SIR: A New Wi eless Senso Ne wo k Rou ing
P o ocol Based on A ificial In elligence
Julio Ba bancho, Ca los Le´on, Ja ie Molina, and An onio Ba bancho
Depa men o Elec onic Technology, Uni e si y o Se ille,
C/ Vi gen de A ica, 7. Se ille 41011, Spain
Tel: (+034) 954 55 71 92, Fax: (+034) 954 55 28 33
{jba bancho, cleon, jmolina, ayboc}@us.es
Abs ac . Cu en ly, Wi eless Senso Ne wo ks (WSNs) a e o med by
hund eds o low ene gy and low cos mic o-elec o-mechanical sys ems.
Rou ing and low powe consump ion ha e become impo an esea ch is-
sues o in e connec his kind o ne wo ks. Howe e , con en ional Quali y
o Se ice ou ing models, a e no sui able o ad hoc senso ne wo ks,
due o he dynamic na u e o such sys ems. This pape in oduces a new
QoS-d i en ou ing algo i hm, named SIR:Senso In elligence Rou ing.
We ha e designed an a ificial neu al ne wo k based on Kohonen sel
o ganizing ea u es map. E e y node implemen s his a ificial neu al
ne wo k o ming a dis ibu ed in elligence and ubiqui ous compu ing
sys em.
Keywo ds: Wi eless senso ne wo ks (WSN); Ad hoc ne wo ks, Qual-
i y o se ice (QoS); Rou ing; A ificial neu al ne wo ks (ANN); Sel -
O ganizing Map (SOM).
1 In oduc ion
Due o he senso ea u es (low-powe consump ion, low adio ange, low mem-
o y, low p ocessing capaci y, and low cos ), sel -o ganizing ne wo k is he bes
sui able ne wo k a chi ec u e o suppo applica ions in such a scena io. Goals
like efficien ene gy managemen , high eliabili y and a ailabili y, communica ion
secu i y, and obus ness ha e become e y impo an issues o be conside ed.
Many esea ch cen e s in he whole wo ld ha e ocused hei in es iga ions in
his kind o ne wo ks [1]. We p esen in his pape a new ou ing algo i hm which
in oduces a ificial in elligence (AI) echniques o measu e he QoS suppo ed
by he ne wo k.
The wi eless senso ne wo ks (WSN) a chi ec u e as a whole has o ake in o
accoun diffe en aspec s, such as he p o ocol a chi ec u e; Quali y-o -Se ice,
dependabili y, edundancy and imp ecision in senso eadings; add essing s uc-
u es, scalabili y and ene gy equi emen s; geog aphic and da a-cen ic add ess-
ing s uc u es; agg ega ing da a echniques; in eg a ion o WSNs in o la ge
ne wo ks, b idging diffe en communica ion p o ocols; e c. [2].
2 SIR: Senso In elligence Rou ing
The necessi y o connec i i y among nodes in oduces he ou ing p oblem. In
a WSN we need a mul i-hop scheme o a el om a sou ce o a des iny. The
p oblem is sol ed by a echnique called ne wo k backbone o ma ion.
We p opose a modifica ion on Dijks a’s algo i hm o o m he ne wo k back-
bone, wi h he minimum cos pa hs om he base s a ion o oo , , oe e y
node in he ne wo k. We ha e named his algo i hm Senso In elligence Rou ing,
SIR, which is desc ibed as ollows in able 1.
Table 1. SIR algo i hm
#1: Se up phase: #3: ∀ i∈T∩Γ( j)calcula e i:= d( j)+wji
d( )=0 I i<d( i)dod( i)= i
d( i)=

w i i i∈Γ( )
∞i i/∈Γ( )

#4: I |T|>0go o#2
Γp( i)=

i
i∈Γ( )
0i i/∈Γ( )

I |T|=0s op
#2: Find a j∈Tsuch as
d( j)=min{d( i)| i∈T}
Do T=T−{ j}
Once i is designed he backbone o ma ion algo i hm, we ha e o define he
way o measu ing he edge weigh pa ame e , wij .
We use a QoS defini ion based on h ee ypes o QoS pa ame e s: imeliness,
p ecision and accu acy. Due o he dis ibu ed ea u e o senso ne wo ks, ou
app oach measu es he QoS le el in a sp ead way, ins ead o an end- o-end
pa adigm. Each node es s e e y neighbo link quali y wi h he ansmissions
o a specific packe named ping. Wi h hese ansmissions e e y node ob ains
mean alues o la ency, e o a e, du y cycle and h oughpu . These a e he ou
me ics we ha e define o measu e he ela ed QoS pa ame e s.
