Monitoring wireless sensor network nodes with a low intrusion hybrid monitor

UNIVERSITY POLYTECHNIC OF VALENCIA
MASTER DEGREE ON COMPUTER ENGINEERING
MASTER FINAL WORK
MONITORING WIRELESS SENSOR NETWORK
NODES WITH A LOW INTRUSION HYBRID
MONITOR
Au ho :
Ma lon Renné Na ia Mendoza
Tu o s:
José Ca los Campelo, Albe o Bonas e Pina
Valencia, Feb ua y 2015
To my amily, my suppo and encou agemen .
- 2 -
Con en s
Abs ac .........................................................................................................................4
1. In oduc ion...............................................................................................................5
2. Objec i es and Jus i ica ion......................................................................................7
3. Moni o ing Tools......................................................................................................8
3.1 Ac i e Moni o s..................................................................................................8
3.2 Passi e Moni o s.................................................................................................9
3.1 O he s Moni o s Pu poses.................................................................................10
4. The Low In usion Ac i e Hyb id Moni o .............................................................12
4.1. A chi ec u e o he Moni o .............................................................................12
4.2. Moni o Ope a ion...........................................................................................15
4.3. Moni o Implemen a ion..................................................................................18
5. E alua ion and Resul s............................................................................................22
5.1. In usion on Time.............................................................................................22
5.2. In usion on Code.............................................................................................24
5.3. In usion on Powe Consump ion....................................................................25
5.4. F equency o E en Log Gene a ion..............................................................26
6. Discussion...............................................................................................................28
7. Conclusions.............................................................................................................31
8. Fu u e Wo k............................................................................................................32
Re e ences...................................................................................................................33
- 3 -
Abs ac
Se e al sys ems ha e been p oposed o moni o senso ne wo ks—
specially Wi eless Senso Ne wo ks (WSN)—in o de o debug and
analyze hei ope a ion. These sys ems a e based on ha dwa e and/o
so wa e echnology, and can be ac i e o passi e. Each one o hem has
p os and cons, bu none is comple e a all. Ac i e moni o s gene a e a lo
o in usion, passi e ones do no collec all ele an in o ma ion abou
nodes, and mos p oposals a e a away om eal sys ems o a e oo
ha dwa e dependen . This wo k p esen s an ac i e hyb id moni o wi h
low in usion, o be applied on senso ne wo k nodes, which o e comes
mos o hese d awbacks. Moni o is based in a new pu posed open
a chi ec u e o Moni o ing Pla o ms which p o ides lexibili y,
uni e sali y, and eu iliza ion. In usion caused o he senso node has
been e alua ed on h ee aspec s: ime, addi ional code, and powe
consump ion. Low in usion has been achie ed in all h ee issues. We
ha e e alua ed in e aces di e en om adi ional se ial ansmission
p oposed by mos o simila app oaches. I has been p o ed ha he use
o pa allel in e ace o Se ial Pe iphe al In e ace (SPI) allows a highe
equency o e en log gene a ion and low in usion.
Keywo ds: Moni o ing, hyb id moni o , low in usion, senso ne wo ks.
- 4 -
1. In oduc ion
Senso ne wo ks ha e unde gone g ea esea ch and de elopmen in ecen yea s.
Howe e hei massi e deploymen has no been oo g ea because senso s ne wo ks
may expe ience p oblems o e o s in hei ope a ion. Many causes can be iden i ied,
such as in e e ences in he ansmission medium, secu i y a acks (especially in
WSN [1]), ad e se en i onmen al condi ions, mal unc ioning nodes, and o he s. The
node aul s, hei sou ces, and de ec ion app oaches a e di e se, as i is de ailed in
[2]. Al hough du ing de elopmen o implemen a ion o his ype o ne wo k
debugging and ope a ion es ing is usually made, when senso s a e deployed he
condi ions can be e y di e en and usually unan icipa ed e en s a ise.
The a ailabili y o a sui able senso ne wo k ailu e diagnos ic ool is a key issue
in p og essing o eal-wo ld deploymen o WSN. Nowadays he e a e no s anda d
ools o s anda d a chi ec u es in his a ea. Mos o p oposals in moni o ing and
debugging do no conside enough aspec s o senso ne wo ks o be ully use ul, o
a e buil o e y speci ic ne wo k a chi ec u es. The e a e many challenges in
se e al aspec s—a chi ec u al, unc ional, and dynamic— ha ha e o be esea ched
acco ding o [3].
The so called moni o ing sys ems—o simply moni o s—a e used o e alua e he
pe o mance and ope a ion o a senso ne wo k, in con olled condi ions o e en in a
eal en i onmen . Moni o s can ocus on many pe o mance pa ame e s, such as
h oughpu , ji e , esponse ime o eliabili y, and e en o secu i y and in usion
de ec ion in he ne wo k, as desc ibed in [4].
