Comparison of different scenarios for path diversity packet wireless networks
Abstract
This article discusses the use of the principles of network coding in wireless networks to increase the overall robustness of the transmission. Path diversity on the packet level is investigated as a software alternative to classical diversity techniques commonly used to improve wireless devices communication. The article describes the initial stage of the research of network coding in elementary form with network traffic duplication. Objects of investigation are two different types of transmission in two scenarios, where the resulting transmission loss is evaluated with dependency on the packet length.
Full text
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE
© 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 199
COMPARISON OF DIFFERENT SCENARIOS FOR PATH DIVERSITY
PACKET WIRELESS NETWORKS
Pe CHLUMSKY1, Zbynek KOCUR1, Ji i VODRAZKA1
1Depa men o Telecommunica ion Enginee ing, Facul y o Elec ical Enginee ing, Czech Technical Uni e si y in
P ague, Technicka 2, 166 27 P ague 6, Czech Republic
[email p o ec ed], zbynek.kocu @ el.c u .cz, [email p o ec ed] u .cz
Abs ac . This a icle discusses he use o he p inciples
o ne wo k coding in wi eless ne wo ks o inc ease he
o e all obus ness o he ansmission. Pa h di e si y on
he packe le el is in es iga ed as a so wa e al e na i e
o classical di e si y echniques commonly used o
imp o e wi eless de ices communica ion. The a icle
desc ibes he ini ial s age o he esea ch o ne wo k
coding in elemen a y o m wi h ne wo k a ic
duplica ion. Objec s o in es iga ion a e wo di e en
ypes o ansmission in wo scena ios, whe e he
esul ing ansmission loss is e alua ed wi h dependency
on he packe leng h.
Keywo ds
Op imiza ion, pa h di e si y, wi eless ansmission.
1. In oduc ion
Op imiza ion o wi eless ne wo ks is a e y b oad ield;
ou wo k is ocused on he op imiza ion wi hin he
meaning o ensu ing he desi ed balance be ween he
ne wo k h oughpu , delay and packe loss size. The e a e
si ua ions when maximal h oughpu and minimal delay
a e equi ed (e.g. oice se ices), howe e , he e a e
si ua ions when e e y single packe is impo an and mus
be deli e ed o i s des ina ion (e.g. ile ans e ). Ou
app oach is o ind he app op ia e connec ion scheme o
unicas ne wo k using he pa h di e si y. Fu he mo e,
a ious packe loss p obabili ies in sepa a e channels,
esul ing om he usage o mo e o less obus
modula ion echniques, a e an in eg al pa o he design
o wo indi idual schemes. This a icle compa es hese
wo di e en ansmission schemes wi h espec o
p obabili y o packe loss.
2. Op imiza ion Design
The i s pa o his chap e consis s o di e en ways o
op imize he da a ansmission, and hen ocuses on he
desc ip ion o ou ansmission schemes.
2.1. Rela ed Wo k
A common way o imp o e he ansmission
cha ac e is ics in he physical laye is he MIMO
echnology (Mul iple-Inpu Mul iple-Ou pu ), i.e. he use
o mul iple an ennas a he ecei e and ansmi e .
MIMO has wo a ian s; he i s one is based on he
spa ial di e si y which means ansmi ing he same da a
h ough se e al an ennas. Spa ial mul iplexing is he
second one and i s co e unc ion is based on he sending
di e en da a h ough he each o an ennas [1]. O he
op imiza ion echniques in he physical laye a ea o
wi eless ne wo ks a e, o example, dynamic subca ie
assignmen , adap i e powe alloca ion 0, o in e e ence
cancela ion [3]. Op imiza ion in he link laye is ypically
based on some kind o adap abili y in o de o adjus he
p o ocol pa ame e s o he ac ual ne wo k condi ions [4],
scheduling [5], o di e en back o mechanisms [6].
Di e en ype o a icles ocused on he ne wo k
op imiza ion co e s new ou ing me hods, such as
dispe si y ou ing. Unlike con en ional di ec o y ou ing
p ocedu es, which ou e a message along a pa icula pa h
be ween he sou ce and des ina ion, his ou ing
mechanism sub-di ides he message and dispe ses i
h ough he maze o pa hs comp ising he ne wo k [7].
2.2. T ansmission Schemes
In he ollowing sec ion is discussed wo sepa a e
wi eless channels sys em op imizing da a ansmission.