Once a node has es ed a neighbo link QoS, i calcula es he dis ance o oo
using he ob ained QoS alue. The exp ession 1 ep esen s he way a node i
calcula es he dis ance o oo h ough node j,whe eqos is a a iable which
alue is ob ained as an ou pu o a neu al ne wo k. This ool is desc ibed in
sec ion 2.1.
d( i)=d( j)·qos (1)
2.1 SOM: Sel O ganizing Map
One o he mos powe ul mechanism de eloped in AI is he Sel -O ganizing
Map (SOM) model [3], c ea ed by Teu o Kohonen in 1982, a he Uni e si y o
Helsinky, Finland.
In SOM we can dis inguish wo phases: lea ning phase,andexecu ion phase.
SOM gi es an ou pu deno ed by qos. This alue is e u ned by a unc ion
Θ defined by he SOM use , acco ding o his aims. Θ depends on he winning
neu on: qos =Θ(g). In sec ion 3 we define his unc ion.
3 Pe o mance E alua ion by Simula ion
Due o he desi e o e alua e he SIR pe o mance, we ha e c ea ed wo simu-
la ion expe imen s unning on ou wi eless senso ne wo k simula o OLIMPO.
E e y node in OLIMPO implemen s a neu al ne wo k (online p ocessing).
We ha e ocused ou simula ion on a wi eless senso ne wo k composed by
4000 nodes co e ing an a ea o 87 Km2. This is he ypical a ea o a eu opean
medium size ci y like Se ille (Spain) o Zu ich (Swi ze land). The densi y o
nodes which a e wi hin he ansmission adius o a node es 7.
Noise influence o e a node has been modelled as an Addi i e Gaussian Whi e
Noise, (AWGN). Noise powe has been modelled as a s ochas ic a iable wi h a
mean alue exp essed as a pe cen age o he an enna sensibili y; and a s anda d
de ia ion exp essed as a pe cen age o i s mean alue.
Ou SOM has a fi s laye o med by ou inpu neu ons, co esponding wi h
e e y me ic (la ency, h oughpu , e o a e and du y cycle); and a second laye
o med by wel e ou pu neu ons o ming a 3x4 ma ix.
Nex , we de ail ou SOM implemen a ion p ocess.
Lea ning Phase. In o de o o ganize he neu ons in a wo dimensional map,
we need a se o inpu samples x( )=[la ency( ), h oughpu ( ),e o - a e( ),du y-
cycle( )]. This samples should conside all he QoS en i onmen s in which a link
communica ion be ween a pai o senso nodes can wo k. Fo ha eason we ha e
o c ea e he special en i onmen s. These scena ios a e implemen ed by diffe en
noise simula ions. In ou esea ch we c ea ed a WSN o e OLIMPO composed by
4000 senso nodes. In his ne wo k, we chose a pai o nodes (le us deno e hem
as 800 and 1250) and in oduced a low powe noise in o one o hem (e.g. 1250).
Acco ding o he inpu equi emen s, we had o measu e he QoS me ics. In ha
sense, we an a ping applica ion 50 imes a node 800. This applica ion pings
a e sen om node 800 o node 1250. Ping equi es acknowledgmen (ACK).
The way node 800 ecei es ACKs will de e mine a specific QoS en i onmen ,
exp essed on he ou elec ed me ics: la ency (seconds), h oughpu (bi s/sec),
e o a e (%) and du y cycle(%) [4]. This p ocess was epea ed 100 imes while
inc easing he noise powe .
Wi h he se o 100 inpu samples we ained ou neu al ne wo k. This p ocess
was implemen ed on a pe sonal compu e using he MATLABneu al oolbox
(offline p ocessing).
Once we had o de ed he neu ons on he Kohonen laye , we iden ified each
one o he se o 100 inpu samples wi h an ou pu laye neu on. Acco ding o
his p ocedu e he se o 100 inpu samples we e dis ibu ed o e he SOM. We
ealized ha inpu samples ob ained om a simila noisy en i onmen and wi h
simila QoS ea u es we e alloca ed in a specific egion o he SOM. Consequen ly
we ob ained a map o med by clus e s, whe e e e y clus e co esponded wi h
a noise le el in oduced a he en i onmen and consequen ly a specific QoS.
Fu he mo e, a synap ic-weigh ma ix is o med, whe e e e y synapsis iden ifies
a connec ion be ween inpu and ou pu laye .
In o de o quan i y he QoS le el we s udied e e y clus e ea u es and as-
signed a alue be ween 0.2 and 10, acco ding o he le el o noise in oduced.
This assignmen was based on ou expe ience as expe s in ne wo ks. In ha
way we defined he ou pu unc ion Θ(i, j),i∈[1,3],j ∈[1,4] wi h wel e alues
co esponding wi h e e y neu on (i, j),i∈[1,3],j ∈[1,4].