Moni o s usually a e buil based on one o wo possible app oaches. Ac i e
moni o s in ol e addi ional ha dwa e and/o so wa e in senso nodes, in e ac ing
wi h i . In his way, ac i e moni o s usually equi e he modi ica ion o he senso
nodes o be moni o ed. This in e e es wi h i s no mal ope a ion and measu ed
pa ame e s may a y om unmoni o ed node. Howe e , he ob ained da a a e
mo e eliable.
- 5 -

On he o he hand, passi e moni o s ely on he obse a ion o he ex e nal
beha io o he moni o ed sys em wi hou any in e e ence wi h i s no mal ope a ion.
Usually, beha io algo i hms a e used o e alua e he p esence o e o s, undesi able
ope a ion o unexpec ed e en s. No incidence on moni o ed nodes pe o mance is
caused, bu only ex e nally obse able a iables can be measu ed.
Besides he e is ano he app oach o moni o s depending i hey a e based on
ha dwa e o so wa e. A so wa e moni o is implemen ed by means o a speci ic
code, applica ion, o plug-in o he ope a ing sys em o node, which access o he
sys em s a us and epo s ele an in o ma ion. Usually, a so wa e moni o has deep
in o ma ion abou he sys em unc ioning, bu i may in e e e wi h he ope a ion o
he moni o ed sys em.
A ha dwa e moni o consis s on elec onic de ices connec ed o he moni o ed
sys em, which ecollec da a om in e es ing sys em poin s. Ha dwa e moni o s use
o be less in usi e han so wa e moni o s, bu hey implies he use o addi ional
componen s.
Each moni o app oach by i sel canno co e all aspec s o moni o ing asks, as
we will s udy in he nex sec ion. Moni o s can also combine bo h app oaches in
o de o achie e he ad an ages o bo h ypes and ob ain a comple e ision o he
sys em, ying o keep he in e e ence o he minimum. These a e he so called
Hyb id moni o s [5].
In his wo k an ac i e hyb id moni o wi h e y low in usion—based on bo h
ha dwa e and so wa e moni o ing me hods—is p esen ed. This moni o can eco d
he e en s which occu in a node o a senso ne wo k and s o e hem in a non- ola ile
memo y o la e analysis. Mo eo e i is capable o being inco po a ed as a pa o a
comple e moni o ing pla o m ha includes o he acquisi ion possibili ies, such as
passi e moni o s, as desc ibed in [6].
- 6 -
2. Objec i es and Jus i ica ion
The main goal is o build a moni o sys em ha le s o combine he ad an ages o
bo h ypes o moni o s in a single moni o ing pla o m able o moni o he comple e
sys em.
The hyb id moni o is going o wo k in a s anda dized way, by using o s anda d
in e aces and a ailable lib a ies, bu wi h low in usion on senso ne wo k nodes.
This moni o sys em allows o egis e he e en s and beha io o senso ne wo k
nodes, as well as o econs uc he ac i i y o he ne wo k. The gene a ed
in o ma ion can be u he analyzed by adequa e ools.
The a chi ec u e used o hyb id moni o de elopmen wan s o become a s anda d
o moni o ing pla o ms, and allows o moni o p o ides lexibili y, uni e sali y, and
eu iliza ion.
- 7 -
3. Moni o ing Tools
The e a e se e al ools and echniques o moni o ing senso ne wo ks, mos o
hem wi h one app oach, and mainly ocused o WSN. In [3], an o e iew and
compa ison o mos o p oposed ools a e made, no only moni o ing ools bu also
debugging ones. In his sec ion a b ie summa y o some o he mos impo an and
ele an is done.
3.1 Ac i e Moni o s
SNMS (Senso Ne wo k Managemen Sys em) [7] is one o he i s and bes
known moni o ing sys ems. SNMS is a comple e managemen sys em, ocused on
wo king wi h any ype o senso ne wo k. I is buil on TinyOS—an open sou ce
ope a i e sys em designed o low-powe de ices [8]—and allows a e iew o he
s a e o a node and e en sa e in o ma ion locally. Ne e heless, i gene a es
subs an ial in usion—i inc eases a ic and powe consump ion—and is o ien ed
a he o managemen ha moni o ing.
Memen o [9] and Ligh weigh T acing [10] a e bo h examples o ac i e moni o s.
Bo h use sho encoding wi h e en s and in o ma ion o senso node. The i s one
adds i s code p o ocol o a message ha is going o be ansmi ed by node. Memen o
can de ec p oblems in a node by basing on he in o ma ion p o ided by hei
neighbo s in he senso ne wo k; bu o de ec ing new kinds o ailu es i equi es
node ep og amming.
In Ligh weigh T acing [10] he e en s a e sa ed by using a e y ligh coding in
non- ola ile memo y o u he econs uc ion and debugging o node and ne wo k
beha io . Because bo h a e ac i e moni o s, hey gene a e subs an ial in usion in
nodes ope a ion.
