic model 2 Gaus-
sian app oxima ion me hod allows access o 2.05 imes
mo e use s ha „Measu ed sum“ algo i hm.
7. Conclusion
In he simula ed node we use he ad an age o Gaus-
sian app oxima ion me hod ha i has good esul s
only o a high numbe o use s. The disad an age o
his me hod may be economical. In compa ison wi h
„Measu ed sum“ algo i hm is mo e expensi e. An-
o he disad an age is ha he e a e mo e compu ing
equi emen s. To he decision ha he connec ion is
enabled o ejec ed he e mus be calcula ed h ee al-
ues. Resul an equi ed bandwid h Cg, he p obabili y
o packe loss Ploss and he p obabili y o line o e low
Po e low. „Measu ed sum“ algo i hm needs o calcu-
la ed only one.
Fo all ou simula ions and calcula ions i is ob i-
ous ha he numbe o allowed equi emen s depends
on he a ic sou ce. When we use as a sou ce o e-
qui emen s only VoIP calls as i is in a ic model 1
he e a e e y good esul s wi h Gaussian app oxima-
ion me hod. In he case o a ic model 2 when
we use as a sou ce o equi emen s VoIP calls and
da a download he e a e s ill good esul s o Gaus-
sian app oxima ion me hod. In bo h cases Gaussian
app oxima ion me hod is be e han „Measu ed sum“
algo i hm. Bu o he a ic model 2 we can expec
ha he e a e AC me hods which a e able o allow ac-
cess o mo e connec ions han Gaussian app oxima ion
me hod. Fo example i could be di usion me hod.
Small and cheap solu ion o imp o e he me hod
o sys em may implemen ce ain wa nings o queues.
F om he iew o he se ing o he use ’s de ice and
he se ings o he ne wo k i sel in eaching maximum
capaci y, he de ice may be eminded o wai un il he
a ailable capaci y. Capaci y may be a e eleased al-
loca ed and de ice may be no i ied abou a ailabili y
o same as i is now de ice can be ejec ed. In he las
case, he de ice will ha e o y he new connec ion.
The i s cases should p e e ably wai in he queue and
would a he be se iced. This would o cou se lead o
g ea e use sa is ac ion bu mainly o mo e e icien
dis ibu ing o he sou ce.
Ou simula ions p o e ha choose o AC me hod
has high impac o a ic managemen . Bu we mus
be e y ca e ul by choosing o AC me hod because i
always depends on a ic sou ce. The e a e o he con-
di ions o only VoIP a ic and o VoIP+ da a a ic.
Acknowledgmen
This a icle was c ea ed wi h he suppo o he Min-
is y o Educa ion, Science, Resea ch and Spo o he
Slo ak Republic wi hin he Resea ch and De elopmen
Ope a ional P og amme o he p ojec "Uni e si y
Science Pa k o STU B a isla a", ITMS 26240220084,
co- unded by he Eu opean Regional De elopmen
Fund.
Re e ences
[1] YEGANEH, H., A. H. DARVISHAN and M.
SHAKIBA. NGN unc ional a chi ec u e o e-
sou ce alloca ion and admission con ol. In:
9 h In e na ional Con e ence on Telecommuni-
ca ion in Mode n Sa elli e, Cable, and B oad-
cas ing Se ices. Nis: IEEE, 2009, pp. 533–
539. ISBN 978-1-4244-4383-3. DOI: 10.1109/TEL-
SKS.2009.5339452.
[2] VOZNAK, M. and J. ROZHON. App oach
o s ess es s in SIP en i onmen based on
ma ginal analysis. Telecommunica ion Sys ems.
2013, ol. 52, iss. 3, pp. 1583–1593. ISSN 1018-
4864. DOI: 10.1007/s11235-011-9525-1.
[3] BARONAK, I., R. TRSKA and P. KVACKAJ.
CAC – Connec ion admission Con ol in ATM
Ne wo ks. Jou nal o Elec ical Eginee ing. 2005,
ol. 56, iss. 5–6, pp. 162–164. ISSN 1335-3632.
[4] YERIMA, S. Implemen a ion and e alua ion
o Measu emen – Based Admission Con ol
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 287
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 13 |NUMBER: 4 |2015 |SPECIAL ISSUE
Schcemes Wi hin a Con e ged Ne wo ks QoS
Managemen F amewo k. In e na ional Jou nal
o Compu e Ne wo ks And Communica ions
(IJCNC). 2011, ol. 3, no. 4, pp. 137–152.
ISSN 0975-2293. DOI: 10.5121/ijcnc.2011.3410.
[5] KOVAC, A., M. HALAS, M. ORGON and M.
VOZNAK. E-model Mos es ima e imp o emen
h ough ji e bu e packe loss modelling. Ad-
ances in Elec ical and Elec onic Enginee ing.
2011, ol. 9, no. 5, pp. 233–242. ISSN 1804-3119.
DOI: 10.15598/aeee. 9i5.542.
