senso s
A icle
LoRa-Based T a ic Flow De ec ion o Sma -Road
Da id Asiain * and Diego An olín


Ci a ion: Asiain, D.; An olín, D.
LoRa-Based T a ic Flow De ec ion o
Sma -Road. Senso s 2021,21, 338.
h ps://doi.o g/10.3390/s21020338
Recei ed: 10 No embe 2020
Accep ed: 3 Janua y 2021
Published: 6 Janua y 2021
Publishe ’s No e: MDPI s ays neu-
al wi h ega d o ju isdic ional clai-
ms in published maps and ins i u io-
nal a ilia ions.
Copy igh : © 2021 by he au ho s. Li-
censee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and con-
di ions o he C ea i e Commons A -
ibu ion (CC BY) license (h ps://
c ea i ecommons.o g/licenses/by/
4.0/).
Escuela Uni e si a ia Poli écnica de la Almunia, C/Mayo 5, La Almunia de Doña Godina,
50100 Za agoza, Spain; dan olin@uniza .es
*Co espondence: dasiain@uniza .es; Tel.: +34-976-600-813 (ex . 327494)
Abs ac :
This pape p esen s a wi eless a ic low de ec ion sys em, mainly ocused on condi ions
in which he a ic low is slow o s opped, which inc eases he isk o highway acciden s. To achie e
his goal, a Low Powe Wide A ea Ne wo k (LPWAN) based on LoRa called Sho LoRa has been
de eloped. This LoRa sub-ne wo k complies wi h he Eu opean Telecommunica ions S anda ds
Ins i u e (ETSI) ha monized s anda d o i s compa ibili y in Eu ope coun ies. In addi ion, he
de elopmen o he de ices has allowed hem o also wo k on a LoRaWAN ne wo k. The in oduced
de elopmen has been compa ed o a e e ence sys em moun ed wi h lase ba ie s ha p o ided a
high accu a e compa ison. Field es s o he sys em ha e been ca ied ou and he da a ob ained in
he measu emen has been analyzed wi h wo di e en me hods, and bo h o hem we e alid o he
applica ion. The esul s can de e mine ehicle speed wi h adequa e p ecision a low speeds. The
a enua ing beha io o he communica ion signal is also analyzed h ough he Radio Signal S eng h
Indica o (RSSI). The ela ionship be ween ehicle speed, ga e dis ances and RSSI a enua ion has
been s udied. The sys em is p o en o ha e e icien esul s in de ec ing a ic low unde he
condi ions o which i has been de eloped.
Keywo ds: LoRa; LoRaWAN; Sma -Road; a ic low; highway signaling
1. In oduc ion
In ecen yea s, con inuous ad ances ha e been made in adio equency communi-
ca ions ha allow low-powe communica ions o co e long dis ances. This has caused
he de elopmen o he In e ne o Things (IoT), new low cos and ene gy e icien de-
ices, using di e en new communica ion echnologies. LoRa is a new p omising (Low
Powe Wide A ea Ne wo k) LPWAN, ha pe mi s communica ion in dis ances up o a ew
kilome e s o no equi e he complex deploymen and main enance o mul i-hop echnolo-
gies [
1
]. LoRa-based ne wo ks ha e been se up and deployed in di e en applica ions
om indoo [2] and u ban [3] en i onmen s, o ma i ime [4] and moun ain scena ios [5].
Nowadays, a g owing esea ch has ocused on IoT based-applica ions, such as sma
ci ies and sma oads. One applica ion ha is cu en ly being esea ched cu en ly is a ic
low de ec ion.
In his las case, se e al senso and de ec ion echniques wi h speci ic ad an ages and
disad an ages a e used. A equen app oach o ehicle de ec ion and classi ica ion is
came a-based sys ems, which achie e a high classi ica ion success a e. This echnique
o en equi es a numbe o came as o analyze he scena io om di e en angles and
pe spec i es. In con as , Hsieh e al. in [
6
] p esen an enhanced isual sys em which is able
o a ange ehicles in o dis inc ehicle classes using a single came a. Rega dless o he
lowe numbe o came as, he use o hese de ices equi es an addi ional e o in e ms o
ins alla ion, main enance and p i acy- ela ed p oblems in eal-wo ld scena ios. Mo eo e ,
he success a e o hese sys ems signi ican ly dec eases when he wea he condi ions
impede he isibili y. In o de o imp o e he success ul de ec ion a e mixed came a
sys ems ha e been de eloped, he e a e app oaches using lase scanne s [
7
], acous ic
senso s [
8
], magne ome e s [
9
] o accele ome e s [
10
]. The i s h ee app oaches p esen
Senso s 2021,21, 338. h ps://doi.o g/10.3390/s21020338 h ps://www.mdpi.com/jou nal/senso s
Senso s 2021,21, 338 2 o 24
simila d awbacks like came a-based de ec ion sys ems; he las one equi es cons uc ion
wo ks (pa emen cu , e c).
O he echniques ha e p o ed o be accu a e in de ec ing di e en elemen s in an
en i onmen , such as Radio Tomog aphic Imaging (RTI) [
11
], which uses a 2.4 GHz WiFi
signal o loca e people, simila ly as i is p oposed in [
1
] o an IoT applica ion. The capaci y
o hese sys ems g an s an abili y o moni o human mo ion by e e ing o he mo emen
o hei limbs [
12
]. O he wo ks o ien ed o moni o ing he a ic low p opose he use o
di e en WiFi and ZigBee signals o de ec a ic low, classi y he di e en ehicles, and
e en moni o hei speed [10,13–16].
The eliabili y o LoRa is e alua ed in [
17
] o di e en se up condi ions. In ou doo
expe imen s, i shows a co ela ion be ween empe a u es, humidi y, packe ecep ion
a e and he s eng h o he signal ecei ed. These en i onmen al condi ions ha e been
conside ed in his p ojec . In [
18
] a loca ion me hod based on Radio Signal S ange Indica o
(RSSI) using LoRa is e alua ed.
The e a e sma oad signs al eady se and unning nowadays, and hey will be seen
equen ly in a nea u u e. The p esen wo k is ocused on hese ype o signals, in which
a communica ion o he signals and a synch onism be ween hem is equi ed, so ha hey
migh ca y ou he pe inen ac ions depending on he a ic si ua ion. In his a ea, LoRa
and LoRaWAN a e app op ia e communica ion echnologies due o hei ela i ely low
consump ion, low cos and long communica ion dis ances.
This p esen documen displays an unexpensi e low o s opped a ic de ec ion
sys em in mo o ways based on LoRa. This is wi hin he amewo k o a p ojec o imp o e
he signaling o a highway. The main p ojec is ocused on a signaling sys em o a dense
og si ua ion ha appea s seasonally and ha has he oad sec ion closed o long pe iods
o he yea . The sys em p oposed in his a icle is an added unc ionali y ha —wi h ze o
ma e ial cos s—allows he de ec ion o s opped ehicles o ehicles wi h a educed speed
ha p oduce a high isk o acciden s wi h dense og on he oad.
The LoRa based ne wo k’s speci ic p oblem on he oad i was de eloped o is he
appea ance o dense og on he oad, which causes a dec ease on he d i e s’ isibili y,
he e o e being likely o spawn acciden s. In hese haza dous en i onmen al condi ions,
o he ca de ec ion echnologies (came as, ada , e c.) ha e shown di icul ies in de ec ing
he p esence o dense and non-mo ing c owds o ehicles on he oad obs uc ing a ic.
The pape is s uc u ed as ollows: Sec ion 2p o ides a b ie backg ound o con ex u-
alize his wo k, he LoRa Ne wo k speci ically p oposed and he expe imen al deploymen
and cases o s udy. Sec ion 3comp ises he expe imen al esul s and di e en da a analysis
me hods. Sec ion 4includes he conclusions and u u e wo k.
2. Expe imen al Deploymen (Ma e ials and Me hods)
This sec ion desc ibes he ha dwa e and so wa e de elopmen , explaining he di e -
en solu ions es ed: ha dwa e and anscei e pla o m, ne wo k and da a ansmission
se up and ca de ec o algo i hm.
2.1. Ha dwa e Desc ip ion
The main objec i e o his job i is allowing o de ec a slowed o s opped a ic on
highways. In his way, he Radio F equency (RF) ansmission echnology should ha e an
ex ensi e co e age, as well as being obus and low powe ed. Se e al RF echnologies ha e
been e alua ed, such us, ZigBee, BLE (Blue oo h Low Ene gy), NB IoT, Sig ox, LoRa o
LTE. Figu e 1shows he ela ionship be ween dis ance anges s. da a ansmission a io.
On he o he hand, he ne wo k opology is impo an o ensu e a obus and simple
communica ion p o ocol. The package in eg i y in his applica ion i is e y impo an o
de ec any s opped ehicle on he highway as soon as possible and gi e he co esponding
ad ice. In addi ion, he oad sec ion be co e ed could ha e a ew hund ed o me e s. To
ensu e he co e age, he ansmission ange should be g ea e han a kilome e . In his
Senso s 2021,21, 338 3 o 24
way, echnologies as ZigBee and BLE a e no app op ia e since hei ansmission ange is
a ound a ew hund ed o me e s.
Senso s 2021, 21, x FOR PEER REVIEW 3 o 26
Figu e 1. Compa ison be ween Low Powe Wide A ea Ne wo k (LPWA) ne wo ks and o he con-
nec i i y echnologies.
On he o he hand, he ne wo k opology is impo an o ensu e a obus and simple
communica ion p o ocol. The package in eg i y in his applica ion i is e y impo an o
de ec any s opped ehicle on he highway as soon as possible and gi e he co esponding
ad ice. In addi ion, he oad sec ion be co e ed could ha e a ew hund ed o me e s. To
ensu e he co e age, he ansmission ange should be g ea e han a kilome e . In his
way, echnologies as ZigBee and BLE a e no app op ia e since hei ansmission ange
is a ound a ew hund ed o me e s.
In [19] he au ho s p esen a e iew o long- ange echnologies o IoT and he Table
1 shown a esume o he main a ibu es o each echnology.