Despi e i s name, Passi e Diagnosis o WSN (PAD) [11] is an ac i e moni o
sys em wi h li le in usion. I is based in a p obabilis ic diagnosis app oach—based
on a Belie (o Bayesian) Ne wo k— o in e he oo causes o abno mal WSN
- 8 -
ope a ion. This adds a p obe in each node ha ma ks he packe s wi h ele an da a
wi h e y li le o e head. . Howe e , PAD has o wai a message ansmission o send
in o ma ion and i migh no de e mine when an e o has happened. Besides, as a
as non-sense nodes, such as ou e nodes, do no send sensed da a, hey a e no able
o send any moni o in o ma ion o in e possible abno mal ope a ion.
3.2 Passi e Moni o s
Sympa hy [12] wo ks as a passi e moni o , and can de ec and debug p e-and-pos
deploymen e o s. I ope a es by analyzing he da a a i ing a he sink o a senso
ne wo k, applying me ics, and in e ing whe e in he ne wo k a aul o ailu e can
be p oduced. The implemen a ion o his mechanism depends on he knowledge o
he ne wo k beha io . I also conside s he agg ega ion o a small o e head on he
ne wo k o inc ease i s accu acy. Howe e , i was no de eloped o e en -d i en
applica ions, losing some impo an in o ma ion and educing i s accu acy.
SNIF (Senso Ne wo k Inspec ion F amewo k) [13], Pimo o [14], Li eNe [15],
SNDS (Senso Ne wo k Dis ibu ed Sni e ) [16], NSSN [17], and EPMOS (Ene gy-
e icien Passi e Moni o ing SysTem o WSN) [18] a e examples o passi e
moni o s. Thei app oach consis on deploying a ne wo k o sni e s— oge he wi h
he a ge senso ne wo k—wi h an in e ace o cap u e all ansmissions om nodes.
The main di e ence be ween hem is how he cap u ed da a is p ocessed.
SNIF [13] and Pimo o [14] ansmi — ia a Blue oo h in e ace— he cap u ed da a
o o he de ice in o de o p ocess i . In he i s case he de ice ha ecei es he
packe s— agged wi h a imes amp—wo ks as a sink and analyzes in o ma ion. The
analysis ool has been de eloped by he au ho s.
In Pimo o he de ice ha ecei es he da a—called ga eway—is a compu e ha
ag he packe s wi h a imes amp and o wa ds i , ia TCP/IP, o a cen al se e o
analysis. Pimo o can also wo ks o e mo e han one senso ne wo k simul aneously.
I has a plugin o Wi esha k [19]—a a ic analysis ool—in o de o analysis da a.
Howe e , Pimo o may no be p ac ical o moni o ing WSN ha has oo many nodes
widely dis ibu ed, because i needs mo e in as uc u es.
- 9 -
shows a scheme o moni o ope a ion. A signal—wi h in o ma ion o he e en —is
sen om senso node o moni o node when a ele an e en — ha we wan o
egis e —occu s.
Figu e 4. Hyb id moni o ope a ion scheme.
The ope a ion ime o he moni o has o be simple in o de o educe he in usion
in he ne wo k nodes. Figu e 5 shows he da a low and ope a ion o hyb id moni o .
Figu e 5. Flowcha o hyb id moni o ope a ion.
- 16 -

Speci ic unc ions ha e been de ined o egis e node e en s, and hey a e
included in he node. The e a e wo unc ions added o senso ne wo k node: One can
be loca ed whe e log in o ma ion equi ed, and he o he is ac i a ed by an
in e up ion when he ACK is ecei ed. When senso node eaches a so wa e ap, i
in okes he S a Send Func ion in o de o make ele an da a (con ained in
TXDa a[ ] s uc u e) o be sen o he moni o node. This unc ion p epa es he
da a o be ansmi ed, and hen i sends he i s pa o da a o he moni o node, and
ac i a es a lag ha indica es a ansmission p ocess. Then he unc ion e u ns he
con ol o he main node applica ion.
The da a sen o he moni o may include addi ional in o ma ion wi h he e en
code. The way o ob ain he addi ional da a depends o he senso node.
This possibili y allows inc easing he accu acy o ob ained in o ma ion. Fo
example, a log o an e o e en wi h an addi ional code ha desc ibes he kind o
e o may be ob ained. I is also possible, when dealing wo h e ansmission e en s,
o add o he ap pa o he en i e message, in o de o di e en ia e be ween
messages.
Finally, i is also possible add a code o indica e he eason o he wakes up o he
node. This is a signi ican imp o emen in compa ison wi h o he p oposals ha only
egis e e en s bu canno handle addi ional in o ma ion abou hem.
The moni o node is always wai ing o in o ma ion sen om he senso node.
When da a om senso node is ecei ed, moni o node s o es i wi h he app op ia ed
imes amp. This imes amp has o be gene a ed in case o a new e en log. Then an
ACK message is sen o he senso node. When he senso ne wo k node ecei es he
ACK he nex pa o in o ma ion may be sen i necessa y; o ansmission lag may
be deac i a ed. This mechanism can be conside ed semi-blocke and educes he
in usion o he senso node.
