ADAPTIVE SCHEMES FOR PACKET DATA IN A
DS-CDMA ENVIRONMENT
O iol Sallen , Ramon Agus i
Depa amen de Teo ia del Senyal
i
Comunicacicons
Uni e si a Poli 6cnica de Ca alunya
C/ G an Capi l s/n
-
08034 Ba celona
-
Spain1
Tel. +34-3-401.71.97 Fax +34-3-401.72.00
E-mail
:
o iol
@
xaloc.upc.es
Abs ac
This pape p esen s a no el Adap i e DS-CDMA Slo ed-
ALOHA packe andom access scheme
o
imp o ing he
h oughpu
o
he con en ional DS-CDMA Slo ed-ALOHA
sys em. Fo his pu pose a Mobile-assis ed algo i hm is
en isaged o con ol he change o he ansmission a e
acco ding o he a ic load. This algo i hm e ealed ha he
op imum beha io may
be
almos eached a a low complexi y
cos . Mo eo e , he p oposed algo i hm was ound o
be
obus o in e cell in e e ence.
I. INTRODUCTION
Slo ed ALOHA (S-ALOHA) has been widely ecognized
o
packe adio applica ions when he da a ansmi ed om he
e minals a e bu s y
o
e en ecen ly o oice cellula
communica ions
[
11.
S-ALOHA DS-CDMA scheme allows
se e al use s o ansmi a he same ime using a di e en
sp eading code o each one [2], [3],
so
ha unsuccess ul
ansmissions a e caused en i ely by mul iple access
in e e ence, p o ided negligible he mal noise e ec s a e
assumed.
This pape is in ended o imp o e he h oughpu achie ed
wi h a mul i- ecei e DS-CDMA S-ALOHA when a ligh
load is o e ed o he sys em. Fo his pu pose we will
ocus
on he ansmission a e used ins ead o he access p o ocol
i sel . In pa icula , di e en ansmission a es a e p oposed
o
be
used in o de o u ilize he mos sui able a e acco ding
o he channel load a any ime slo . Then, a as and simple
algo i hm ha command he p ocessing gain o he Sp ead
Spec um DS scheme as a unc ion o he channel load is
en isaged. Fo low channel loads, a high CDMA p ocessing
gain is no longe necessa y and i can be educed
so
as o
inc ease he ac ual bi a e. The e o e, in an ex eme si ua ion
wi h a p ocessing gain equal o one, all he use s would
access he same channel a hei maximum bi a e acco ding
o a pu e S-ALOHA.
In a eal cellula scena io, in e e ence om neighbough cells
mus be aken in o accoun . To keep hese in e e ence a a
easonable le el, he concep o cell clus e a ises in
echniques wi h a FDMA componen . Howe e , i is he case
nei he o DS-CDMA based sys ems no
o
cellula
S-
ALOHA
[I],
whe e a euse ac o o
1
is no mally conside ed.
Accep ing ha a h oughpu dec ease will ollow when
conside ing ihe cellula s uc u e, we a e in e es ed on
assessing his dec ease in e ms o h oughpu as well as
whe he he p oposed change o he ansmission a e
algo i hm is s ill alid
o
no . We will show ha no changes
a e equi ed in he algo i hm, despi e he in e e ence is no
only dependen on he o e ed load bu also on he
p opaga ion condi ions ( ha s ongly a ies he in e e ence
le el a ec ing o a e e ence use ). Fu he mo e, some powe
con ol based policies a e p oposed o imp o e he sys em
pe o mance.
The pape is o ganized as ollows. In Sec ion
2
an analy ical
model
o
he DS-CDIMA S-ALOHA sys em is p esen ed,
which is used o e alua e he op imum achie able h oughpu
wi h an adap i e change o ansmission a e algo i hm. In
Sec ion 3 a pa icula S-ALOHA adap i e algo i hm is
p oposed and assessed. In Sec ion 4 a cellula model is
conside ed and he impac o o he -cell in e e ence is
a ained. The change o he ansmission a e algo i hm is
also s udied unde lhese condi ions and limi ed-powe
s a egies a e e alua ed. Conclusions close he pape in
Sec ion
5.