Op imiza ion o h oughpu o esilience, in sys ems wi h
pa h di e si y in use, has i s e lec ion in MIMO me hods
o spa ial di e si y and mul iplexing. P oposed
echniques o ou wo k, on he con a y, a e in ended o
be uni e sal and independen on he physical laye .
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE
200 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING
1) Load Balance Sys em Model
Conside a ansmission sys em wi h wo sepa a e
channels as depic ed in Fig. 1. Sys em wi h uni o mly
dis ibu ed load elies on he assump ion ha i he e a e
wo pa allel da a s eams and each will ansmi s he hal
o inpu da a, hen, hanks o a lowe bi a e o a single
da a channel will be achie ed highe esis ance using
mo e obus modula ion. O e all h oughpu will be he
same as in he basic sys em wi h a single channel.
he impac o ype o modula ion on he packe e o a e
(PER). The bi e o a e depends on he bi o symbol
mapping, o SNR >> 1 (Signal- o-noise a io) and G ay-
coded assignmen is possible o assume ha each symbol
e o causes only one bi e o ; hen he p obabili y o bi -
e o pe ca ie is [8]:
n
bb SNR
b
e c
b
PER
22
3
2
1
1
2
11 1
2
, (1)
Fig. 1: Load balance ansmission sys em.
2) Backup Sys em Model
Sys em wi h da a mi o ing (see Fig. 2) akes ad an age
o he si ua ion ha wo pa allel channels ansmi s he
comple ely same da a. The e is necessa y o ha e a less
obus modula ion, compa ed o he p e ious me hod,
which is capable o ansmi da a wi h he same o e all
h oughpu . Fo example, i he e is a ne wo k wi h wo
channels based on he s anda d IEEE 802.11g and eques
o a 24 Mbps h oughpu , hen wi h he i s sys em is
necessa y o ha e a wo QPSK (Quad a u e Phase Shi
Keying) channels (s anda d de ines bi a e o 12 Mbps).
Wi h he second sys em, one can use a less obus 16-
QAM (Quad a u e Ampli ude Modula ion) modula ion
which is de ined o he bi a e o 24 Mbps in he
men ioned s anda d (see Tab. 1). The o al bi a e will no
inc ease, because o he selec ion jus o he o iginal da a
a he des ina ion, bu due o pa allel ansmission o da a
will be p ese ed high du abili y and e y low e o a e
despi e highe e o a e pe channel caused by less
esilien modula ion echnique.
whe e b is numbe o bi s pe symbol; numbe o symbols
in modula ion cons ella ion M = 2b, n is packe leng h in
bi s and e c(x) is Gaussian complemen a y e o
unc ion.
Modula ion echniques and coding a es, which
indica es he a io o he numbe o bi s o use ul
in o ma ion o he o al numbe o bi s (including he
added edundancy), o each ype o modula ion is
de ined by IEEE s anda d 802.11g [9] and p esen ed in
Tab. 1.
3. Compa ison o Schemes o
Di e en Scena ios
Ou compa ison o p esen ed schemes in e ms o packe
loss wi h espec o packe leng h is di ided in o wo
scena ios. The i s one deals wi h addi i e whi e Gauss.
Fig. 2: Backup ansmission sys em.
2.3. Packe E o and Th oughpu
The backg ound o his comes om he ac ha
obus ness is dec eased wi h he g ea e numbe o
symbols in modula ion cons ella ion. Equa ion 1 shows
noise and wi h di e en modula ions in use o backup
and load balance sys em. The second one is based on he
same p obabili ies o in e up o bo h sys ems and
shows how hese sys ems a e capable o deal wi h hem.
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE
© 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 201
3.1. Packe Loss o Gaussian Noise
Only addi i e whi e Gaussian noise is conside ed in he
i s scena io. The di e ence be ween load balance and
backup sys em is app oxima ely ou o de s o magni ude
g ea e in a o o he load balance sys em. This is so
because sys ems a e designed o ope a e wi h di e en
modula ions and, he e o e, is impo an o wo k wi h
app op ia e p obabili ies o bi e o . These a e possible o
ge om well-known equa ions o a pa icula ype o
modula ion (mo e in [8]).
Fo his compa ison, we ha e chosen 16-QAM
modula ion o he backup sys em channels and he
QPSK modula ion o he load balance sys em.
App op ia e bi e o p obabili ies o 15 dB SNR a e
shown in Fig. 3.