The e en s a e coded in 4 bi s. These codes ha e al eady been de ined ea lie in
[22] and a e shown in Table 1, al hough his de ini ion can be modi ied o inc eased,
in o de o enhance he accu acy.
- 17 -
Moni o node s o es in o ma ion wi h a p ede ined o ma . This o ma includes
he imes amp, he log code, and addi ional in o ma ion i apply; and i mus be easy
o analyze wi h an adequa e applica ion.
Table 1. E en codes de ined by au ho s in [22].
Code Meaning
#de ine Log_Rese 0x00 //Node Rese /Ini ializa ion
#de ine Log_Sense0 0x01 //Read senso 0 ( i s /unique)
#de ine Log_Sense1 0x02 //Read senso 1 (second i i ’s)
#de ine Log_Sense2 0x03 //Read senso 2 ( hi d i i ’s)
#de ine Log_Wakeup 0x04 //Wake up om sleep/s op
#de ine Log_RxDa a 0x05 //Node ecei es da a
#de ine Log_TxDa a 0x06 //Node sends da a
#de ine Log_RxACK 0x07 //Node ecei es ACK
#de ine Log_RRou e 0x08 //Node e ou es da a (i apply)
#de ine Log_Sleep 0x09 //Node goes o sleep mode
#de ine Log_S op 0x0A //Node goes o s op mode
#de ine Log_LowBa 0x0B //Low ba e y indica ion
#de ine Log_SinkRx 0x0C //Sink ecei es da a
#de ine Log_SinkTx 0x0D //Sink sends da a
#de ine Log_SinkE 0x0E //E o in sink
#de ine Log_E o 0x0F //E o in node
4.3. Moni o Implemen a ion
The moni o node has been de eloped on a STM32F051R8T6 ARM Co ex-M0
mic ocon olle (MCU) [25] implemen ed on a STM32F0 Disco e y Boa d [26].
This mic ocon olle is a 32-bi s co e high pe o mance MCU, and i has hese main
ea u es:
ARM Co ex-M0 co e up o 48MHz.
Memo y: 64KB o lash memo y, 8KB o RAM.
Se e al GPIO (Gene al Pu pose Inpu -Ou pu ) po s.
Connec i i y: USART (Uni e sal Synch onous Asynch onous Recei e -
T ansmi e ), SPI, I2C.
Con ol: 3-phase PWM (Pulse-Wid h Modula ion) con ol ime , PWM
ime s, basic ime , compa a o s.
- 18 -
Real Time Clock onboa d o ex e nally supplied.
Digi al-Analog and Analog-Digi al con e e s.
The moni o boa d has been connec ed o a SD ca d h ough i s SPI in e ace.
Figu e 6 shows he implemen a ion o he hyb id moni o . The moni o node—a le
side—is connec ed o a senso node—a igh side— ha is used o es ing. I has
a ached he SD ca d used o s o e da a. In case o Figu e 6 he SPI in e ace is also
used o connec he moni o node o he senso node.
Figu e 6. Pic u e o a senso node connec ed o he moni o node.
In o de o e alua e he mos app op ia e communica ions in e ace be ween
senso node and moni o node, h ee di e en in e aces implemen a ion—usually
ound in mos o mic ocon olle s—ha e been used: SPI, USART, and inally
pa allel ansmission, using 16 GPIO po s. Pa allel and SPI da a wid h is 16 bi s, and
USART da a wid h is 8 bi s. Besides, h ee ansmission speeds we e e alua ed o
SPI in e ace. To compa e he pe o mance o each in e ace, ou sizes o
ansmi ed da a (common sizes o 16, 32, 64, and 128 bi s pe ap) ha e been
assumed. . The bigge he da a being ansmi ed, mo e de ailed he in o ma ion
p o ided in each ap, bu highe he in e e ence.
A lib a y should be added o he code implemen a ion on senso node wi h he
ins uc ion se s o ini ializa ion and da a ansmission. As a as he ARM
- 19 -
a chi ec u e may be he mos used in senso node implemen a ion, and in o de o
p o ide a s anda dized le el CMSIS (Co ex Mic ocon olle So wa e In e ace
S anda d) has been used o c ea e his lib a y. CMSIS is a endo -independen
ha dwa e abs ac ion laye and can be used on many mic ocon olle s [27]. This
lib a y can be adap ed easily o o he s a chi ec u es wi h he co esponden
ins uc ions. As a as he messages sen h ough he in e ace be ween senso node
and moni o node (SPI, USART o Pa allel) a e he same ha p e iously ci ed ones,
moni o node needs no modi ica ion.
Fo e alua ion o he moni o node a senso node based in an ARM Co ex-M0 a
48MHz was used. E e y ime his senso node wakes up om sleep mode, i akes a
measu emen om a empe a u e senso , ansmi s he measu emen ia wi eless, and
u ns back o sleep mode o 60 seconds again.