Table 1. A ibu es o long- ange echnologies o he In e ne o Things (IoT).
A ibu e LTE-M NB-IoT Sig ox LoRa
F equency Band 700‒900 MHz 700‒900 MHz 868, 902 MHz Sub-GHz ISM
Da a Ra e 375 kbps 25‒65 kbps 0.1 kbps 0.3‒37.5 kbps
Bandwid h 1.08 MHz 200 kHz 100 Hz <500 kHz
Range <15 km <35 km Ru al: 30‒50 km
U ban: 3‒10 km
Ru al: 10‒15 km
U ban: 3‒5 km
LTE is adequa e o a high bandwid h and high da a ansmission a io ou o low
powe equi emen s. NB IoT, Sig ox and LoRa a e LPWAN whe e hese ne wo ks ha e a
long- ange da a ansmission also p esen s a low-powe consump ion. These p o ocols
a e sui able o he moni o ing and signaling o highways.
NB-IoT can be deployed in di e en modes o ope a ion. I deployed in gua d band
ope a ion mode, i wo ks in a equency band simila o LTE, as indica ed in Table 1. NB-
IoT has a highe powe consump ion han LoRa.
Sig ox wo ks well o simple de ices and low da a a e, as in his case. This p o ocol
is no deployed e e ywhe e, making i di icul o use. In addi ion, communica ion is be -
e di ec ed om he end poin o he base s a ion, and as i will be seen la e in he wo k
de elopmen , he applica ion also equi es communica ion be ween ne wo k nodes and
he possibili y o bo h uplink and downlink communica ion om he se e o he ga e-
way. Addi ionally, i has a small da a a e and a sho bandwid h.
LoRa allows you o con igu e and manage you own ne wo k, being a good op ion
when bi-di ec ionali y is equi ed because i has a symme ical uplink and downlink con-
Figu e 1.
Compa ison be ween Low Powe Wide A ea Ne wo k (LPWA) ne wo ks and o he
connec i i y echnologies.
In [
19
] he au ho s p esen a e iew o long- ange echnologies o IoT and he Table 1
shown a esume o he main a ibu es o each echnology.
Table 1. A ibu es o long- ange echnologies o he In e ne o Things (IoT).
A ibu e LTE-M NB-IoT Sig ox LoRa
F equency Band 700–900 MHz 700–900 MHz 868, 902 MHz Sub-GHz ISM
Da a Ra e 375 kbps 25–65 kbps 0.1 kbps 0.3–37.5 kbps
Bandwid h 1.08 MHz 200 kHz 100 Hz <500 kHz
Range <15 km <35 km Ru al: 30–50 km
U ban: 3–10 km
Ru al: 10–15 km
U ban: 3–5 km
LTE is adequa e o a high bandwid h and high da a ansmission a io ou o low
powe equi emen s. NB IoT, Sig ox and LoRa a e LPWAN whe e hese ne wo ks ha e a
long- ange da a ansmission also p esen s a low-powe consump ion. These p o ocols a e
sui able o he moni o ing and signaling o highways.
NB-IoT can be deployed in di e en modes o ope a ion. I deployed in gua d band
ope a ion mode, i wo ks in a equency band simila o LTE, as indica ed in Table 1. NB-IoT
has a highe powe consump ion han LoRa.
Sig ox wo ks well o simple de ices and low da a a e, as in his case. This p o ocol is
no deployed e e ywhe e, making i di icul o use. In addi ion, communica ion is be e
di ec ed om he end poin o he base s a ion, and as i will be seen la e in he wo k
de elopmen , he applica ion also equi es communica ion be ween ne wo k nodes and
he possibili y o bo h uplink and downlink communica ion om he se e o he ga eway.
Addi ionally, i has a small da a a e and a sho bandwid h.
LoRa allows you o con igu e and manage you own ne wo k, being a good op ion
when bi-di ec ionali y is equi ed because i has a symme ical uplink and downlink
connec ion. I also enables communica ion be ween nodes, al hough LoRaWAN does no
con empla e i because LoRaWAN ne wo k opology is s a . Lo a allows an in e media e
da a ans e a e in compa ison wi h he p e ious wo. These cha ac e is ics make i
app op ia e o he applica ion, so LoRa has been selec ed o he de elopmen o his wo k.
Senso s 2021,21, 338 4 o 24
LoRa de ice a chi ec u e consis s o he high-pe o mance mic ocon olle , i is a
low-powe ARM
®
Co ex
®
-M0 + based ATSAMD21G18 wi h 256 kB o lash, 32 KB o
SRAM and ope a ing equency up o 48 MHz. I implemen s he LoRaWAN s ack.
The long ange anscei e is he module RFM95 wi h a equency o 868 MHz. I s main
ea u es a e i s high sensi i i y below
−
148 dBm combined wi h he +20 dBm in eg a ed
powe ampli ie . I s modula ion modes a e FSK, GFSK, MSK, GMSK, LoRaTM and OOK.
The connec ion in e ace be ween he mic ocon olle and he anscei e is SPI (Se ial
Pe iphe al In e ace). Finally, he whole sys em is inished wi h an omnidi ec ional an enna
o 1dBi gain. In addi ion, he module inco po a es a iaxial MEMS lis3dh accele ome e
and a gene al-pu pose inpu -ou pu connec o . Las ly, he Powe Supply consis ing o a
TPS7A05 low quiescen cu en low d op egula o . Figu e 2shows he block diag am o
he module.
Senso s 2021, 21, x FOR PEER REVIEW 4 o 26
nec ion. I also enables communica ion be ween nodes, al hough LoRaWAN does no con-
empla e i because LoRaWAN ne wo k opology is s a . Lo a allows an in e media e da a
ans e a e in compa ison wi h he p e ious wo. These cha ac e is ics make i app op i-
a e o he applica ion, so LoRa has been selec ed o he de elopmen o his wo k.
LoRa de ice a chi ec u e consis s o he high-pe o mance mic ocon olle , i is a
low-powe ARM
®
Co ex
®
-M0 + based ATSAMD21G18 wi h 256 kB o lash, 32 KB o
SRAM and ope a ing equency up o 48 MHz. I implemen s he LoRaWAN s ack.
The long ange anscei e is he module RFM95 wi h a equency o 868 MHz. I s
main ea u es a e i s high sensi i i y below −148 dBm combined wi h he +20 dBm in e-
g a ed powe ampli ie . I s modula ion modes a e FSK, GFSK, MSK, GMSK, LoRaTM and
OOK. The connec ion in e ace be ween he mic ocon olle and he anscei e is SPI (Se-
ial Pe iphe al In e ace). Finally, he whole sys em is inished wi h an omnidi ec ional
an enna o 1dBi gain. In addi ion, he module inco po a es a iaxial MEMS lis3dh accel-
e ome e and a gene al-pu pose inpu -ou pu connec o . Las ly, he Powe Supply con-
sis ing o a TPS7A05 low quiescen cu en low d op egula o . Figu e 2 shows he block
diag am o he module.
Figu e 2. Shows he block diag am o he module, in he cen e he ARM M0 MCU, (W) LoRa
anscei e , (A) omni-di ec ional an enna, (Ldo) powe supply and (AT) iaxial accele ome e .
Up o his poin , he mos app op ia e ne wo k p o ocol has been selec ed o he
moni o ing and/o con ol o he in elligen signaling o a highway. Fo a ic de ec ion,
i is necessa y ha he ne wo k has o be able o ansmi a smalle dis ances o make a
co ec de ec ion h ough he signal ansmission powe . This is why a sho ange wi e-
less ne wo k is de eloped ha is capable o coexis ing wi h a LoRaWAN ne wo k, which
we will call Sho LoRa. This ne wo k will be p esen ed in he nex subsec ion.
2.2. Sho LoRa Ne wo k Topology
Se e al ne wo k opologies ha e been s udied in Wi eless Senso Ne wo ks (WSN).
The mos p omising opologies consis o a mesh whe e all de ices a e pee o pee and
he e a e no hie a chical ela ionships. This opology is e y complex and p esen s high
main enance equi emen s. In his way, he s a opology shown in Figu e 3 is iable o
his applica ion due o he la ge ansmission ange, i is he opology used o LoRaWAN
p o ocol. S a opology has less e sa ili y communica ion op ions bu i has small main e-
nance equi emen s, i is obus as i is no a mul i-hop opology whe e he ga eway e-
cei es he in o ma ion om all End-De ices (ED) and con ol he ne wo k communica-
ions. This ac gi es a high obus echnology.
Figu e 2.
Shows he block diag am o he module, in he cen e he ARM M0 MCU, (
W
) LoRa
anscei e , (A) omni-di ec ional an enna, (Ldo) powe supply and (AT) iaxial accele ome e .
Up o his poin , he mos app op ia e ne wo k p o ocol has been selec ed o he
moni o ing and/o con ol o he in elligen signaling o a highway. Fo a ic de ec ion,
i is necessa y ha he ne wo k has o be able o ansmi a smalle dis ances o make a
co ec de ec ion h ough he signal ansmission powe . This is why a sho ange wi eless
ne wo k is de eloped ha is capable o coexis ing wi h a LoRaWAN ne wo k, which we
will call Sho LoRa. This ne wo k will be p esen ed in he nex subsec ion.
2.2. Sho LoRa Ne wo k Topology
Se e al ne wo k opologies ha e been s udied in Wi eless Senso Ne wo ks (WSN).
The mos p omising opologies consis o a mesh whe e all de ices a e pee o pee and
he e a e no hie a chical ela ionships. This opology is e y complex and p esen s high
main enance equi emen s. In his way, he s a opology shown in Figu e 3is iable o
his applica ion due o he la ge ansmission ange, i is he opology used o LoRaWAN
p o ocol. S a opology has less e sa ili y communica ion op ions bu i has small main e-
nance equi emen s, i is obus as i is no a mul i-hop opology whe e he ga eway ecei es
he in o ma ion om all End-De ices (ED) and con ol he ne wo k communica ions. This
ac gi es a high obus echnology.
