on the use of lorawan for indoor industrial iot...

Research ArticleOn the Use of LoRaWAN for Indoor Industrial IoT Applications

Michele Luvisotto 1 Federico Tramarin 2 Lorenzo Vangelista1 and Stefano Vitturi2

1Department of Information Engineering University of Padova Via Gradenigo 6B 35131 Padova Italy2CNR-IEIIT National Research Council of Italy Via Gradenigo 6B 35131 Padova Italy

Correspondence should be addressed to Federico Tramarin federicotramarinieiitcnrit

Received 29 January 2018 Accepted 12 April 2018 Published 17 May 2018

Academic Editor Vicente Casares-Giner

Copyright copy 2018 Michele Luvisotto et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

Low-Power Wide-Area Networks (LPWANs) have recently emerged as appealing communication systems in the context of theInternet ofThings (IoT) Particularly they proved effective in typical IoT applications such as environmental monitoring and smartmetering Such networks however have a great potential also in the industrial scenario and hence in the context of the IndustrialInternet ofThings (IIoT) which represents a dramatically growing field of application In this paper we focus on a specific LPWANnamely LoRaWAN and provide an assessment of its performance for typical IIoT employments such as those represented by indoorindustrialmonitoring applications In detail after a general description of LoRaWANwediscuss how to set someof its parameters inorder to achieve the best performance in the considered industrial scenario Subsequentlywe present the outcomes of a performanceassessment based on realistic simulations aimed at evaluating the behavior of LoRaWAN for industrial monitoring applicationsMoreover the paper proposes a comparison with the IEEE 802154 network protocol which is often adopted in similar applicationcontextsThe obtained results confirm that LoRaWAN can be considered as a strongly viable opportunity since it is able to providehigh reliability and timeliness while ensuring very low energy consumption

1 Introduction and Motivation

The Internet of things (IoT) concept was introduced towardsthe turn of the century [1] to indicate an interconnectedsystem of uniquely identifiable objects equipped with radiondashfrequency identification (RFID) technology Nowadays theIoT paradigm has expanded embracing a wide range of com-munication technologies software architectures and applica-tions and can be best defined as ldquoa network of networks wheretypically a massive number of objectsthingssensorsdevicesare connected through communication and information infras-tructures to provide valuendashadded servicesrdquo [2] Over the pastdecade several IoT solutions have been developed by bothindustry and academia for various kinds of applicationsincluding smart wearables smart homes and smart cities toname a few [2]

In the next years the IoT vision is expected to be appliednot only to the consumer market but also to productivesectors dramatically changingmanufacturing [3] energy [4]automotive [5] agriculture [6] and other industrial applica-tions in what has already been termed as ldquoIndustrial IoTrdquo

(IIoT) According to a report by the World Economic Forum[7] the IIoT revolution will impact economic sectors thataccount for nearly two-thirds of the global gross domesticproduct changing the basis of competition and redrawingindustry boundaries New connected ecosystemswill emergeallowing significant improvements in operational efficiencyas well as the advent of a new outcome economy wherecompanies no longer deliver products and services but rathermeasurable results that create value for their customers [7]

The deployment of IIoT solutions is a complex processthat as addressed in [8] where a functionalityndashbased archi-tecture for IIoT is proposed impacts on several disciplinessuch as communication and computer science

From a traditional communication aspect IoT encom-passes several heterogeneous systems such as Local AreaNet-works (LANs) Wireless Sensor Networks (WSNs) cellularnetworks mesh and ad hoc networks whose interoperabilityis ensured by the common use of existing Internet protocolssuch as the Internet Protocol version 6 (IPv6) [8]

Moving to the industrial scenario applications oftenhave stringent quality of service (QoS) requirements in

HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 3982646 11 pageshttpsdoiorg10115520183982646

2 Wireless Communications and Mobile Computing

terms of robustness reliability latency determinism energyefficiency and security [9 10] Therefore a careful selectionof the most appropriate network for a specific application isnecessary in order to meet those requirements and provideeffective IIoT solutions Also according to [11] the truepotential of the IIoT paradigm can be unlocked only when awireless communication architecture is envisioned As a con-sequence it is necessary to analyze the suitability of differentwireless networks in view of their deployment in IndustrialIoT applications To this regard several wireless communica-tion systems have been considered for IoT applications [12]They range from very shortndashrange solutions such as nearfield communication (NFC) to extremely longndashrange onessuch as WiMAX and from low-power technologies such asBluetooth Low Energy (BLE) to highndashpower ones as cellularnetworks (2G3G4G) The several amendments to the IEEE80211 standard for Wireless LANs (WLANs) and to theIEEE 802154 standard for Wireless Personal Area Networks(WPANs) are also highly regarded Moreover industrialwireless sensor networks [13 14] and more specificallydedicated wireless networks for industrial applications existsuch as WirelessHART and ISA 10011a and are currentlybeing expanded to guarantee IP support under the 6TiSCHfamily of standards [15]

In addition to the aforementioned solutions Low-PowerWide-Area Networks (LPWANs) have recently emerged inthe IoT scenario [16] themost popular beingNB-IoT SigFoxIngenu Weightless and LoRaWAN These networks that areavailable on licensed as well as unlicensed bands combinea very long communication range (up to several kms) withextremely long battery life at the cost of a limited throughputNowadays LPWANs aremostly used for outdoormonitoringapplications such as environmental monitoring [17] andsmart metering [18] However their features are appealingfor IIoT applications as well and hence they have beenvery recently started to be considered also in this scenario[11 19] Indeed the significantly high energy efficiency ofLPWAN devices can reveal truly interesting for cost-effectiveIIoTdeploymentsMoreover the remarkable communicationrobustness that allows LPWANs to achieve longndashrange com-munications can be useful in industrial applications wherethe wireless channel is often impaired by multipath andfading [20] thus giving them an edge against other low-power wireless technologies