T aps ha e been inse ed in he p og am o he senso node o egis e h ee kinds
o e en s: wake up, ansmission, and sleep. These codes a e sen o he moni o
node, which e lec s hem in o ASCII ex sepa a ed by commas, and s o es hese
da a and i s imes amp in a SD ca d. Figu e 7 shows pa o he in o ma ion
gene a ed by he hyb id moni o and eco e ed om his SD ca d.
Figu e 7. In o ma ion eco e ed om he SD ca d a ached o he moni o node.
In case o Figu e 7 in o ma ion includes da e, ime, mic oseconds, e en code, and
addi ional da a. E en codes 04, 06 and 09 (Table 1) ha e been used o wake up,
- 20 -
ansmission and sleep e en s, espec i ely. Addi ional alues o wake up and
ansmission e en s—also sen as addi ional in o ma ion—indica e he i e a ion
numbe . All his da a can be deli e ed o he isualiza ion and con ol subsys em
(Figu e 1) by using he in o ma ion laye se ices.
- 21 -

5. E alua ion and Resul s
In o de o e alua e he cha ac e is ics o he p oposed moni o , in usion analysis
is equi ed. Th ee a e he main in usion aspec s ha mus be conside ed. Time
in usion deals wi h he inc emen o execu ion ime on senso node caused by he
moni o . Taking in o accoun ha senso nodes use o be limi ed in RAM memo y,
Code in usion e alua es how many by es ha e o been added o sou ce p og am o
his senso node. Finally, Powe in usion e alua es he amoun o ene gy ha his
moni o ing echnique equi es om he senso node, and hus educing i s ba e y
li e. Se e al expe imen s ha e been pe o med, conside ing he p e iously ci ed
in e aces and da a sizes, o measu e he in usion o he ac i e hyb id moni o .
Ob ained esul s a e shown in his sec ion.
5.1. In usion on Time
Time in usion was de e mined by measu ing he amoun o ime needed o ul ill
one housand wake-up i e a ions in he senso node wi hou aps ( e e ence ime),
and hen measu ing when aps we e added. Th ee in e ace implemen a ions and
ou da a sizes we e combined o p o ide he able 2. This able shows he di e ence
be ween he ob ained ime and he e e ence ime, and hus he ime in usion.
Table 2. In usion on ime o each log e en wi h di e en in e aces and da a sizes
(mic oseconds).
Da a Size Pa allel
(16bi s) SPI 18mbps SPI
4.5mbps
SPI
2.25mbps
USART
115.2kHz
16bi s 3.80 3.10 3.10 3.20 6.20
32bi s 7.30 6.10 6.20 6.20 13.60
64bi s 16.50 13.00 13.10 13.10 28.80
128bi s 31.90 28.10 28.10 28.20 60.60
As shown, he in usion ime is simila o di e en in e aces wi h he same da a
size, due o he implemen a ion o he moni o code in he senso node: As a as da a
ansmission h ough he in e ace is pe o med by speci ic ha dwa e (SPI con olle ,
USART, o he s), ap ou ines only w i e he ou going da a in o he in e ace bu e s
- 22 -
and he senso node p og am may con inue. These ha dwa e con olle s ake ca e o
ansmission in pa allel o his p og am execu ion.
Pa allel in e ace has no dedica ed ha dwa e con olle . Tha is he eason ha he
ime is li le bigge han he o he s. An addi ional line is used o gene a e he
ansmission in e up ion in he moni o node. As expec ed, in usion ime inc eases
o la ge da a sizes.
Howe e , alues in Table 2 canno be used as a e e ence o know he maximum
equency o he e en log gene a ion, because i depends on he pe iphe al using
ime (communica ion ha dwa e) and he moni o node ha dwa e. Besides, he imes
o Table 2 a e a sum o non-con inue alues. Time limi a ions on SPI con olle ,
USART and pa allel po s mus also be conside ed. In case o USART he in usion
ime is abou double because i only sends 8 bi s pe ansmission and i has o
manage wice o in e up ions.
The p ocessing ime in he moni o node was measu ed oo. Each e en log has o
be p ocessed in one o mo e da a ecep ion in e up ion, acco ding o da a size
ansmi ed om senso node. The p ocessing ime is conside ed om he a i al o
he i s da a o he ecei ing o he las piece o in o ma ion. The imes amp
gene a ion akes 15.8 mic oseconds, and is done when he i s piece o da a a i es.
Table 3 shows he ob ained alues.
Table 3. P ocessing ime in moni o node o each e en log (mic oseconds).
Da a Size Pa allel
(16bi s) SPI 18mbps SPI
4.5mbps
SPI
2.25mbps
USART
115.2kHz
16bi s 17.80 17.40 17.40 17.40 88.40
32bi s 22.90 23.40 25.40 27.80 262.00
64bi s 32.90 34.40 40.10 48.20 612.00
128bi s 52.80 56.20 69.80 88.00 1300.00
As shown in Table 3, when dealing wi h 16 bi s e en s he ob ained imes a e
simila o e e y in e ace, as a as p ocess ime depends mainly on imes amp
gene a ion. Fo la ge da a size he p ocessing ime is a bi less when using pa allel
in e ace ins ead o SPI, especially when he SPI speed is low.