The ne wo k p oposed he e wo ks unde a LoRaWAN ne wo k like he one p esen ed
in Figu e 4and i is implemen ed so ha bo h ne wo ks can wo k oge he .
The Sho Lo a subne equi es i e nodes, ou o hem a e used o de ec ehicle a ic
and a i h ha does he wo k o ne wo k coo dina o , mainly esponsible o sequencing
and es ablishing he wo k cycles o he measu emen s.
Following he LoRaWAN s anda d implemen a ion Sho LoRa use he same s a
opology, whe e he coo dina o node is he ame sende in mode mul icas , and his
node is esponsible o ne wo k synch oniza ion and he inal collec ion o da a o u he
p ocessing. The es o he nodes (#1, #2, #3, #4) a e p omiscuous mode in e ace ne wo k
o packe sni ing so ha hey can also measu emen RSSI wi h each o he and be able o
de ec ehicles. In addi ion, ED nodes a e de ined as Class A in LoRaWAN ne wo k, ha
acco ding o LoRa [
20
] allows o bi-di ec ional communica ions. Sho LoRa sub-ne wo k
use a speci ic single channel o wo k, which makes i di e en om he channels used by
Senso s 2021,21, 338 5 o 24
LoRaWAN. The speci ic ope a ion o he ne wo k o ehicle de ec ion will be p esen ed
in Sec ion 2.4.
Senso s 2021, 21, x FOR PEER REVIEW 5 o 26
Figu e 3. LoRaWAN s anda d opology.
The ne wo k p oposed he e wo ks unde a LoRaWAN ne wo k like he one p e-
sen ed in Figu e 4 and i is implemen ed so ha bo h ne wo ks can wo k oge he .
The Sho Lo a subne equi es i e nodes, ou o hem a e used o de ec ehicle
a ic and a i h ha does he wo k o ne wo k coo dina o , mainly esponsible o se-
quencing and es ablishing he wo k cycles o he measu emen s.
Figu e 4. Sho LoRa ne wo k opology.
Following he LoRaWAN s anda d implemen a ion Sho LoRa use he same s a o-
pology, whe e he coo dina o node is he ame sende in mode mul icas , and his node
is esponsible o ne wo k synch oniza ion and he inal collec ion o da a o u he p o-
cessing. The es o he nodes (#1, #2, #3, #4) a e p omiscuous mode in e ace ne wo k o
packe sni ing so ha hey can also measu emen RSSI wi h each o he and be able o
de ec ehicles. In addi ion, ED nodes a e de ined as Class A in LoRaWAN ne wo k, ha
acco ding o LoRa [20] allows o bi-di ec ional communica ions. Sho LoRa sub-ne wo k
Figu e 3. LoRaWAN s anda d opology.
Senso s 2021, 21, x FOR PEER REVIEW 5 o 26
Figu e 3. LoRaWAN s anda d opology.
The ne wo k p oposed he e wo ks unde a LoRaWAN ne wo k like he one p e-
sen ed in Figu e 4 and i is implemen ed so ha bo h ne wo ks can wo k oge he .
The Sho Lo a subne equi es i e nodes, ou o hem a e used o de ec ehicle
a ic and a i h ha does he wo k o ne wo k coo dina o , mainly esponsible o se-
quencing and es ablishing he wo k cycles o he measu emen s.
Figu e 4. Sho LoRa ne wo k opology.
Following he LoRaWAN s anda d implemen a ion Sho LoRa use he same s a o-
pology, whe e he coo dina o node is he ame sende in mode mul icas , and his node
is esponsible o ne wo k synch oniza ion and he inal collec ion o da a o u he p o-
cessing. The es o he nodes (#1, #2, #3, #4) a e p omiscuous mode in e ace ne wo k o
packe sni ing so ha hey can also measu emen RSSI wi h each o he and be able o
de ec ehicles. In addi ion, ED nodes a e de ined as Class A in LoRaWAN ne wo k, ha
acco ding o LoRa [20] allows o bi-di ec ional communica ions. Sho LoRa sub-ne wo k
Figu e 4. Sho LoRa ne wo k opology.
The Figu e 5shows he LoRaWAN s. Sho Lo a ansmission in e als. I is obse ed
ha he idle pe iods o he LoRaWAN ne wo k a e high, his ac is used o implemen
he Sho LoRa ne wo k ha will ca y ou he communica ions du ing hese idle pe-
iods o LoRaWAN. To a oid da a collisions, he subne wo ks on a di e en channel
han LoRaWAN.
The ansmission ime diag am is shown in he Figu e 6, in which we can see ha
he coo dina ing node o sho Lo a sends ame in mul icas mode (a beacon signal)
o synch onize he ne wo k and s a he measu emen , a e which each o he nodes
pe o ms a mul icas ansmission o collec RSSI in o ma ion om communica ion wi h
he o he nodes ha a e pa o he subne , so LoRa.

Senso s 2021,21, 338 6 o 24
Senso s 2021, 21, x FOR PEER REVIEW 6 o 26
use a speci ic single channel o wo k, which makes i di e en om he channels used by
LoRaWAN. The speci ic ope a ion o he ne wo k o ehicle de ec ion will be p esen ed
in Sec ion 2.4.
The Figu e 5 shows he LoRaWAN s. Sho Lo a ansmission in e als. I is ob-
se ed ha he idle pe iods o he LoRaWAN ne wo k a e high, his ac is used o imple-
men he Sho LoRa ne wo k ha will ca y ou he communica ions du ing hese idle
pe iods o LoRaWAN. To a oid da a collisions, he subne wo ks on a di e en channel
han LoRaWAN.
Figu e 5. LoRaWAN s. Sho Lo a ansmission in e als.
The ansmission ime diag am is shown in he Figu e 6, in which we can see ha he
coo dina ing node o sho Lo a sends ame in mul icas mode (a beacon signal) o syn-
ch onize he ne wo k and s a he measu emen , a e which each o he nodes pe o ms
a mul icas ansmission o collec RSSI in o ma ion om communica ion wi h he o he
nodes ha a e pa o he subne , so LoRa.
Figu e 6. Sho Lo a ansmission ime diag am.
2.3. Sho Lo a Ne wo k P o ocol
The objec i e o he de elopmen o his applica ion is he de ec ion o s opped ehi-
cles o slow a ic on Eu ope highways. The a ge egion is impo an due o he e-
s ic ions p esen ed by he egula ions each one is subjec o. In his case, i is egula ed by
EU Ha monized NRI o he 863–870 MHz band. The speci ic egula ion o his kind o
ne wo ks is gi en by
he Eu opean Telecommunica ions S anda ds Ins i u e (ETSI) ETSI
EN 300 220-2 V3.2.1.33 [21].
On he o he hand, i is necessa y o es ablish a limi ha we unde s and o de e mine
a slow a ic speed. Assuming ha 50 km/h (13.9 m/s) is slow a ic, and es ima ing he
size o a ehicle o app oxima ely 4 m. Based on hese da a, we can es ablish ha he cycle
ime o he obs acle measu emen wi h he adio equency sys em exposed a he nex
poin is 288 ms as seen in Equa ion (1).
Figu e 5. LoRaWAN s. Sho Lo a ansmission in e als.
Senso s 2021, 21, x FOR PEER REVIEW 6 o 26
use a speci ic single channel o wo k, which makes i di e en om he channels used by
LoRaWAN. The speci ic ope a ion o he ne wo k o ehicle de ec ion will be p esen ed
in Sec ion 2.4.
The Figu e 5 shows he LoRaWAN s. Sho Lo a ansmission in e als. I is ob-
se ed ha he idle pe iods o he LoRaWAN ne wo k a e high, his ac is used o imple-
men he Sho LoRa ne wo k ha will ca y ou he communica ions du ing hese idle
pe iods o LoRaWAN. To a oid da a collisions, he subne wo ks on a di e en channel
han LoRaWAN.
Figu e 5. LoRaWAN s. Sho Lo a ansmission in e als.
The ansmission ime diag am is shown in he Figu e 6, in which we can see ha he
coo dina ing node o sho Lo a sends ame in mul icas mode (a beacon signal) o syn-
ch onize he ne wo k and s a he measu emen , a e which each o he nodes pe o ms
a mul icas ansmission o collec RSSI in o ma ion om communica ion wi h he o he
nodes ha a e pa o he subne , so LoRa.
Figu e 6. Sho Lo a ansmission ime diag am.
2.3. Sho Lo a Ne wo k P o ocol
The objec i e o he de elopmen o his applica ion is he de ec ion o s opped ehi-
cles o slow a ic on Eu ope highways. The a ge egion is impo an due o he e-
s ic ions p esen ed by he egula ions each one is subjec o. In his case, i is egula ed by
EU Ha monized NRI o he 863–870 MHz band. The speci ic egula ion o his kind o
ne wo ks is gi en by
he Eu opean Telecommunica ions S anda ds Ins i u e (ETSI) ETSI
EN 300 220-2 V3.2.1.33 [21].
On he o he hand, i is necessa y o es ablish a limi ha we unde s and o de e mine
a slow a ic speed. Assuming ha 50 km/h (13.9 m/s) is slow a ic, and es ima ing he
size o a ehicle o app oxima ely 4 m. Based on hese da a, we can es ablish ha he cycle
ime o he obs acle measu emen wi h he adio equency sys em exposed a he nex
poin is 288 ms as seen in Equa ion (1).
Figu e 6. Sho Lo a ansmission ime diag am.
2.3. Sho Lo a Ne wo k P o ocol
The objec i e o he de elopmen o his applica ion is he de ec ion o s opped
ehicles o slow a ic on Eu ope highways. The a ge egion is impo an due o he
es ic ions p esen ed by he egula ions each one is subjec o. In his case, i is egula ed
by EU Ha monized NRI o he 863–870 MHz band. The speci ic egula ion o his kind o
ne wo ks is gi en by he Eu opean Telecommunica ions S anda ds Ins i u e (ETSI) ETSI
EN 300 220-2 V3.2.1.33 [21].