Tab.1: Th oughpu o modula ion echniques and coding a es
de ined in IEEE s anda d 802.11g.
Modula ion
Coding a e
Bi a e
BPSK
1/2
6 Mbps
BPSK
3/4
9 Mbps
QPSK
1/2
12 Mbps
QPSK
3/4
18 Mbps
16-QAM
1/2
24 Mbps
16-QAM
3/4
36 Mbps
64-QAM
2/3
48 Mbps
64-QAM
3/4
54 Mbps
Fig. 3: Bi e o p obabili ies o QPSK and 16-QAM modula ion.
Figu e 4 shows how looks he si ua ion wi h
di e en modula ions in use. The load balance sys em is
signi ican ly mo e esilien agains bi e o s. This is
caused by g ea di e ence o p obabili ies o e o o
QPSK and 16-QAM modula ion.
3.2. Packe Loss o In e up and
In e e ence in Wi eless Channel
Fo compa ison which o schemes is mo e obus and less
e o -p one is good o know how is p obable ha an e o
will happen. The packe e o should be mo e p obable in
he load balance sys em because i is su icien ha e o
occu ed only a one o channels and he packe is los .
On he con a y, when he packe is los in one channel o
he backup sys em, is s ill deli e ed o i s des ina ion by
he second channel. These wo di e en beha iou s a e
desc ibed in ollowing equa ions.
Fig. 4: Compa ison o sys ems wi h di e en modula ions in use.
The p obabili y ha an e o occu s in he in e al
o he packe leng h n in he load balance sys em is:
nn
LppP 21 1111
nn pp 21 1111
. (2)
The p obabili y ha an e o occu s in he in e al
o he packe leng h n in he backup sys em is:
nn
BppP 21 1111
. (3)
whe e p1 and p2 a e p obabili ies o e o in one bi .
Compa ison o hese equa ions o a ious packe
leng hs and o he same in e up e o a e 10-5 o bo h
sys ems is depic ed in Fig. 5. The di e ence be ween load
balance and backup sys em is app oxima ely h ee o de s
o magni ude g ea e in a o o he backup sys em.
The e o e, o a wi eless ansmission a ec ed by ading
o an enna shading [10] is a backup sys em mo e
app op ia e.
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE
202 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING
Fig. 5: Compa ison o sys ems o in e up in channel.
4. Conclusion
This a icle p esen s wo di e en pa h di e si y wi eless
ansmission schemes and hei compa ison o packe loss
o a ious packe leng hs. The i s scheme is based on
load balancing o sen da a in o wo sepa a e channels.
Thanks o lowe bi a e o a single da a channel is
achie ed highe esis ance using mo e obus modula ion.
The second scheme uses bo h ansmission channels o
pa allel sending o he same da a. The e o e, high
esis ance agains e o is achie ed by da a backup.
Compa ison o hese schemes is done om wo poin s o
iew. Fi s ly, o he addi i e whi e Gaussian noise
channel model i is shown ha load balance sys em is a
mo e esilien o all es ed packe leng hs. Subsequen ly,
o he si ua ions wi h signal in e up (e.g. shaded
an enna) he backup sys em p o ed o be mo e sui able.
These esul s will be used o u he wo k on hese
schemes and i s usage o speci ic pu poses.
Acknowledgemen s
This wo k was suppo ed by he G an o he Minis y o
he In e io o he Czech Republic, No. VG20122014095,
„De elopmen o adap able and da a p ocessing sys ems
o high-speed, secu e and eliable communica ion in
ex eme condi ions“, and was esea ched in coope a ion
wi h CERTICON and he S uden ´s g an o Czech
echnical uni e si y in P ague, No.
SGS11/124/OHK3/2T/13.
Re e ences
[1] GESBERT, D., M. SHAFI, P. SMITH and A. NAGUIB. F om
heo y o p ac ice: An o e iew o MIMO space- ime coded
wi eless sys ems. IEEE jou nal on Selec ed A eas in
Communica ions. Ap il 2003, ol. 21, iss. 3, pp. 281-302. ISSN
0733-8716. DOI: 10.1109/JSAC.2003.809458.
[2] GUOCONG, Song and Ye LI. C oss-laye op imiza ion o
OFDM wi eless ne wo ks-pa I: Theo e ical amewo k. IEEE
T ansac ions on Wi eless Communica ions. Ma ch 2005, ol. 4,
iss. 2, pp. 614-625. ISSN 1536-1276. DOI:
10.1109/TWC.2004.843065.