11. MODEL FOR A S-ALOHA DS-CDMA
ACCESS SYSTEM
F om now onwa ds
N
egis e ed use s will be conside ed.
These use s can be in wo di e en ope a ion modes: ‘idle
mode’ and ‘backlogged mode’. In he o me mode he e is no
packe o be e ansmi ed and new packe s a e gene a ed wi h
p obabili y
po.
Te minals en e he backlogged mode when an
a emp o ansmi a new packe ails. In his mode, he
e ansmission o he backlogged packe
occu s
in any gi en
slo wi h p obabili y
p
.
While in he backlogged mode he
use does no gene a e any new packe .
The p ocess de ined by he numbe o backlogged use s a he
beginning o he
k
h
slo is a Ma ko chain, whose
pe o mance has been s udied among o he s in [4]. When
conside ing DS-CDMA, he o mula ion mus be modi ied as
i was done in
[5].
Analy ical h oughpu measu emen s can
be ob ained om he ma ko ian model.
0-7803-3659-3/97
$1
0.00
01
997
IEEE
1019
A. S-ALOHA DS-CDMA ACCESS
SYSTEM
A
BPSK
DS-CDMA access sys em wi h a p ocessing gain
gi en by
Gp
is conside ed. All use s ha e been assigned
andom PN signa u e sequences. A pe ec powe con ol
capable o mi iga ing adings o he channel is in oduced.
Wi h a iew o achie ing his, a con inuous link be ween
BS
and MS
is
needed
so
ha he mobile ansmi s a a e
when
he e is in o ma ion o be sen and a a e
’
(in gene al much
lowe ) when he e minal is no ac i e.
’
should be high
enough o allow he ading on he uplink pa h o be acked by
upda ing ansmi ed powe in esponse o he BS commands.
An ins an aneous powe con ol pe mi s a huge imp o emen
in he sys em pe o mance when compa ed wi h an open loop
powe con ol, and i is, in ac , conside ed in he al eady
ope a i e
[6]
and he p oposed CDMA sys ems
[7]
espec i ely. Al hough an ins an aneous powe con ol could
be en isaged o packe adio on a packe by packe basis
[8],
we ha e e ained he con inuous link app oach because in his
case no synch oniza ion o e head a he beginning o ime
slo would be equi ed, since he mobile is al eady
synch onized. Whe he con inuous powe con ol is e ained
o no i is in any case i ele an o he p oposed S-ALOHA
scheme aken o g an ed ha a h oughpu dec ease should
be conside ed in he non-con inuous packe based powe
con ol app oach.
B.
S-ALOHA DS-CDMA PERFORMANCE
By assuming an ideal ins an aneous powe con ol, he
channel can be seen as an AWGN i we use he gaussian
hypo hesis o model he in e e ence o igina ed by o he use s
[9],
[2].
Unde hese condi ions he ollowing exp essions
hold o he e alua ion o he
BER
whe e
IZ
is he numbe o simul aneous use s. The p obabili y
o de ec ing co ec ly a packe con aining
aB
bi s is
Keeping he ansmission bandwid h cons an , and aking as a
e e ence a ansmission a e o
bi ds,
packe leng h o
B
bi s and a p ocessing gain
Gp,
he ansmission a e could
be
inc eased by a ac o
a
(
a>l)
a he expense o a educ ion in
he sp ead capaci y. Tha is, ansmi ing a a e
a
bids
would allow a p ocessing gain o
G,Ja
.
Howe e , in
exchange o his educ ion in p ocessing gain,
aB
bi s pe
ime slo can be alloca ed ins ead o only
B.
a emp ed ansmissions pe ime slo
-,
and
po
is se equal o
p
so
ha
po=p,=p=G/N
.
G
is assumed o a y slowly.
Mo eo e , o nume ical esul s
‘
has
beem
se o ze o.
Al hough he e a e no es ic ions on he alues
o
a,
o he
sake o b e i y and cla i y only h ee a es
, 2
and
4
bi s/.
(a=l,2,4)
will be conside ed in he sequel.
I is clea om Fig.