On he o he hand, i is necessa y o es ablish a limi ha we unde s and o de e mine
a slow a ic speed. Assuming ha 50 km/h (13.9 m/s) is slow a ic, and es ima ing he
size o a ehicle o app oxima ely 4 m. Based on hese da a, we can es ablish ha he cycle
ime o he obs acle measu emen wi h he adio equency sys em exposed a he nex
poin is 288 ms as seen in Equa ion (1).
=x
; =x
=4 m
50 km
h
=4 m
13.9 m
s
=288 ms, (1)
whe e: is eloci y, x is space and is ime.
Following he nodal dis ibu ion p esen ed in [
15
] which has shown good esul s o
o he wi eless communica ion echnologies, bu main aining he LoRaWAN s uc u e ( o
which 5 de ices a e equi ed which will be jus i ied la e ), i is necessa y ha each node
ansmi communica es e e y 50 ms.
On he o he hand, he use o a s anda d LoRaWAN ne wo k has some wo k cycle and
ansmission powe equi emen s ha make i impossible o use his p o ocol in a s anda d
way, as can be seen in Table 2.
Senso s 2021,21, 338 7 o 24
Table 2.
Bands and egula ions acco ding o he Eu opean Reseach Council (ERC) Recommenda ion
70-03 and LoRaWAN Speci ica ions [20–23].
Band Numbe F equency (MHz) Du y Cycle Powe
g0 865.0–868.0 1% o LBT + AFA 125 mW = 14 dBm
g1 868.0–868.6 1% o LBT + AFA 25 mW = 14 dBm
g2 868.7–869.2 0.1% o LBT + AFA 25 mW = 14 dBm
g3 869.4–869.65 10% o LBT + AFA 500 mW = 27 dBm
g4 869.7–870.0 1% o LBT + AFA
25 mW (no du y-cycle
equi emen i powe < 5
mW/7 dBm)
1LBT + AFA: Lis en Be o e Talk (LBT) wi h Adap i e F equency Agili y (AFA).
Pa ame e s gi en in Table 2, de e mine he minimum ne wo k ansmission pe iod.
The heo e ical analysis s a s de ining he ime on he ai , T
on- he-ai
(2), o packe ansmis-
sion du a ion. This ime de e mines in u n he minimum ime ha he ne wo k should
emain wi hou ansmi ing in o ma ion.
Ton- he-ai =np eamble +npayload·Tsymbol =nsymbols·Tsymbol (2)
T
on- he-ai
depends on he numbe o symbols ansmi ed, including hose o he
p eamble and he payload, whose sum is he numbe o symbols ansmi ed; and om
he ime necessa y o ansmi each o hese symbols (T
symbol
). T
symbol
is ob ained om he
a e o sending o symbols (R
Symbol
) which depends on he Sp eading Fac o (SF) and he
signal bandwid h (BW).
Tsymbol =1
Rsymbol
(3)
Rsymbol =BW
2SF (4)
Wi h hese equa ions, he ime on he ai and ansmission pe iods a e shown in
Table 3 o 6 by es and he di e en ne wo k pa ame e s.
Table 3.
Ne wo k con igu a ion s. du y cycle, ime on ai and da a ansmission pe iod wi h 6 by es o da a ansmission.
F equency
[MHz] Powe [mW] Modula ion
[kHz]
Sp eading
Fac o
Time on Ai
[ms] Du y Cycle [%] Cycle Scan
Ne wo k [s]
868.3 25 250 SF7 18.05 1 1.805
869.525 25 125 SF9 123.90 10 1.239
869.525 500 125 SF7 36.1 1 3.61
868.8 25 125 FSK 3.871 1 0.386
869.850 5 125 SF7 36.1 100 10.1805
1Wi h a spec um access echnique such as LBT o equi alen and a maximum ansmi pe iod o 1 min o each ansmission.
As a esul , he needs o a ic de ec ion wi h a speed below 50 km/h, and conside ing
he ne wo k limi a ions, in e ms o he du y cycle and sampling pe iod obse ed in
Table 3
, we can conclude ha i is necessa y o implemen a subne wo k ha can li e wi h
LoRaWAN. This subne wo k will wo k wi h 869.85 MHz, 125 kHz o modula ion and SF7
5 mW (7 dBm) and a du y 100%, complying wi h he egula ions indica ed in ERC [
21
]. This
ne wo k wo ks ou o LoRaWAN speci ica ions bu inside o ETSI EN 300 220-2 V3.2.133
egula ion o Sho Range De ices (SRD) [24].
In o de o ha e disposable all channel ime and no o a ec he compa ibili y o he
sys em wi h LoRaWAN, a subne wo k compa ible has been implemen ed. To mee his
equi emen , he new ne wo k will base i s implemen a ion o he LoRaWAN p o ocol.
Sho LoRaWAN implemen a ion s. s anda d LoRaWAN a e compa ed below.
Senso s 2021,21, 338 8 o 24
Figu e 7shown he s anda d LoRaWAN Radio PHY laye message s uc u e in o de
o es ablish a compa ison line wi h he p o ocol p oposed in his pape .
Senso s 2021, 21, x FOR PEER REVIEW 8 o 26
869.850 5 125 SF7 36.1 100
1 0.1805
1 Wi h a spec um access echnique such as LBT o equi alen and a maximum ansmi pe iod o 1
min o each ansmission.
As a esul , he needs o a ic de ec ion wi h a speed below 50 km/h, and conside ing
he ne wo k limi a ions, in e ms o he du y cycle and sampling pe iod obse ed in Table
3, we can conclude ha i is necessa y o implemen a subne wo k ha can li e wi h Lo-
RaWAN. This subne wo k will wo k wi h 869.85 MHz, 125 kHz o modula ion and SF7 5
mW (7 dBm) and a du y 100%, complying wi h he egula ions indica ed in ERC [21]. This
ne wo k wo ks ou o LoRaWAN speci ica ions bu inside o ETSI EN 300 220-2 V3.2.133
egula ion o Sho Range De ices (SRD) [24].
In o de o ha e disposable all channel ime and no o a ec he compa ibili y o he
sys em wi h LoRaWAN, a subne wo k compa ible has been implemen ed. To mee his
equi emen , he new ne wo k will base i s implemen a ion o he LoRaWAN p o ocol.
Sho LoRaWAN implemen a ion s. s anda d LoRaWAN a e compa ed below.
Figu e 7 shown he s anda d LoRaWAN Radio PHY laye message s uc u e in o de
o es ablish a compa ison line wi h he p o ocol p oposed in his pape .
Figu e 7. LoRaWAN packe s uc u e [20,22].
The LoRaWAN s anda d has a p eamble alue o 0 × 34, his pa ame e is used o
es ablish a synch onized LoRaWAN ne wo k. The subne wo k p oposed equi es a di -
e en p eamble, i is 0 × 12. The ollowing pa ame e s, PHDR and PHDR a e used in ex-
plici mode, de aul con igu a ion o LoRa de ices, bu a e anspa en o he use and
no used by hem. Fo his eason, PHDR and PHDR_CRC ha e been simpli ied in Sho
LoRa, in he same way as MIC.
MHDR is made up o MType, RFU and Majo . MHDR has been educed only o a
simpli ied MType, since he p oposed ne wo k does no pe o m a nego ia ed ne wo k
Figu e 7. LoRaWAN packe s uc u e [20,22].
The LoRaWAN s anda d has a p eamble alue o 0
×
34, his pa ame e is used
o es ablish a synch onized LoRaWAN ne wo k. The subne wo k p oposed equi es a
di e en p eamble, i is 0
×
12. The ollowing pa ame e s, PHDR and PHDR a e used in
explici mode, de aul con igu a ion o LoRa de ices, bu a e anspa en o he use and
no used by hem. Fo his eason, PHDR and PHDR_CRC ha e been simpli ied in Sho
LoRa, in he same way as MIC.
MHDR is made up o MType, RFU and Majo . MHDR has been educed only o a
simpli ied MType, since he p oposed ne wo k does no pe o m a nego ia ed ne wo k join
o equi e an access code. The Table 4shows di e ences be ween S anda d LoRa MType
and Sho LoRa MType.
The nex LoRaWAN pa ame e , FHDR, is composed by 4 by es o de ice add ess,
1 by e o F ame Con ol, 2 by es o F ame Coun e and up o 15 by es o ame op ions. In
Sho LoRa, hese equi emen s ha e been educed, FHDR is composed o 4 bi s add ess
(16 nodes), 2 bi s o con ol (FC l) and 4 bi s o coun he payload leng h (FCn ). The
2 bi s o con ol (FC l) a e: he i s one is he acknowledge (ACK) and he second is
Rese ed o Fu u e Usage (RFU). In addi ion, F ame Op ions (FOp ) ha e been emo ed
in o de o simpli y he ne wo k p o ocol.
Senso s 2021,21, 338 9 o 24
Table 4. Media Access Con ol (MAC) message ypes.
S anda d LoRa
MType Value
Sho LoRa
MType Value Desc ip ion
000 No A ailable Join Reques
001 No A ailable Join Accep
010 00 Uncon i med Da a Up
011 01 Uncon i med Da a Down
100 10 Con i med Da a Up
101 11 Con i med Da a Down
110 No A ailable RFU
111 No A ailable P op ie a y
The ADR and ADRACKReq ields wi hin FC l a e supp essed espec o he LoRaWAN
s anda d because his subne wo ks by se ing he da a a e and ansmission powe .
The FOp sLen ield wi hin FC l and FOp s is supp essed because MAC commands
can be sen in he FRMPayload ield i he FPo ield is se o 0.
Las ly, he payload ame (FRMPayload) con ained he da a collec ion ha could be
sen h ough he ne wo k, ha could ha e up o 14 by es. In his way using he enc yp ion
gi en by Ad anced Enc yp ion S anda d (AES), we ma ch he da a packe sen o he
minimum packe encoded by AES 128 bi s. They mus always be enc yp ed and mus no
exceed he maximum leng h o FRMPayload.
The Figu e 8shows g aphically he Radio PHY laye message s uc u e p oposed o
Sho LoRa.