- 23 -
In he case o he USART in e ace, he p ocessing ime is e y la ge , because i s
low ansmission speed and he ac ha i only can send 8 bi s a a ime, whe eas
o he conside ed op ions can send 16 bi s a a ime. As expec ed, he p ocessing ime
inc eases o la ge da a sizes.
Should be no ed, alues o Tables 2 and 3 may change in some cases. A moni o
node based on a as e mic ocon olle will educe he imes amp gene a ion ime,
hence will educe he p ocessing ime and inc ease he pe o mance. Besides, he
ime in usion and p ocessing ime could change depending o senso ne wo k node
cha ac e is ics, such as co e equency, a ailabili y o bu e s, a chi ec u e, in e ace
o moni o , and o he s.
USART in e ace has no been conside ed as an op ion o be used in ou moni o
because o i s low speed and high p ocessing ime—compa ed wi h he o he s
in e aces—, which esul s in low pe o mance. Thus i is no conside ed in
u he esul s.
5.2. In usion on Code
As a as memo y esou ces a e limi ed o senso nodes, he e alua ion o he
in usion on p og am code is ele an . Code used in he senso node was gene a ed
by Keil MDK (Mic ocon olle De elopmen Ki ) e sion 5, a comp ehensi e
so wa e de elopmen en i onmen o Co ex-M p ocesso based mic ocon olle s
ha includes IDE (In eg a ed De elopmen En i onmen ), Compile , and
Debugge [28].
Size di e ence in by es has been conside ed be ween he pu e applica ion code
and he ap-modi ied code, in bina y compiled p og am. Main di e ences be ween
bo h codes consis on he addi ion o he po ini ializa ion sub ou ine and se e al
da a ansmission ins uc ions.
Table 4 shows he in usion in by es on p og am code in senso node. As expec ed,
in usion is no ela ed o he communica ion speed, hus his pa ame e has no been
conside ed. Since ansmission code is eused o all he aps, a new e en may be
moni o ed by jus adding o he code o he senso node a ap call o 8 by es.
- 24 -
Table 4. In usion on Code: Ini ializa ion and one e en log (By es).
In e ace 16bi s 32bi s 64bi s 128bi s
Pa allel 228 252 268 276
SPI 428 452 468 476
Code in usion (SIn usion) may be p edic ed by means o he Equa ion (1), whe e Sini
is he ini ializa ion alue appea ing in Table 4, and n is he numbe o ap calls
included in he applica ion code. As expec ed, o a small numbe o aps moni o ed,
he code in usion is mos ly de e mined by he ini ializa ion code.
SIn usion=Sini +8×
(
n−1
)
(1)
5.3. In usion on Powe Consump ion
Taking in o accoun ha senso nodes usually a e powe ed wi h ba e ies o
ha es ing echniques, powe consump ion is a key aspec in senso ne wo k nodes.
Due he wo king ime is e y low—some milliseconds as maximum—powe
consump ion was de e mined by modi ying he applica ion p og am o he senso
node, a oiding sleep mode and keeping used pe iphe als enabled. Abou 7000
samples o senso node elec ical powe we e aken—18 samples pe second—and
a e aged hem. Measu emen s we e aken wi h an Agilen 34405A Mul ime e . This
mul ime e has a 5.5 digi esolu ion, wi h an accu acy o ±(0.05% o eading +
0.015% o ange) o ou measu ing condi ions [29].
The ob ained esul s showed ha he ins an aneous powe consump ion was he
same o all he e alua ed communica ion speeds. Da a size was also ound o be
i ele an in powe consump ion.
The senso node elec ical cu en equi ed in clea ope a ion, wi hou moni o ing,
was ound o be 53.4 mA. The o al elec ical cu en equi ed when moni o ing was
pe o med h ough each in e ace and he di e ences wi h no mal ope a ion a e
shown in Table 5.
- 25 -
8. Fu u e Wo k
Adding he moni o o a passi e moni o ing sys em—e.g. a sni e s ne wo k like
men ioned p oposed abo e—can esul in a comple e moni o ing sys em ha can
obse e he whole ne wo k beha io . Hence, a u u e wo k ha in eg a es he ac i e
moni o o a passi e moni o ing ne wo k (sni e s) should be de eloped.
Fo a as e analysis o log da a cap u ed by he moni o , a adio in e ace can be
added o he moni o node o ans e he da a om memo y ca d o a hos . This hos
can be he same de ice ha wo ks as sni e .
The o ma o s o ed da a—in he in o ma ion laye —can be s anda dized, so i
can comply wi h he main goal o pu posed a chi ec u e in which he hyb id moni o
is based. A p ac ical and easy way o do i is o use an exis ing s anda d o ma o
adap i o his pu pose.
A model o de e mina e in usion is necessa y. The conside ed in usion aspec s o
his model mus be a leas he h ee co e ed in his wo k. The model should de ine
le els and h esholds o in usion, as well as o mulas o calcula e hese ones.