1
ha he highe a e
(4
bi ds)
is
in e es ing when he sys em is ligh ly loaded: mo e bi s pe
packe can be sen h ough he channel, since he li le
in e e ence obse ed pe mi s co ec ansmission. When he
o e ed load inc eases so does he in e e ence le el, and
he e o e he a e
4
bi s/s
is no longe in e es ing because o
he highe
BER
compa ed o he
2
bi sh
a e. Using
2
bi ds
in his ange is be e han
4
bi s/.
in he sense ha a highe
h oughpu is achie ed al hough he packe con ains hal he
numbe
o
bi s pe packe (e o s occu qui e o en wi h
4
bi s/s,
bu
2
bi s/s
can s ill bea he in e e ence because o
he highe
Gp
compa ed o ha o
4
bi s/.
a e). Fo highe
o e ed loads he same ade-o appea s be ween
2
bi s/.
and
1
bi sh.
7000
1
I
6000
5
5000
5
4000
m
ln
I
2
3000
Q
g
2000
1000
Fig.
1.
Th oughpu pe o mance o di e en
ansmission a es.
C. ADAPTIVE S-ALOHA DS-CDMA
In he ligh o he abo e esul s, an algo i hm able o change
he ansmission a e used by he
MS
as a unc ion o he
a ic load o he sys em could be o eseen
so
ha he
maximum possible h oughpu could always be ob ained.
The op imum h oughpu a ained can
be
analy ically
ob ained om he Ma ko model, and his esul can
be
la e
used as a e e ence o assess he pe o mance o he
ansmission a e change algo i hm p oposed below. In o de
o
a ain his op imum pe o mance, he bes combina ion o
ansmission a es should be ob ained p o ided ha
n
simul aneous use s a e p esen . The esul ing op imal able
would be eached a e an exhaus i e sea ch, as i is u he
explained in
[lo].
Fig.
2
shows his op imum h oughpu in
compa ison wi h he indi idual beha io s.
The achie able global h oughpu wi h di e en ansmission
a es ob ained wi h he Ma ko model can
be
seen
in
Fig.
1,
whe e
G
s ands o he o e ed load
-
a e age numbe o
1020
7000
l
7000
6000
-
5
5000
L
4000
m
cn
c
3000
13
@
2000
c
1000
Fig.
2.
S-ALOHA DS-CDMA op imum h oughpu .
111. CHANGE
OF
THE TRANSMISSION RATE
ALGORITHM
In his sec ion a p oposal o an adap i e change o he
ansmission a e algo i hm is add essed. The basic idea
consis in sensing he a ic load h ough he channel in
o de o accommoda e ansmission a es acco dingly.
The p oposed simple algo i hm ca ied ou by he mobile
s a ion (MS) wo ks as ollows: each e minal aces i s own
e olu ion du ing he ansmission ime, ha is, e minals
coun hei success ul and e oneous packe s. In he absence
o e o s he mobile will assume a low a ic load and ies o
use
a
highe ansmission a e. The h oughpu should be
inc eased in his way. I e o s occu , he mobile decides ha
he channel is oo loaded and ies a lowe ansmission a e.
In his case ewe bi s pe packe a e ansmi ed, bu a global
imp o emen o he h oughpu should also ollow because
hese bi s can
be
now de ec ed co ec ly since p ocessing gain
inc eases acco dingly.
Le
us
no e ha his decision is aken by he mobile wi hou
any exchange o in o ma ion wi h he Base S a ion (BS)
excep o he packe s acknowledgmen . E en he
MS
does
no need o indica e i s choice o ansmission a e be o e
using i because he e y same BS could be able o de ec
which one is a i ing.
Speci ically, he MS only needs o es ablish wo pa ame e s:
he numbe o consecu i e packe ailu es be o e changing o
a lowe a e
(max- )
and he numbe o consecu i e packe
successes be o e ying a highe a e
(min-suc).
Fig. 3 shows he h oughpu a ained o
N=60
egis e ed
mobiles. Fig.
1
has been aken as a e e ence in o de o
app ecia e whe he mobiles choose he mos sui able a e o
no . In spi e o he simplici y o he algo i hm, he en elope o
he h ee indi idual g aphs is almos eached, and his is no
a om
he
op imum
beha io .