The ne wo k nodes ha e been de eloped o wo k in Class A in o de o hey can be
used o a s anda d LoRaWAN ne wo k when i is equi ed by he applica ion and he es
o he ime you can wo k in he p oposed low consump ion ne wo k.
Unde he p oposed con igu a ion, 869.85 MHz, 125 kHz o modula ion and SF7, he
di e en imes in he ai o he ansmi ed in o ma ion ha e been calcula ed and i is
shown in Table 5.
Al hough 50 km/h may seem like a low speed o de ec ing a ic on a highway, bu
his limi is jus i ied below.
Eu opean ETSI egula ion es ablishes a single channel o wo k on FSK wi hin he
LoRaWAN speci ica ion. I is ue ha wo king in FSK would allow he de ec ion and
measu emen o highe ci cula ion speeds co esponding o he Da a Ra e 7 (DR7) o he
LoRa FSK speci ica ion wi h 50 kbi /s [
22
] compa ed o 11 kbi /s in LoRa SF7 (DR6) in
he EU 863–870 MHz ISM Band. This limi a ion ound o LoRa in speed measu emen
imp o es egula ion in o he coun ies such as he Uni ed S a es whe e using DR13 and
SF7 he da a a e is 21.9 kbi /s in US 902–928 MHz ISM Band.
Fu he mo e, analyzing he in o ma ion p esen ed in [
25
] whe e he LoRa commu-
nica ion is compa ed wi h he FSK communica ion, i can be deduced ha as he bi a e
inc eases, he LoRa sensi i i y dec eases, app oaching he FSK sensi i i y. Al hough, in his
same documen , we no e ha immuni y agains noise p oduced by o he adio equency
signals is be e in LoRa han in FSK. This is an impo an ac o since i is in ended o
de ec communica ion dis u bances caused by obs acles in he di ec line o sigh and no
by o he RF signals.
On he o he hand, in he da ashee o he RF anscei e used [
26
] explains he
di e ence in he way o calcula ing he RSSI in bo h LoRa and FSK. In LoRa we ha e access
o an a e age RSSI alue o he sen packe while in FSK he RSSI alue is smoo hed on a
use de ined numbe o measu ed RSSI samples, he g ea e he numbe o samples, he
g ea e he p ecision bu he g ea e he delay in he measu emen . This ea u e o LoRa is
in e es ing since i allows us o wo k independen ly o each da a packe , imp o ing he
obus ness in he de ec ion o obs acles.
Senso s 2021,21, 338 16 o 24
Senso s 2021, 21, x FOR PEER REVIEW 17 o 26
ame e s ha may in e ene in communica ion and hey a e no con olled. This in o -
ma ion will be used o compensa e o hese losses in communica ions a ga es and c osses
and o ha e an adequa e base le el o bias le el.
Figu e 14. Miscellaneous losses calcula ion subsys em.
Fo he heo e ical calcula ion o communica ion losses, bo h he ini ial ansmission
powe le el and he elemen s ha con ibu e gain o communica ion, such as an ennas,
and hose ha p oduce losses, such as cables and connec o s, a e aken in o accoun . F ee
space losses ha e also been aken in o accoun and a e ela ed o he dis ance ha sepa-
a es emi e and ecei e as seen in Equa ion (5).
𝐿 =20 𝑙𝑜𝑔
4𝜋𝑑
𝜆 (5)
whe e 𝐿 is pa h loss, usually ee space loss (dB), 𝜆 is he signal wa eleng h and 𝑑 is
he dis ance be ween he an ennas in he same uni s as wa eleng h.
This alue o he powe losses in he communica ion ha we will call miscellaneous
losses as seen in Equa ion (6), due o he possibili y ha hey a e p oduced by elemen s o
unknown na u e, is calcula ed in bo h di ec ions o he same, ha is, in he communica ion
om node #1 o #3 and om #3 o #1, o la e a e age i . The same is done o he commu-
nica ion o nodes #2 and #4.
𝐿=𝑃
 +𝐺 −𝐿 −𝐿 +𝐺 −𝐿 +𝑃
 (6)
whe e: 𝐿 is miscellaneous losses (dB), 𝑃 is ansmi e ou pu powe (dBm), 𝐺 is
ansmi e an enna gain (dBi), 𝐿 is ansmi e losses (dB), 𝐿 is pa h loss, usually
ee space loss (dB), 𝐺 is ecei e an enna gain (dBi), 𝐿 is ecei e losses (coax,con-
nec o s...) (dB) and 𝑃 is ecei ed powe RSSI (dBm).
Figu e 15 shows he es o he de ec ion algo i hm. In i we can see how he miscel-
laneous loss da a calcula ed using he blocks in Figu e 14 is en e ed in o a new block. In
Figu e 14. Miscellaneous losses calcula ion subsys em.
This alue o he powe losses in he communica ion ha we will call miscellaneous
losses as seen in Equa ion (6), due o he possibili y ha hey a e p oduced by elemen s o
unknown na u e, is calcula ed in bo h di ec ions o he same, ha is, in he communica ion
om node #1 o #3 and om #3 o #1, o la e a e age i . The same is done o he
communica ion o nodes #2 and #4.
LM=PTx +GTX −LTX −LFS +GRX −LRX +PRX (6)
whe e:
LM
is miscellaneous losses (dB),
PTx
is ansmi e ou pu powe (dBm),
GTX
is
ansmi e an enna gain (dBi),
LTX
is ansmi e losses (dB),
LFS
is pa h loss, usually ee
space loss (dB),
GRX
is ecei e an enna gain (dBi),
LRX
is ecei e losses (coax,connec o s...)
(dB) and PRX is ecei ed powe RSSI (dBm).
Figu e 15 shows he es o he de ec ion algo i hm. In i we can see how he miscel-
laneous loss da a calcula ed using he blocks in Figu e 14 is en e ed in o a new block. In
his calcula ion block, he heo e ical calcula ion o communica ion losses be ween ga es o
be ween c osses is ob ained, aking in o accoun he ee space losses o he sepa a ion
dis ance o hese communica ions o hese nodes. Miscellaneous losses a e added o his
calcula ion, which a e assumed o be he same o all communica ions in he sys em, since
hey a e wi hin he same en i onmen . Wi h his, a base band o bias le el is es ablished on
which o compa e luc ua ions in he ecep ion powe .
PRX =PTx +GTX −LTX −LFS −LM+GRX −LRX (7)
whe e:
PRX
is ecei ed powe RSSI (dBm),
PTx
is ansmi e ou pu powe (dBm),
GTX
is
ansmi e an enna gain (dBi),
LTX
is ansmi e losses (dB),
LFS
is pa h loss, usually ee
space loss (dB),
LM
is miscellaneous losses (dB),
GRX
is ecei e an enna gain (dBi) and
LRX is ecei e losses (coax, connec o s...) (dB).

Senso s 2021,21, 338 17 o 24
Senso s 2021, 21, x FOR PEER REVIEW 19 o 26
Figu e 15. Simulink diag am o link budge compensa ed da a analysis me hod.
Figu e 15. Simulink diag am o link budge compensa ed da a analysis me hod.
Finally, his bias ecep ion powe (
PRX
) is sub ac ed om he RSSI alues ecei ed
a each o he bidi ec ional ga es and c osses ecep ion alues, as app op ia e. Wi h his
compensa ed base le el he new base le el is a ound ze o, wo limi s a e se , one uppe
and one lowe . The chosen limi alues ha e been ob ained expe imen ally o each
communica ion pai . When one o he signals c osses he lowe limi i p oduces a logical
‘1
0
. In his me hod, unlike he p e ious one, he ou pu alue o he compa ison is passed
h ough an OR logic ga e so ha when one o he wo RSSI signals p oduces a posi i e, i is
de ec ed. I his de ec ion exceeds he maximum ime es ablished based on he leng h o
he ehicle and he minimum speed, we ace slow o s opped a ic.
This de ec ion is ca ied ou a each o he ga es and c osses. The esul s will be
explained in he nex subsec ion.
Senso s 2021,21, 338 18 o 24
3.3. Resul s
Figu e 16 shows he esul s o he ehicle de ec ion algo i hms. Fo he compa ison
and analysis o he esul s, he da a ob ained by means o he wo lase ba ie s loca ed
be ween he links o he ga es a e aken as a e e ence. The esponse ime o hese lase
ba ie s has a high accu acy, p o iding an adequa e e e ence o sys em alida ion.
Senso s 2021, 21, x FOR PEER REVIEW 20 o 26
3.3. Resul s
Figu e 16 shows he esul s o he ehicle de ec ion algo i hms. Fo he compa ison
and analysis o he esul s, he da a ob ained by means o he wo lase ba ie s loca ed
be ween he links o he ga es a e aken as a e e ence. The esponse ime o hese lase
ba ie s has a high accu acy, p o iding an adequa e e e ence o sys em alida ion.
The case p esen ed in Figu e 16 co esponds o one o he es s ca ied ou in which
he ehicle uns a 30 km/h and he dis ance be ween ga es is 20 m. In he es , he ca
en e s he de ec ion zone wice, he i s o which en e s h ough he ga e ha we will now
call Ga e 2, o med by nodes #3 and #4, and exi s h ough wha we will call Ga e 1, co e-
sponding o nodes #1 and #2.
In he igu e we can see se e al hings, he i s o which a e lase ba ie s ha p o-
ide an answe i s , consis en wi h i s use as a e e ence sys em. Nex , we obse e ha
he link budge compensa ed me hod has a as e and equal de ec ion on bo h ga es; while
he me hod based on he disc e e de i a i e esponds somewha la e and wi h signals o
non-uni o m wid h.
Despi e he di e ences shown, he wo da a p ocessing me hods a e equally alid o
de ec ing he passage o ehicles.
Figu e 16. Abo e, i is shown he digi al signals achie ed a e o apply he da a analysis me hod compa ed wi h lase
signals. Below a e RSSI signal a ia ions o e 0 dB. Bo h g aphics co espond wi h 30 km/h o ca eloci y and 20 m
be ween ga es.