Execu ion Phase. E e y senso node measu es he QoS o i s links collec ing
inpu samples and unning he wining neu on elec ion algo i hm. Fo example, i
a specific inpu sample is qui e simila han he synap ic-weigh - ec o o neu on
(2,2), his neu on will be ac i a ed. A e he winning neu on is elec ed, he node
uses he ou pu unc ion Θ o assign a QoS es ima ion, qos. Finally his alue is
employed o modified he dis ance o oo (eq. 1).
Ou SIR algo i hm has been e alua ed by he ealiza ion o wo expe imen s
de ailed as ollows.
Expe imen #1. Fi s , a wi eless senso ne wo k wi h 4000 nodes is c ea ed.
The ne wo k backbone is o med using SIR algo i hm, as de ailed in able 2.
Howe e , no SOM is applied, so, he dis ance om a node i o oo d( i)isno
modified by he neighbo link quali y, qos (eq.1). We ha e called his algo i hm
as No AI algo i hm.
Nex , a high le el o noise is in oduced a nodes 100 and 200,figu e1.c.
Finally, a specific node (e.g. node numbe 300) uns he ‘T ansmi clock o
base s a ion’ applica ion. Node 300 sends 10 packe s o oo o measu e he
la ency. E e y packe con ains ‘clock’ in o ma ion.
(a) (b)
noisy s a ions noisy s a ions
0 20 40 60 80 100 120 140 160 180 200
0
5
10
15
20
25
30
Mean noise (% Ps) (when noise s anda d de ia ion is se o 50% o Mean Noise)
Clock la ency
(sec)
SIR ou ing
algo i hm
NoAI ou ing
algo i hm
(c) (d)
0
50
100
150
200
0
50
100
0
5
10
15
20
25
30
Noise s anda d
de ia ion
(% Mean noise)
Mean noise (% Ps)
Clock la ency
(sec)
Fig. 1. (a) Clock la ency measu emen : No AI pe o mance (b) Clock la ency measu e-
men : SIR e sus No AI (c) Ne wo k backbone o ma ion based on No AI algo i hm
(d) Ne wo k backbone o ma ion based on SIR algo i hm.
Figu e 1.a ep esen s clock la ency depending on he le el o he noise powe
in oduced a nodes 100 and 200.
Expe imen #2. This expe imen is simila han expe imen #1, bu in his
case, he dis ance d( i) is modified by he neighbo link quali y, using equa ion
1. The ne wo k backbone is o med in such a way ha he pa h c ea ed om
he node 300 o oo does no con ain he noisy nodes, figu e 1.d.
Figu e 1.b shows SIR algo i hm pe o mance compa ed wi h No AI algo i hm
pe o mance. As depic ed in his figu e, i he mean noise is low, bo h algo i hm
pe o mances a e excellen . In his case, he pa h om he node 300 o oo
con ains nodes 100 and 200. Howe e , when he mean noise g ows up abo e
he an enna sensibili y SIR algo i hm pe o mance imp o es No AI algo i hm
pe o mance, main aining he QoS. In his case, he pa h om he node 300 o
oo does no con ain he noisy nodes.
4 Conclusion and Fu u e Wo ks
SIR has been p esen ed in his pape as an inno a i e QoS-d i en ou ing algo-
i hm based on a ificial in elligence. This ou ing p o ocol can be used o e wi e-
less senso ne wo ks s anda d p o ocols, such as IEEE 802.15.4 and Blue oo h,
and o e o he well known p o ocols such as A achne,SMACS,EAR,LEACH,
e c.
The inclusion o AI echniques (e.g. neu al ne wo ks) in wi eless senso ne -
wo ks has been p o ed o be an use ul ool o imp o e ne wo k pe o mances.
An addi ional ad an age is he low cos he AI implemen a ion ep esen s.
The g ea effo made o implemen a SOM algo i hm inside a senso node
means ha he use o a ificial in elligence echniques can imp o e he WSN pe -
o mance. Acco ding o his idea, we a e wo king on he design o new p o ocols
using his kind o ools.
Re e ences
1. I.F. Akyildiz, Y. Su, W. Sanka asub amaniam, and E. C¸ayi ci. Wi eless senso
ne wo ks: A su ey. Compu e Ne wo ks, Else ie , 38:393–422, Decembe 2002.
2. F.J. Molina, J. Ba bancho, and J. Luque. Au oma ed me e eading and SCADA
applica ion. Lec u e No es in Compu e Science, Sp inge Ve lag, 2865:223–234,
Oc obe 2003.
3. T. Kohonen. The sel -o ganizing map. In P occedings o he IEEE, olume 78, pages
1464–1480, 1990.
4. R. Iye and L. Klein ock. QoS con ol o senso ne wo ks. In IEEE In e na ional
Con e ence on Communica ions, ICC’03, olume 1, pages 517–521. IEEE P ess,
May 2003.