[3] SHABADANOV, S., P., MITRAN and C. ROSEMBERG.
Op imiza ion Using Ad anced Physical Laye Techniques in
Wi eless Mesh Ne wo ks. IEEE T ansac ions on Wi eless
Communica ions. Ap il 2012, ol. 11, iss. 4, pp. 1622-1631.
ISSN 1536-1276. DOI: 10.1109/TWC.2012.070212.111859.
[4] FARAGO, A., A. D. MYERS, V. R. SYROTIUK and G. V.
ZARUBA. A new app oach o MAC p o ocol op imiza ion. In:
Global Telecommunica ions Con e ence, GLOBECOM. San
F ancisco: IEEE, 2000, ol. 3, pp. 1742-1746. ISBN 0-7803-
6451-1. DOI: 10.1109/GLOCOM.2000.891934.
[5] FASOLO, E., M. ROSSI, J. WIDMER and M. ZORZI. On
MAC Scheduling and Packe Combina ion S a egies o
P ac ical Random Ne wo k Coding. In: IEEE In e na ional
Con e ence on Communica ioins. Glasgow: IEEE, 2007, pp.
3582-3589. ISBN 1-4244-0353-7. DOI:
10.1109/ICC.2007.591.
[6] SHI, Q. and C. COMANICIU. C oss-laye MAC Op imiza ion
o Wi eless Senso Ne wo ks. In: IEEE Sa no Symposium.
P ince on: IEEE, 2008, pp. 1-5. ISBN 978-1-4244-1843-5.
DOI: 10.1109/SARNOF.2008.4520060.
[7] MAXEMCHUK, N.F. Dispe si y Rou ing. In: P oceedings o
ICC '75. San F ancisco. 1975, pp. 10-13.
[8] DOBES, Jose a Vacla ZALUD. Mode n Radio Enginee ing
[in Czech]. P aha: BEN, 2006. ISBN 80-7300-132-2.
[9] IEEE S anda d o In o ma ion Technology-Pa 11: Wi eless
LAN Medium Access Con ol (MAC) and Physical Laye (PHY)
Speci ica ions. 2007. A ailable a :
h p://s anda ds.ieee.o g/ge ieee802/download/802.11-
2007.pd .
[10] CHLUMSKY. P., Z. KOCUR and V. MACHULA. Simula ion
o da a ans e om he ae oba ic plane. In: Knowledge in
Telecommunica ion Technologies and Op ics 2011
P oceedings. Szczy k: VSB-Technical Uni e si y o Os a a,
2011, pp.113-116. ISBN 978-80-248-2399-7.
Abou Au ho s
Pe CHLUMSKY was bo n in 1985. He ecei ed his
M.Sc. om he Czech Technical Uni e si y in P ague in
2010. Since 2010 he has been s udying Ph.D. deg ee. His
esea ch in e es s include wi eless ansmission, ne wo k
coding and ne wo k simula ion. He is eaching ne wo k
echnologies, da a ne wo ks and digi al enginee ing. He is
cu en ly in ol ed in p ojec ocused on de elopmen and
op imiza ion o modules o simula ion o communica ion
ne wo ks.
Zbynek KOCUR was bo n in 1982. He ecei ed his
M.Sc. deg ee in elec ical enginee ing om he Czech
Technical Uni e si y in P ague in 2008. Since 2008 he
has been s udying Ph.D. deg ee. He is eaching
communica ion in da a ne wo ks and ne wo king
echnologies. His esea ch is ocused on wi eless
ansmission and da a low analysis, simula ion and
op imiza ion. He is cu en ly ac i ely in ol ed in p ojec s
ocused on high speed da a ansmission om as
mo ing objec s and da a op imiza ion ia sa elli e
ne wo k.
INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE
© 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 203
Ji i VODRAZKA was bo n in P ague, Czech Republic
in 1966. He joined he Depa men o Telecommunica ion
Enginee ing, FEE, CTU in P ague in 1996 as a esea ch
assis an and ecei ed his Ph.D. deg ee in elec ical
enginee ing in 2001. He has been he head o he
T ansmission Media and Sys ems scien i ic g oup since
2005 and became Associa e P o esso in 2008. He
pa icipa es in nume ous p ojec s in coope a ion wi h
ex e nal bodies. Cu en ly he ac s also as ice-head o he
Depa men .