The lowe g aph o Figu e 16 shows he a enua ions p oduced in he RSSI alue wi h
he passage o ehicles wi h espec o an a enua ion le el o 0dBs. The a enua ions o
he ansmissions o bo h ga es and o he bidi ec ional communica ion o he pai s o
nodes ha o m each o hem a e p esen ed sepa a ely. The g aph e e s o hese signals
as Ga e 1 Link 1, his being he communica ion om node #1 o #2; Ga e 1 Link 2 om #2
o #1; Ga e 2 Link 1 om #3 o #4 and Ga e 2 Link 2 om #4 o #3, hus ep esen ing all he
de ec ion signals p esen be ween he ga es communica ions.
The RSSI da a o his analysis is collec ed in Table 6, bo h a ga e and c osslinks.
S udying wha his a enua ion looks like shows ha he ga e signs a e mo e a enua ed
han he c ossing signals. This educ ion in a enua ion is due o he ac ha in he c oss-
ings he e a e mo e de la ions, mo e ebounds and; he e o e, less a enua ion and less
sensi i i y.
Figu e 16.
Abo e, i is shown he digi al signals achie ed a e o apply he da a analysis me hod compa ed wi h lase
signals. Below a e RSSI signal a ia ions o e 0 dB. Bo h g aphics co espond wi h 30 km/h o ca eloci y and 20 m
be ween ga es.
The case p esen ed in Figu e 16 co esponds o one o he es s ca ied ou in which
he ehicle uns a 30 km/h and he dis ance be ween ga es is 20 m. In he es , he ca
en e s he de ec ion zone wice, he i s o which en e s h ough he ga e ha we will
now call Ga e 2, o med by nodes #3 and #4, and exi s h ough wha we will call Ga e 1,
co esponding o nodes #1 and #2.
In he igu e we can see se e al hings, he i s o which a e lase ba ie s ha p o ide
an answe i s , consis en wi h i s use as a e e ence sys em. Nex , we obse e ha he
link budge compensa ed me hod has a as e and equal de ec ion on bo h ga es; while
he me hod based on he disc e e de i a i e esponds somewha la e and wi h signals o
non-uni o m wid h.
Despi e he di e ences shown, he wo da a p ocessing me hods a e equally alid o
de ec ing he passage o ehicles.
The lowe g aph o Figu e 16 shows he a enua ions p oduced in he RSSI alue wi h
he passage o ehicles wi h espec o an a enua ion le el o 0dBs. The a enua ions o he
ansmissions o bo h ga es and o he bidi ec ional communica ion o he pai s o nodes
ha o m each o hem a e p esen ed sepa a ely. The g aph e e s o hese signals as Ga e 1
Link 1, his being he communica ion om node #1 o #2; Ga e 1 Link 2 om #2 o #1; Ga e
2 Link 1 om #3 o #4 and Ga e 2 Link 2 om #4 o #3, hus ep esen ing all he de ec ion
signals p esen be ween he ga es communica ions.
The RSSI da a o his analysis is collec ed in Table 6, bo h a ga e and c osslinks.
S udying wha his a enua ion looks like shows ha he ga e signs a e mo e a enua ed
han he c ossing signals. This educ ion in a enua ion is due o he ac ha in he
c ossings he e a e mo e de la ions, mo e ebounds and; he e o e, less a enua ion and
less sensi i i y.
Senso s 2021,21, 338 19 o 24
Table 6. RSSI measu emen s om communica ion ansmissions du ing he es wi h 20 m o dis ance be ween ga es.
Connec ion Max. Signal
(dB)
Min. Signal
(dB)
Mean Signal
(dB)
RMS Signal
(dB)
RMS Noise
(dB)
RMS
SNR 1
Ga es Dis ance 20 m
Ga e1(#1–#2) 27.00 8.00 16.54 16.96 0.46 16.50
Ga e1(#2–#1) 27.00 7.00 15.46 16.13 0.51 15.62
Ga e1(#3–#4) 27.00 10.00 16.21 16.65 0.94 15.71
Ga e1(#4–#3) 26.00 8.00 15.95 16.50 1.00 15.50
Mean Ga es 26.75 8.25 16.04 16.56 0.73 15.83
C osses1(#1–#4) 19.00 6.00 8.83 9.44 0.35 9.09
C osses1(#4–#1) 16.00 6.00 8.96 9.42 0.85 8.57
C osses2(#2–#3) 12.00 6.00 7.14 7.34 0.73 6.60
C osses2(#3–#4) 20.00 6.00 7.63 8.30 1.21 7.09
Mean C osses 16.75 6.00 8.14 8.62 0.79 7.84
Ga es Dis ance 10 m
Ga e1(#1–#2) 29.00 14.00 19.68 19.89 0.20 19.69
Ga e1(#2–#1) 29.00 13.00 19.71 20.11 0.50 19.61
Ga e1(#3–#4) 33.00 9.00 18.00 19.01 0.97 18.03
Ga e1(#4–#3) 28.00 10.00 16.95 17.61 0.53 17.08
Mean Ga es 29.75 11.50 18.58 19.15 0.55 18.60
C osses1(#1–#4) 22.00 7.00 12.34 13.01 0.40 12.61
C osses1(#4–#1) 24.00 8.00 12.03 12.62 0.31 12.31
C osses2(#2–#3) 24.00 7.00 11.64 12.63 0.42 12.21
C osses2(#3–#4) 19.00 6.00 9.92 10.43 0.63 9.80
Mean C osses 22.25 7.00 11.48 12.17 0.44 11.73
1SNR: Signal Noise Ra io.
On he o he hand, he a e age ga e a enua ion is a leas 15 dB, and his RSSI
a enua ion alue occu s in bo h di ec ions o communica ion. These wo cha ac e is ics
make he sys em e y obus .
C ossings can be used o sense he di ec ion o he ehicle o o make he ga e de ec ion
algo i hm mo e obus . Al hough he de ec ion sys em wo ks sa is ac o ily using only he
ga e signals.
Fo unobs uc ed links (#1–#3 and #2–#4), hey ha e a signal RMS a enua ion o
10 dBs and RMS noise o 1 dB.
Based on he RSSI in o ma ion p ocessed and he co esponding digi al signal ob ained
o each da a analysis me hod, oge he wi h he e e ence lase sys em, he ime i akes
o a ehicle o pass om one ga e o he nex has been measu ed. F om he in o ma ion
shown in Figu e 16 he pulse ins an s a e ob ained o he calcula ion o he ime di e ence.
Since he dis ance be ween he ga es is known, which is ixed o each case s udy (10 and
20 m), and he ime i akes o he ehicle o a el ha gap is measu ed, he speed o he
ca can be calcula ed.
Table 7shows hese esul s. In he able, when he ehicle en e s h ough Ga e 1 and
exi s h ough Ga e 2, i is conside ed posi i e. The opposi e di ec ion is conside ed nega i e
(Ga e 2 o Ga e 1).
F om he in o ma ion p esen ed, bo h o 10 and 20 m o ga e dis ances, i is obse ed
ha o low speeds o 10 o 30 km/h he measu emen h ough he wi eless ne wo k is
qui e accu a e and i s esponse ime is qui e as compa ed o he e e ence measu emen
ob ained using he lase ba ie s. The accu acy dec eases as he speed inc eases and his
is due o he sampling ime wi h which he ne wo k wo ks. This aspec can be imp o ed,
bu i would be necessa y o en e an ope a ing mode ou side he ETSI EN 300 220-2
s anda d. The o he op ion is o wo k in FSK mode ha con lic s wi h he wo king modes
o LoRaWAN ga eways, comp omising compa ibili y be ween LoRaWAN and Sho LoRa.
Senso s 2021,21, 338 20 o 24
Table 7. Da a analysis and speed es ima ion o he di e en p ocessing me hods.
Lase Measu emen s
Sho LoRa Ne wo k
Measu emen s
(De i a i e Me hod)
Sho LoRa Ne wo k
Measu emen s (Link Budge
Compensa ed Me hod)
Ca Veloci y
(Km/h)
Di e en ial
Ga es Time (s)
Veloci y
(km/h)
Di e en ial
Ga es Time (s)
Veloci y
(km/h)
Di e en ial
Ga es Time (s)
Veloci y
(km/h)
Ga es Dis ance 20 m
10 6.77 10.63 6.8 10.58 6.8 10.58
10 −6.78 −10.61 −6.5 −11.07 −6.7 −10.74
20 3.54 20.32 3.8 18.94 3.5 20.57
20 −3.78 −19.04 −3.7 −19.45 −3.7 −19.45
30 2.59 27.71 2.8 25.71 2.5 28.8
30 −2.54 −28.27 −2.5 −28.8 −2.2 −32.72
40 1.92 37.42 2 36 2 36
40 −1.87 −38.35 −1.7 −42.35 −1.8 −40
50 1.51 47.58 1.5 48 1.5 48
50 −1.51 −47.65 −1.5 −48 −1.3 −55.38
Ga es Dis ance 10 m
10 3.69 9.92 3.5 10.28 3.7 9.72
10 −2.92 −12.32 −2.8 −12.85 −2.7 −13.33
20 1.92 18.73 2 18 2 18
20 −1.93 −18.65 −1.7 −21.17 −1.7 −21.17
30 1.27 28.27 1.3 27.69 1.5 24
30 −1.21 −29.72 −1.3 −27.69 −1−36
40 0.93 38.37 1 36 1 36
40 −0.93 −38.66 −0.7 −51.42 −0.8 −45
50 0.72 50.06 0.8 45 0.7 51.42
50 −0.75 −48 −0.8 −45 −0.6 −60
Wi h he da a shown in Table 7, he absolu e and ela i e e o s in he measu emen
we e calcula ed o bo h me hods, always aking as a e e ence he measu emen ob ained
om he lase ba ie s. The e o alues a e shown in Table 8.
In gene al, i can be seen ha he e o inc eases wi h speed and is g ea e when he
dis ance be ween ga es is 10 m, since he ime i akes o he ehicle o c oss bo h a es is
less han in he case o 20 m. As al eady men ioned, his is p oduced by he sampling ime.