A P og am o collec ed da a analysis mus be de eloped. This p og am can be a
plugin o be used wi h any known ne wo k a ic analyze —e.g. Wi esha k—o a
new applica ion de eloped o his goal.
- 32 -

Re e ences
1. Pa el, M.; Agga wal, A. Secu i y A acks in Wi eless Senso Ne wo ks: A
Su ey. In P oceedings o he IEEE In e na ional Con e ence on In elligen
Sys ems and Signal P ocessing (ISSP), Anand, India, 1–2 Ma ch 2013; pp. 329–
333.
2. Mahapa o, A.; Khila , P.M. Faul Diagnosis in Wi eless Senso Ne wo ks: A
Su ey. Communica ions Su e ys & Tu o ials, 2013, Vol. 15, No. 4, pp 2000–
2026.
3. Rod igues, A.; Camilo, T.; Sil a, J.S.; Boa ida, F. Diagnos ic Tools o Wi eless
Senso Ne wo ks: A Compa a i e Su ey. Jou nal o Ne wo k and Sys ems
Managemen , 2012, Vol. 21 Iss. 3, pp. 408–452.
4. Rachedi, A. Moni o ing Mechanism o Wi eless Senso Ne wo ks: Challenges
and Solu ions. In Wi eless Senso Ne wo ks, El Ema y, I,; Ramak ishan, S., Eds;
CRC P ess: Boca Ra on, FL, USA, 2013; pp. 595–621.
5. Schoo s, A.; O'Ha e, G.M.P.; Ruzzelli, A.G. Debugging Low-Powe and Lossy
Wi eless Ne wo ks: A Su ey. IEEE Communica ions Su eys & Tu o ials,
2012, Vol. 14, No. 2, pp. 311–321.
6. Pique as, I.; Campelo, J. C.; O s, R.; Se ano, J. J. Hyb id moni o ing o
wi eless senso ne wo ks. In P oceedings o he IEEE In e na ional Con e ence
on Wi eless In o ma ion Technology and Sys ems (ICWITS), Maui, Hl, USA,
11–16 No embe 2012; pp. 1–4.
7. Tolle, G.; Culle , D. Design o an Applica ion-Coope a i e Managemen Sys em
o Wi eless Senso Ne wo ks. In P oceedings o 2nd Eu opean Wo kshop on
Wi eless Senso Ne wo ks (EWSN), Is anbul, Tu key, 31 Jan. –2 Feb ua y 2005;
pp. 121–132.
8. FAQ – TinyOS Wiki. A ailable online: h p:// inyos.s an o d.edu/ inyos-
wiki/index.php/FAQ (accessed on 15 Decembe 2014).
9. Ros , S.; Balak ishnan, H. Memen o: A Heal h Moni o ing Sys em o Wi eless
Senso Ne wo ks. In P oceedings o 3 d Annual IEEE Communica ions Socie y
on Senso and Ad Hoc Communica ions and Ne wo ks (SECON), Res on, VA,
USA, 25–28 Sep embe 2006; Vol. 2, pp. 575–584.
10. Sunda am, V.; Eugs e , P.; Zhang, X. Ligh weigh acing o wi eless senso
ne wo ks debugging. In P oceedings o he 4 h In e na ional Wo kshop on
Middlewa e Tools, Se ices and Run-Time Suppo o Senso Ne wo ks
(MidSens 09), U bana Champaign, Il, USA, 30 No . – 4 Decembe 2009; pp.
13–18.
- 33 -
11. Liu, Y.; Liu, K.; Li, M. Passi e Diagnos ics o Wi eless Senso Ne wo ks.
IEEE/ACM T ansac ions on Ne wo king, 2010, Vol. 18, No. 4; pp. 1132–1144.
12. Ramana han, N.; Chang, K.; Kapu , R.; Gi od, L.; Kohle , E.; Es in, D.
Sympa hy o he Senso Ne wo k Debugge . In P oceedings o 3 d ACM
Con e ence on Embedded Ne wo ked Senso Sys ems (ACM SenSys 2005), San
Diego, CA, USA, 2–4 No embe 2005; pp. 255–267.
13. Ringwald, M.; Röme , K.; Viale i, A. SNIF: Senso Ne wo k Inspec ion
F amewo k. Depa men o Compu e Science, ETH Zu ich, Technical Repo
No. 535; 2006.
14. Awad, A.; Nebel, R.; Ge man, R.; D essle , F. On he Need o Passi e
Moni o ing in Senso Ne wo ks. In P oceedings o 2008 11 h EUROMICRO -
Con e ence on Digi al Sys em Design A chi ec u es, Me hods and Tools (DSD);
Pa ma, I aly, 3–5 Sep embe 2008; pp. 693–699.
15. Chen, B.; Pe e son, G.; Mainland, G.; Welsh, M. Li eNe : Using Passi e
Moni o ing o Recons uc Senso Ne wo k Dynamics. In P oceedings o IEEE
Con e ence on Dis ibu ed Compu ing in Senso Sys ems (DCOSS), San o ini
Island, G eece, 11–14 June 2008; pp. 79–98.