Resul s
ha e
been
ob ained
wi h
m- -1
and
min_suc='l,
which is he bes choice, once
o he possibili ies ha e been s udied.
6000
-
5
5000
4000
m
c
1
lz
3
2
3000
2000
c
I000
0
0
10
20
30
40 50
60
G
Fig.
3.
Th oughpu pe o mance o MS algo i hm.
IV. MS ALGORITHM IN A CELLULAR
ENVIRONMENT
When conside ing mul iple-cell CDMA sys ems, which will
no mally be he case, he si ua ion becomes mo e
complica ed. The e a e basically wo new p oblems ha mus
be aken in o accoun :
1-.
The in e e ence le el o igina ing om use s in
he o he cells. This in e e ence a ies no only acco ding o
he a enua ion in he pa h o he subsc ibe 's cell si e, bu also
in e sely o he a enua ion om he in e e ing use o his
own
cell si e, which h ough powe con ol by ha cell si e
may inc ease
o
dec ease he in e e ence o he desi ed cell
si e.
2-.
The necessa y hando e p ocedu es. In o de o
keep he gene al in e e ence le el as low as possible, i would
be desi able o each mobile o be a ached o he cell si e o
which he adio pa h a enua ion is minimized. Hando e
decisions a e usually based on pilo -s eng h measu emen s
( adio dis ance measu emen s). Howe e , some ime is
necessa y o powe measu emen a e aging and o hando e
execu ion.
In o de
o
e alua e he
MS
algo i hm in a cellula
en i onmen se e al compu e simula ions ha e o
be
ca ied
ou , and he i s equi emen is o es ablish a channel model.
The p opaga ion a enua ion is gene ally modeled as he
p oduc o
a
shadowing ading and a sho - e m ading. The
o me is due o he ei ain con igu a ion, whe eas he la e
is la gely due o inul ipa h e lec ions. An accep ed
ep esen a ion o he ecei ed powe is
(3)
whe e
P,
is
he
ecei ed
powe ,
P
is
he
ansmi ed powe ,
a2
is an exponen ial andom a iable, is he dB a enua ion
due o shadowing, wi h ze o mean and s anda d de ia ion
(3
1021
MAX-POW means educing he in e e ence caused o bo h
his
own
and he o he cell si es. Howe e , i makes i mo e
di icul o he powe con ol o main ain he signal e e ence
le el a he base s a ion o he single use . The e e se
applies i MAX-POW is inc eased. Which o hese
phenomena has mo e impac on he global h oughpu
depends on he le el o which he powe is limi ed as well as
on he o e ed load, which de e mines he o al in e e ence
le el.
Resul s o di e en alues o MAX-POW a e p esen ed in
Fig. 6. Only he pe o mance o he MS algo i hm is shown,
and he gain ob ained when limi ing he maximum
ansmi ed powe can be no ed. The h oughpu deg ada ion
o low loads when dec easing MAX-POW could be a oided
by adap i ely changing his pa ame e join ly wi h he
ansmission a e, acco ding o he MS algo i hm. The e o e,
he in e e ence e ec s o Rayleigh ading could in p ac ice be
s ongly mi iga ed.
4500
I
4000
1
.........
;
........
/.:?-
...................
,
...........
1
.L.
-._
~
,’,
3500
...........
;..
.....
i.
.:
...........
..:...
......
..;
.;Y:,
.!
...........
i/’:
,,
.
<.
1
2
3000
2500
b-
I
c
3
8
2000
0
OL
E
1500
1000
500
.......
...............
.....
.....
........
.;
..............
,
..........
......
....
..........
.;
...........
.I
.
....
I
....
1.1..
I
....
I
....
I
0
10
20
30
40
50
60
G
Fig. 6. A powe limi a ion condi ion is imposed.
V.
CONCLUSIONS
A new Adap i e DS-CDMA S-ALOHA echnique o packe
mobile communica ions access based on he choice o he
mos sui able ansmission a e in any ime slo has been
add essed. The mobile assis ed algo i hm allows
us
o ob ain
almos he maximum a ainable pe o mance a a e y low
complexi y cos , and by a ou pe o ms hose ob ained wi h a
con en ional DS-CDMA S-ALOHA scheme. Simula ions in a
cellula en i onmen show ha he p oposed algo i hm is also
obus o in e cell in e e ence. Mo eo e , limi ing he
maximum ansmi ed powe p o ed o be a good policy o
gene al h oughpu imp o emen .