I can also be seen ha he e o in he de i a i e me hod is gene ally less han in he link
budge compensa ed me hod.
3.4. Dicussion
We can ind in he li e a u e di e en communica ion p o ocols o he In elligen
T anspo Sys em (ITS). Two o hese p o ocols, pe haps he mos ele an in his a ea a e:
6LoWPAN and IEEE 802.11p. In [
28
,
29
] he au ho s alk abou he use o he 6LoWPAN
and IEEE 802.11p p o ocols, espec i ely, in ITS applica ions.
Bo h p o ocols ha e g ea e po en ial o hese applica ions, as hey allow Vehicle- o-
Vehicle (V2V), Vehicle- o-In as uc u e (V2I) communica ion, as well as global communi-
ca ion ia he In e ne o Things (IoT). On he o he hand, hese p o ocols equi e a g ea e
in as uc u e and ha he ehicles ha e a compa ible communica ion sys em.
The implemen a ion o his ype o sys em is much mo e complex and a p esen in
he a ge coun y o he p ojec , i is no easible o de elop a communica ion ne wo k
o his ype, in ol ing bo h he oad communica ion/moni o ing in as uc u e and he
ehicles ha ci cula e h ough i .
Senso s 2021,21, 338 21 o 24
Table 8. A e age absolu e and ela i e e o s o bo h ca de ec ion me hods.
De i a i e Me hod Link Budge
Compensa ed Me hod
Ca Veloci y
(Km/h) Absolu e E o Rela i e E o Absolu e E o Rela i e E o
Ga es Dis ance 20 m
10 0.05 0.004 0.05 0.004
10 0.46 0.04 0.13 0.01
20 1.38 0.07 0.24 0.01
20 0.42 0.02 0.42 0.02
30 2.00 0.07 1.09 0.04
30 0.52 0.02 4.45 0.16
40 1.42 0.04 1.42 0.04
40 3.99 0.10 1.64 0.04
50 0.41 0.01 0.41 0.01
50 0.35 0.01 7.73 0.16
Ga es Dis ance 10 m
10 0.37 0.04 0.05 0.02
10 0.53 0.04 0.46 0.08
20 0.73 0.04 1.38 0.04
20 2.52 0.14 0.42 0.14
30 0.59 0.02 2.00 0.15
30 2.04 0.07 0.52 0.21
40 2.38 0.06 1.42 0.06
40 12.76 0.33 3.99 0.16
50 5.07 0.10 0.41 0.03
50 3.00 0.06 0.35 0.25
The objec i e o he p oposed wo k is much mo e modes , in his sense, and aims o
moni o he condi ion and clima e o a highway, lea ing aside he in o ma ion om he
ehicles. The main objec i e is o be able o de ec an acciden o a si ua ion o collapse
on a mo o way, wi h ex eme impo ance in ci cums ances in which wea he condi ions
can hinde isibili y du ing a ic on he oad, inc easing he isk o an acciden . This
is wi hou he need o ehicles o ca y a communica ion sys em compa ible wi h ha
ins alled on he oad, ha is, i is a sys em solely in ended o moni o he condi ion and
en i onmen o he oad ha uses he communica ion signal i sel o de ec ci cums ances
in which may be a ehicle s opped o mo ing slowly on he oad.
This wo k p esen s a a ic de ec ion sys em based on a LoRa ne wo k ha wo ks
wi h a communica ion equency o 868 MHz. We can ind in he li e a u e o he wo ks
o a simila na u e ha use WiFi communica ion echnologies, IEEE 802.11b [
14
,
15
], IEEE
802.11n [
13
] o Low Powe Wi eless Senso Ne wo ks (LPWN) as IEEE 802.15.4 [
16
] wi h
communica ion equencies in he 2.4 GHz band. Bo h communica ion equencies a e
wi hin he Indus ial, Scien i ic and Medical (ISM) bands.
These wo ks a e o ien ed o applica ion in ci ies, places whe e 2.4 GHz wi eless
ne wo ks a e abundan and whose in as uc u e can be exploi ed. The esul s p esen ed
by he ci ed au ho s a e p omising. The p oblem wi h his equency is i s sho ange, such
ne wo ks a e no longe a ailable on a highway, and due o hei sho ange hey p esen
ce ain p oblems.
In his wo k, he design o a ne wo k sys em based on LoRa wi h a communica ion e-
quency o 868 MHz is p oposed, his allows communica ion o e long dis ances, al hough
in he wo k a sho - ange subne based on LoRa is p oposed, i s capaci y coexis ing and
wo king wi h LoRaWAN means ha hese small subne s can be dis ibu ed and con olled
by a single LoRa Ga eway.
Rega ding he algo i hms in he ea men o he da a, hey use Suppo Vec o Ma-
chines (SVM) [
14
–
16
] and machine lea ning algo i hms such as k-Nea es Neighbo s (k-

Senso s 2021,21, 338 22 o 24
NN) [
13
]. These algo i hms o da a analysis allow hem o classi y he ype o ehicle is
going. The algo i hms p oposed in he a icle a e simple bu jus as e ec i e in de ec ing
he passage o ehicles.
Table 9shown a compa a i e summa y o di e en communica ion p o ocols dis-
cussed in his subsec ion.
Table 9. Compa a i e summa y o di e en communica ion s anda ds.
Classi ica ion
Low Powe Wi eless
Pe sonal A ea Ne wo k
(LWPAN)
Wi eless Access Spaces o
Vehicula En i onmen
(WAVE/DSRC)
Low Powe Wide A ea
Ne wo k (LPWAN)
S anda d IEEE 802.15.4 IEEE 802.11p LoRa
OS No Yes No
Range 10–100 m 100–1000 m 1000–10,000 m
Powe Low Medium Low
High Laye s ZigBee, 6LoWPAN IP 6, WSMP (WAVE Sho
Message P o ocol) Lo aWAN
Modula ion Type BPSK, OQPKS BPSK, QPSK, 16QAM,
64QAM LoRa, FSK
Bi Ra e (Mbps) 0.020 o 0.25 3 o 27 0.003 o 0.050
F equency Bands o
Ope a ion
868 MHz, 915 MHz and 2.4
GHz 5.9 GHz 433 MHz, 868 MHz and 915
MHz
Ne wo k A chi ec u e Pee - o-pee o s a ne wo ks
Pee - o-pee ad hoc ne wo k
in opology and loca ion
based
S a -o -s a s opology in
which ga eways
Wi h he in o ma ion se ou in his subsec ion, we can conclude ha he p oposed
sys em, wi h simple calcula ion algo i hms, has sa is ac o y esul s o he de ec ion o
ehicle a ic, al hough wi h ce ain es ic ions when i complies wi h he Eu opean
egula ion ETSI EN 300 220-2 V3.2.133. These es ic ions mainly a ec when calcula ing
he speed es ima e a speeds abo e 50 km/h, since i is a he limi o wha he subne
sampling ime allows.
4. Conclusions
A new LoRa Ne wo k P o ocol has been de eloped. I is p esen ed in Sec ion 2. This
p o ocol is compa ible wi h LoRaWAN and can li e and wo k wi h his in as uc u e a
he same ime on he de ices. The design allows he de ice o wo k on bo h ne wo ks wi h
he same anscei e .
The ne wo k p o ocol p oposed in his pape wo ks is a Pe sonal A ea Ne wo k (PAN),
so he powe consump ion o wo k in his mode is lowe han o a LoRaWAN; a LoRa
(Long Range) ne wo k is a Low-Powe Wide-A ea Ne wo k (LPWAN) echnology.
The ne wo k has been deployed in a oad and es s ha e been ca ied ou o alida e
he a ic low de ec ion sys em based on Radio Signal S ange Indica o (RSSI) in LoRa
Ne wo ks. The es alida es di e en cases: di e en ehicle speeds, one ehicle on he
oad, wo ehicles in he same way and wo ehicles in opposi e di ec ions.
A powe consump ion analysis has been ca ied ou in Sec ion 2.5. I shows how
he consump ion o he applica ion p oposed in his wo k does no p esen a signi ican
inc ease in a luminous a ic signal.
Two da a p ocessing sys ems ha e been es ed. The i s one is shown in Sec ion 3.1. I
is based on he RSSI de i a i e alue o he communica ion signal compa ed o a h eshold
alue. I is a simple bu e ec i e me hod, al hough i has he limi a ion o no being able o
de ec s opped a ic.
The second me hod is mo e complex, in his case, i is possible o de ec s opped a ic.
Fo i s ope a ion, he communica ion losses a e calcula ed based on he RSSI da a be ween
he nodes whose communica ion is no in e e ed wi h by he low o a ic, nodes #1 and
#2 and nodes #3 and #4.
Senso s 2021,21, 338 23 o 24
The sys em is synch onous, making i possible o ob ain he speed o he ehicles ha
a e a eling h ough he moni o ing a ea. This allows he de elopmen o a ic signs ha
adap hei esponse o he speed o ehicles on a oad.
Field es s ha e been ca ied ou ha ha e allowed an in-dep h analysis o he sys em:
calcula ing he speed o he ehicles in he es s, as well as he e o s in his calcula ion
hanks o ha ing lase ba ie s as e e ence; and he RSSI a enua ion has been analyzed o
he di e en communica ion links.
In all es cases o bo h analysis me hods, he esul s shown in Sec ion 3.3 a e success ul.
LoRa is a p omising echnology in IoT, used o de elop a new WSN o he imp o e-
men o highway sa e y. A slow and s opped a ic de ec ion sys em has been de eloped
based on he measu emen o in ensi y o RF signals, using RSSI. This allows hese haza dous
oad condi ions o be de ec ed when isibili y condi ions a e educed, such as dense og.
I is deduced om he esul s ob ained ha he sys em p oposed in his wo k is
adequa e o he de ec ion o slow o dense ehicle a ic.
In he u u e, he in o ma ion o he in e sec ions (C osses) can be p ocessed o ob ain
he di ec ion o mo emen o he ehicles.