16. Kuang, X,; Shen, J. SNDS: A Dis ibu ed Moni o ing and P o ocol Analysis
Sys em o Wi eless Senso Ne wo k. In P oceedings o 2010 Second
In e na ional Con e ence on Ne wo ks Secu i y, Wi eless Communica ion and
T us ed Compu ing (NSWCTC), Wuhan, China, 24–25 Ap il 2010; pp. 422–
425.
17. Zhonghua, Z.; Huang u, W.; Linmin, S. NSSN: A Ne wo k Moni o ing and
Packe Sni ing Tool o Wi eless Senso Ne wo ks. In P oceedings o 8 h
In e na ional Wi eless Communica ions and Mobile Compu ing Con e ence
(IWCMC), Limassol, Cyp us, 27–31 Augus 2012; pp. 537–542.
18. Ga cia, F.; And ade, R.; Oli ei a, C.; de Souza J.N. EPMOS : And Ene gy-
E icien Passi e Moni o ing Sys em o Wi eless Senso Ne wo ks. Senso s,
2014, 14, pp. 10804–10828.
19. Wi esha k. A ailable online: h ps://www.wi esha k.o g/ (accessed on 20
Janua y 2015)
20. Somme , P.; Kusy, B. Mine a: Dis ibu ed T acing and Debugging in Wi eless
Senso Ne wo ks. In P oceedings o he 11 h ACM Con e ence on Embedded
Ne wo ked Senso Sys ems (SenSys), Rome, I aly, 11–14 No embe 2013.
21. Hossain, M.S.; Lee, W.S.; Raghuna han, V. Spi-Noope : A Ha dwa e-So wa e
App oach o T anspa en Ne wo k Moni o ing in Wi eless Senso Ne wo ks. In
P oceedings o he 8 h IEEE/ACM/IFIP In e na ional Con e ence on
Ha dwa e/So wa e Codesign and Sys ems Syn hesis (CODES+ISSS), Tampe e,
Finland, 7–12 Oc obe 2012; pp. 53–62.
- 34 -
22. Na ia, M.; Campelo, J.C.; Bonas e, A.; Se ano, J.J. Low In usion Ac i e
Hyb id Moni o o Nodes o Senso Ne wo ks. In P oceedings o Wo kshop on
Inno a ion on In o ma ion and Communica ion Technologies (ITACA-WIICT),
Valencia, Spain, 4 July 2014; pp. 111–120.
23. Capella, J.V.; Campelo, J.C.; Bonas e, A.; O s, R. P oposal o a Re e ence
Model o WSN’s Moni o ing Pla o ms. Jou nal o Ne wo k and Compu e
Applica ions. Recei ed (unde e ision.)
24. ISO/IEC. In o ma ion Technology – Open Sys ems In e connec ion – Basic
Re e ence Model. In e na ional S anda d ISO/IEC 7498-1, second edi ion, 1994.
25. STM32F051R8 ARM Co ex-M0 MCU. A ailable online:
h p://www.s .com/web/ca alog/mmc/ (accessed on 2 Decembe 2014).
26. STM32F0DISCOVERY Disco e y Ki o STM32F051. A ailable online:
h p://www.s .com/web/ca alog/ ools/FM116/SC959/SS1532/PF253215
(accessed on 20 Janua y 2015)
27. CMSIS - Co ex Mic ocon olle So wa e In e ace S anda d. A ailable online:
h p://www.a m.com/p oduc s/p ocesso s/co ex-m/co ex-mic ocon olle -
so wa e-in e ace-s anda d.php (accessed on 8 Decembe 2014).
28. Keil MDK-ARM Ve sion 5. A ailable online: h p://www2.keil.com/mdk5/
(accessed on 10 Janua y 2015).
29. 34405A Digi al Mul ime e , 5½ digi | Keysigh (Agilen ). A ailable online:
h p://www.keysigh .com/en/pd-686884-pn-34405A/ (accessed on 10 Janua y
2015).
30. Gha ghan, S.; No din, R.; Ismail, M. Ene gy-E icien ZigBee-Based Wi eless
Senso Ne wo k o T ack Bicycle Pe o mance Moni o ing. Senso s, 2014, 14,
pp. 15573–15592.
31. Molina-Ga cia, A.; Fuen es, J.A.; Gómez-Láza o, E.; Bonas e, A.; Campelo,
J.C.; Se ano, J.J. De elopmen and Assessmen o a Wi eless Senso and
Ac ua o Ne wo k o Hea ing and Cooling Loads. IEEE T ansac ions on Sma
G id, 2012, Vol. 3 No. 3, pp. 1992-1202.
32. Lee, D.S.; Liu, Y.H.; Ling, C.R. A Wi eless Senso Enabled by Wi eless Powe .
Senso s, 2012, 12, pp. 16116–16143.
- 35 -