VI.
ACKNOWLEDGMENTS
The wo k desc ibed in his pape was ca ied ou wi hin he
CICYT TIC94-0870-C02-01 p ojec , in he amewo k o
Na ional Plan o Spain.
VII.
REFERENCES
[l] M. ZORZI,
S.
PUPOLIN, “Slo ed
ALOHA
o High-
Capaci y Cellula Communica ions”, IEEE T ansac ions on
Vehicula Technology, Vol. 43, No. 4, No embe 1994, pp.
1011-1021.
[21 D. RAYCHAUDHURI, “Pe o mance Analysis o Random
Access Packe -Swi ched Code Di ision Mul iple Access
Sys ems”, IEEE T ansac ions on Communica ions, Vol. Co n.
29, No. 6, June 1981, pp. 895-901.
[3] R. K. MORROW, J.S. LEHNERT “Packe Th oughpu in
Slo ed ALOHA DSISSMA Radio Sys ems wi h Random
Signa u e Sequences”, IEEE T ansac ions on Communica-
ions, Vol. Co n. 40, No. 4, July 1992, pp. 1223-30.
[4] L. KLEINROCK,
S.S.
LAM, “Packe Swi ching in a
Mul i-access B oadcas Channel: Pe o mance E alua ion
”,
IEEE T ansac ions on Communica ions, Vol. Con-23, No.
4, Ap il 1975, pp.410-422.
[5]
Z.
LIU, M. EL
ZA.RKI,
“Pe o mance Analysis o DS-
CDMA wi h Slo ed ALOHA Random Access o Packe
PCNs
”,
Wi eless Ne wo ks 1 (1995), pp. 1-16.
[6] A. SALMASI, K.
S.
GILHOUSEN, “On he Sys em
Design Aspec s o CDlMA Applied o Digi al Cellula and
Pe sonal Communica iions Ne wo ks”, P oc. 4lS IEEE
Vehicula Technology Con e ence, S . Louis, May 1991.
[7] A. BAIER
e
al.,
“Design S udy o a CDMA-Based
Thi d-Gene a ion Mobile Radio Sys em
”,
IEEE Jou nal on
Selec ed A eas in Connmunica ions, Vol. 12, No. 4, May
[8] M. J. McTIFFIN
‘e
al.,
“Mobile Access o an ATh4
Ne wo k Using a CDMA Ai In e ace
”,
IEEE Jou nal on
Selec ed A eas in Conimunica ions, Vol. 12, No. 5, June
[9] M. B. PURSLEY, “Pe o mance E alua ion o Phase-
Coded Sp ead-Spec um Mul iple-Access Communica ion
-
Pa I: Sys em Analysis
”,
IEEE T ansac ions on Communi-
ca ions, Vol. Com.-25,
No.
8, Augus 1977, pp.795-799.
[lo]
0.
SALLENT,
F:.
AGUSTI, “A Mobile Con olled
Algo i hm o Imp o ing he Th oughpu in a S-ALOHA DS-
CDMA Sys em”, P oceedings o he 7 h Pe sonal, Indoo and
Mobile Radio Communilca ions (PIMRCP6), pp. 1192-96.
[ll] M. KWOK, H. WANG, “Adjacen Cell In e e ence
Analysis o Re e se-Lin k in CDMA Cellula Radio Sys ems”
Vehicula Technology Con e ence VTC’95, pp.446-450.
[12] K.
S.
GILHOUSIZN
e
al.
,
“On he Capaci y o a
Cellula CDMA Sys e n”, IEEE T ansac ions on Vehicula
Technology,
Vol.
40,
No.
2, May 1991, pp.303-312.
[13] M. GUDMUNDSON, “Co ela ion
Model
o Shadow
Fading in Mobile Radio Sys ems”, IEE Elec onic Le e s,
Vol. 27,
No.
23, 7 h No embe 1991, pp. 2145-2146.
1994, pp. 733-743
.
1994, pp. 900-908
.
1023