[6] VOZNAK, M. and J. ROZHON. Me hodology o
SIP in as uc u e pe o mance es ing. WSEAS
T ansac ions on Compu e s. 2010, ol. 9, iss. 9,
pp. 1012–1021. ISSN 1109-2750.
[7] CHROMY, E., M. JADRON, M. KAVACKY and
S. KLUCIK. Admission Con ol in IMS Ne wo ks.
Ad ances in Elec ical and Elec onic Enginee ing.
2011, ol. 9, no. 5, pp. 373–379. ISSN 1804-3119.
DOI: 10.15598/aeee. 11i5.875.
[8] LIAO, J., J. WANG, T. LI, J. WANG
and X. ZHU. A oken-bucke based no i ica-
ion a ic con ol mechanism o IMS p es-
ence se ice. Compu e Communica ions. 2011,
ol. 34, iss. 10, pp. 1243–1257. ISSN 0140-3664.
DOI: 10.1016/j.comcom.2010.12.017.
[9] VOZNAK, M. and J. SAFARIK. DoS a acks a -
ge ing SIP se e and imp o emen s o obus ness.
In e na ional Jou nal o Ma hema ics and Com-
pu e s in Simula ion. 2012, ol. 6, iss. 1, pp. 177–
184. ISSN 1998-0159.
[10] CHROMY, E., J. DIEZKA, M. KAVACKY and
M. VOZNAK. Ma ko models and hei use o
calcula ions o impo an a ic pa ame e s o
con ac cen e . WSEAS T ansac ions on Com-
munica ions. 2011, ol. 10, iss. 11, pp. 341–350.
ISSN 1109-2750.
[11] BARONAK, I. and P. KVACKAJ. S a ical CAC
Me hods in ATM. Radioenginee ing. 2006, ol. 15,
iss. 2, pp. 53–56. ISSN 1805-9600.
[12] KARAMDEEP, S. and K. GURMEET. Connec-
ion Admission Con ol Me hods Based on Fuzzy
Logic. In: 6 h In e na ional Mul i Con e ence on
In elligen Sys ems and Nano echnology. Yamu-
nanaga : Ins i u e o Science and Technology,
2010, pp. 68–70.
[13] CHAMRAZ, F. and I. BARONAK. Con ibu ion
o he managemen o a ic in ne wo ks. Ad-
ances in Elec ical and Elec onic Enginee ing.
2014, ol. 12, no. 4, pp. 334–340. ISSN 1804-3119.
DOI: 10.15598/aeee. 12i4.1213.
[14] CHROMY, E., M. JADRON and T. BEHUL. Ad-
mission Con ol Me hods in IP Ne wo ks. Ad-
ances in Mul imedia. 2013, ol. 2013, iss. 1, pp. 1–
7. ISSN 1687-5680. DOI: 10.1155/2013/918930.
[15] DE RANGO, F., M. TROPEA, P. FAZIO and S.
MARANO. Call Admission Con ol o Agg ega e
MPEG-2 T a ic O e Mul imedia Geo-Sa elli e
Ne wo ks. IEEE T ansac ions on B oadcas ing.
2008, ol. 54, iss. 3, pp. 612–622. ISSN 0018-9316.
DOI: 10.1109/TBC.2008.2002716.
[16] DE RANGO, F., M. TROPEA, P. FAZIO and
S. MARANO. Call admission con ol wi h s a-
is ical mul iplexing o agg ega e MPEG a -
ic in a DVB-RCS sa elli e ne wo k. In: IEEE
Global Telecommunica ions Con e ence, GLOBE-
COM ’05. S . Louis: IEEE, 2005, pp. 3231–
3236. ISBN 0-7803-9414-3. DOI: 10.1109/GLO-
COM.2005.1578372.
Abou Au ho s
Filip CHAMRAZ was bo n in Malacky, Slo akia,
on Ma ch 1992. Nowadays he is a s uden a he
Ins i u e o Telecommunica ions, Facul y o Elec ical
Enginee ing and In o ma ion Technology, Slo ak
Uni e si y o Technology in B a isla a. He comple ed
bachelo deg ee in 2014. He ocuses on p oblems
o a ic managemen , IMS and admission con ol
me hods.
I an BARONAK was bo n in Zilina, Slo akia,
on July 1955. He ecei ed Mas e o Science deg ee
(elec ical enginee ing) om Slo ak Technical Uni-
e si y B a isla a in 1980. Since 1981 he has been a
lec u e a Depa men o Telecommunica ions Slo ak
Uni e si y o Technology in B a isla a. Nowadays
he wo ks as a p o esso a Ins i u e o Telecom-
munica ions o Facul y o Elec ical Enginee ing
and In o ma ion Technology, Slo ak Uni e si y o
Technology in B a isla a. Scien i ically, p o essionally
and pedagogically he ocuses on p oblems o digi al
swi ching sys ems, elecommunica ion ne wo ks,
elecommunica ion managemen (TMN), IMS, VoIP,
QoS, p oblem o op imal modeling o p i a e elecom-
munica ion ne wo ks and se ices.
c
2015 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 288