Au ho Con ibu ions:
D.A. (Da id Asiain) designed and implemen ed he sys em and de eloped
he exposed ne wo k p o ocol; D.A. (Da id Asiain) and D.A. (Diego An olín) alida ed he sys em
expe imen ally; D.A. (Da id Asiain) analyzed he esul s and p o ided heo e ical suppo o he
wo k; D.A. (Diego An olín) w o e he pape . All au ho s ha e ead and ag eed o he published
e sion o he manusc ip .
Funding: This esea ch ecei ed no ex e nal unding.
Acknowledgmen s:
This wo k has been suppo ed in pa by he company TECNIVIAL, S.A and i s
pa icipa ion in he inno a i e public pu chase ende app o ed by he Minis y o De elopmen
o he Go e nmen o Spain whose objec i e is o “sea ch o inno a i e solu ions in ela ion o he
design and implemen a ion o og p o ec ion sys ems on he A-8 mo o way be ween Mondoñedo
and A Xes a “(Re : BOE-B-2015-20619). Finally, we app ecia e he collabo a ion o he enginee Se gio
Sánchez Romanillos, head o he R&D depa men o TECNIVIAL.
Con lic s o In e es : The au ho s decla e no con lic o in e es .
Re e ences
1.
Sadowski, S.; Spachos, P. RSSI-Based Indoo Localiza ion wi h he In e ne o Things. IEEE Access
2018
,6, 30149–30161. [C ossRe ]
2.
Bo , M.; Roedig, U. LoRa T ansmission Pa ame e Selec ion. In P oceedings o he 2017 13 h In e na ional Con e ence on
Dis ibu ed Compu ing in Senso Sys ems (DCOSS), O awa, ON, USA, 5–7 June 2017; pp. 27–34. [C ossRe ]
3.
Bo , M.; Vidle , J.E.; Roedig, U. LoRa o he In e ne o Things. In P oceedings o he 2016 In e na ional Con e ence on Embedded
Wi eless Sys ems and Ne wo ks, G az, Aus ia, 15–17 Feb ua y 2016; pp. 361–366.
4.
Pe ajaja i, J.; Mikhaylo , K.; Roi ainen, A.; Hanninen, T.; Pe issalo, M. On he co e age o LPWANs: Range e alua ion and chan-
nel a enua ion model o LoRa echnology. In P oceedings o he 2015 14 h In e na ional Con e ence on ITS Telecommunica ions
(ITST), Copenhagen, Denma k, 2–4 Decembe 2015; pp. 55–59.
5.
Io a, O.; Mu phy, A.L.; Picco, G.P.; Ghi o, L.; Mol eni, D.; Ossi, F.; Cagnacci, F. LoRa om he Ci y o he Moun ains: Explo a ion
o Ha dwa e and En i onmen al Fac o s. In P oceedings o he 2017 In e na ional Con e ence on Embedded Wi eless Sys ems
and Ne wo ks, Uppsala, Sweden, 20–22 Feb ua y 2017; pp. 317–322.
6.
Hsieh, J.-W.; Yu, S.-H.; Chen, Y.-S.; Hu, W.-F. Au oma ic a ic su eillance sys em o ehicle acking and classi ica ion.
IEEE T ans. In ell. T ansp. Sys . 2006,7, 175–187. [C ossRe ]
7.
Xu, Z.; Wei, J.; Chen, X. Vehicle ecogni ion and classi ica ion me hod based on lase scanning poin cloud da a. In P oceedings o
he 2015 In e na ional Con e ence on T anspo a ion In o ma ion and Sa e y (ICTIS), Wuhan, China, 25–28 June 2015; pp. 44–49.
8.
Geo ge, J.; Ma y, L.; Riyas, K.S. Vehicle de ec ion and classi ica ion om acous ic signal using ANN and KNN. In P oceedings
o he 2013 In e na ional Con e ence on Con ol Communica ion and Compu ing (ICCC), Thi u anan hapu am, India, 13–15
Decembe 2013; pp. 436–439.
9.
Tagh aeeyan, S.; Rajamani, R. Po able Roadside Senso s o Vehicle Coun ing, Classi ica ion, and Speed Measu emen . IEEE T ans.
In ell. T ansp. Sys . 2014,15, 73–83. [C ossRe ]
10.
Ma, W.; Xing, D.; McKee, A.; Bajwa, R.; Flo es, C.; Fulle , B.; Va aiya, P. A Wi eless Accele ome e -Based Au oma ic Vehicle
Classi ica ion P o o ype Sys em. IEEE T ans. In ell. T ansp. Sys . 2014,15, 104–111. [C ossRe ]
11.
Wilson, J.; Pa wa i, N. Radio Tomog aphic Imaging wi h Wi eless Ne wo ks. IEEE T ans. Mob. Compu .
2010
,9, 621–632.
[C ossRe ]
Senso s 2021,21, 338 24 o 24
12.
Blum osen, G.; Lu wak, A. Human Body Pa s T acking and Kinema ic Fea u es Assessmen Based on RSSI and Ine ial Senso
Measu emen s. Senso s 2013,13, 11289–11313. [C ossRe ]
13.
Gup a, S.; Hamzin, A.; Degbelo, A. A Low-Cos Open Ha dwa e Sys em o Collec ing T a ic Da a Using Wi-Fi Signal S eng h.
Senso s 2018,18, 3623. [C ossRe ]
14.
Al-Husseiny, A.; Yousse , M. RF-Based T a ic De ec ion and Iden i ica ion. In P oceedings o he 2012 IEEE Vehicula Technology
Con e ence (VTC Fall), Quebec Ci y, QC, Canada, 3–6 Sep embe 2012; pp. 1–5.
15.
Kassem, N.; Kosba, A.E.; Yousse , M. RF-Based Vehicle De ec ion and Speed Es ima ion. In P oceedings o he 2012 IEEE 75 h
Vehicula Technology Con e ence (VTC Sp ing), Yokohama, Japan, 6–9 May 2012; pp. 1–5.
16.
Sen, R.; Mau ya, A.; Raman, B.; Meh a, R.; Kalyana aman, R.; Singh, A. Road-RFSense: A P ac ical RF Sensing–Based Road T a ic
Es ima ion Sys em o De eloping Regions. Acm T ans. Sen. Ne w. 2014,11, 1–32. [C ossRe ]
17.
Ca ani, M.; Boano, C.A.; Röme , K. An Expe imen al E alua ion o he Reliabili y o LoRa Long-Range Low-Powe Wi eless
Communica ion. J. Sens. Ac ua o Ne w. 2017,6, 7. [C ossRe ]
18.
Goldoni, E.; P ando, L.; Vizziello, A.; Sa azzi, P.; Gamba, P. Expe imen al da a se analysis o RSSI-based indoo and ou doo
localiza ion in LoRa ne wo ks. In e ne Technol. Le . 2019,2, e75. [C ossRe ]
19.
Pou sa a , N.; Alahi, M.E.E.; Mukhopadhyay, S. Long- ange wi eless echnologies o IoT applica ions: A e iew. In P oceedings
o he 2017 Ele en h In e na ional Con e ence on Sensing Technology (ICST), Sydney, NSW, Aus alia, 4–6 Decembe 2017; pp.
1–6.
20.
LoRaWAN
®
Speci ica ion 1.0.3|LoRa Alliance
TM
. A ailable online: h ps://lo a-alliance.o g/ esou ce-hub/lo awan -
speci ica ion- 103 (accessed on 5 No embe 2019).
21.
ERC Recommenda ion 70-03. A ailable online: h ps://docdb.cep .o g/download/25c41779-cd6e/Rec7003e.pd (accessed on
5 Janua y 2021).
22.
LoRaWAN
®
Regional Pa ame e s 1.1 B|LoRa Alliance
TM
. A ailable online: h ps://lo a-alliance.o g/ esou ce-hub/lo awan -
egional-pa ame e s- 11 b (accessed on 5 No embe 2019).
23.
LoRaWAN
®
Back-End In e aces 1.0|LoRa Alliance
TM
. A ailable online: h ps://lo a-alliance.o g/ esou ce-hub/lo awan -
back-end-in e aces- 10 (accessed on 5 No embe 2019).
24.
Final D a ETSI EN 300 220-2 V3.2.1. 33. A ailable online: h ps://www.e si.o g/deli e /e si_en/300200_300299/30022002/03.
02.01_30/en_30022002 030201 .pd (accessed on 5 Janua y 2021).
25.
AN1200.22 LoRaTM Modula ion Basics 2015. A ailable online: h ps://sem ech.my.sales o ce.com/s c/p/#E0000000JelG/a/2R0
000001OJk/yDEc AkD9qEz6oG3PJ yoHKas3UMsMDa3TFqz1UQOkM (accessed on 5 Janua y 2021).
26.
SX1276/77/78/79—137 MHz o 1020 MHz Low Powe Long Range T anscei e Da ashee 2020. A ailable online: h ps://sem ech.
my.sales o ce.com/s c/p/#E0000000JelG/a/2R0000001Rb /6E VZUo poKF aF_Fkpgp5kzjiNyiAbqcpqh9qSjE (accessed on
5 Janua y 2021).
27.
Casals, L.; Mi , B.; Vidal, R.; Gomez, C. Modeling he Ene gy Pe o mance o LoRaWAN. Senso s
2017
,17, 2364. [C ossRe ]
[PubMed]
28.
Naga aj, V.; Ki an, V.; Padmaja, K.V. 6LoWPAN based in elligen anspo sys em o su eillance—A model o ubiqui ous
compu ing echnology. In P oceedings o he 2016 Eigh h In e na ional Con e ence on Ubiqui ous and Fu u e Ne wo ks (ICUFN),
Vienna, Aus ia, 5–8 July 2016; pp. 801–803.
29.
A ena, F.; Pau, G.; Se e ino, A. A Re iew on IEEE 802.11p o In elligen T anspo a ion Sys ems. J. Sens. Ac ua o Ne w.
2020
,9,
22. [C ossRe ]