The above considerations represent the main motivationof this paper which basically addresses the use of LPWANsand in particular of LoRaWAN for indoor industrial moni-toring applications and supports this claim with a thoroughperformance assessment To this aim an accurate simulationmodel for LoRaWANnetworks is developed starting from thework in [21] including a realistic channel model for indoorindustrial environments

To the best of the authorsrsquo knowledge this is one ofthe first works that widely investigates the performance ofLoRaWAN for indoor industrial monitoring applicationsIndeed in [22] a case study is presented that addresses aspecific indoor LoRaWAN industrial application whereasin [23] some possible industrial applications of LPWANsare described in general In [24] indoor applications of

LoRaWAN are considered but not for industrial scenariosIn [25] other LPWAN solutions are addressed for IIoT andfinally [17 21 26] deal with the LoRaWAN performance inoutdoor scenarios

In detail the rest of this paper is organized as followsSection 2 introduces the most widespread LPWAN solutionswith a particular focus on LoRaWAN Section 3 describesthe scenario considered in this work namely indoor indus-trial monitoring Section 4 presents a realistic model forLoRaWAN networks employed in industrial environmentsfollowed by a detailed performance analysis in Section 5Finally Section 6 concludes the paper and outlines somefuture directions of research

2 Low-Power Wide-Area Networks

LPWANs are designed to offer affordable connectivity to ahigh number of low-power devices distributed over largegeographical areas In this section the most widespreadLPWAN solutions are discussed with a particular focus onLoRaWAN

21 LoRa and LoRaWAN LoRaWAN is an open networkstandard [27] developed by the LoRa Alliance which mainlydefines the Medium Access Control (MAC) layer and mes-sage formats It is based on LoRa a proprietary physical(PHY) layer developed by Semtech Corporation and derivedfrom chirp spread spectrum (CSS) modulation In this tech-nology each symbol is spread in a fixed bandwidth119861 and thetime duration of the symbol is varied according to an indexcalled spreading factor (SF) which can range between 7 and12 Consequently the duration of a symbol varies from 1119861 times27 and 1119861 times 212 This spreading technique allows recoveringdata even when the received power is very low (also underthe noise level) thus offering very robust communication atthe cost of a reduced data rate [28] which does not exceed547 Kbps Moreover transmissions with different spreadingfactors are somewhat orthogonal to each other increasingnetwork capacity as better discussed in Section 4

LoRaWAN networks are deployed in the unlicensed in-dustrial scientific andmedical (ISM) bands of 863ndash870MHzin Europe and 902ndash928MHz in the US According to theregulations in these bands the transmitting devices mustlimit their maximum power to 14 dBm (27 dBm in the8694ndash86965MHz subndashband) and adopt either a dutyndashcycledtransmission (01 1 or 10 percent according to the subndashband)or a listenndashbeforendashtalkndashadaptivendashfrequencyndashagility (LBTAFA) behavior

A LoRaWAN network includes three types of entitiesnamely End Devices (EDs) Gateways (GWs) and NetworkServer (NS) EDs are typical field devices that collect sensorinformation from the field and possibly send commandsThey are connected (via wireless links) to one or moreGWs that in turn are connected (either through a wiredor cellular backhaul link) to a single NS which managesthe entire network and originates downlink transmissions (ifany)There is no exclusive association between EDs andGWsand the same uplink message can be received by several GWswith different signal qualities

Wireless Communications and Mobile Computing 3

The LoRaWAN specifications define three functionalclasses namely Classes A B and C with the first one beingmandatory for all LoRaWAN end devices (EDs) Class AEDs access the channel in a random fashion following anALOHAndashlike scheme and open (at most) two receptionwindows at predefined slots in time and frequency after eachuplink transmission whereas they remain in sleep mode forthe rest of the time Classes B and C devices differentiatemostly for their management of receive windows class BEDs can open them at scheduled time intervals (they aresynchronized with the NS by means of beacon messagesbroadcasted by the GWs) while class C ones keep themalways open clearly sacrificing energy efficiency for lowlatency Finally authentication and encryption mechanismsat different levels (device network and application) are envi-sioned by LoRaWAN specifications to ensure the integrityand security of communications

22 Other LPWAN Solutions Besides LoRaWAN severalother LPWAN technologies are available in both licensed andunlicensed bands

The former includes mainly NarrowBand IoT (NB-IoT) which is part of 3GPP Release 13 NB-IoT reusesthe LTE design adopting Orthogonal FrequencyndashDivisionMultiple Access (OFDMA) for downlink and SinglendashCarrierFrequencyndashDivision Multiple Access (SC-FDMA) for uplinktransmissions with resource blocks of 15 kHz times 05ms anda maximum bandwidth of 180 kHz offering a peak data rateof 250Kbps NB-IoT can be deployed inndashband using LTEresources in the guard band between two LTE bands or asstandndashalone by replacing a 200 kHz GSM carrier

In the unlicensed band the most widespread LPWANsbesides LoRaWAN are SigFox Ingenu and Weightless Thefirst one is a proprietary protocol based on ultra narrowband(UNB)modulation with data rate limited to 100 bps in uplinkand 600 bps in downlink SigFox works in the same bands asLoRaWAN and limits the operation of the connected devicesto 140 uplink messages and 4 downlink messages per dayIngenu is also a proprietary technology which works in the24GHz band and adopts a patented Random Phase MultipleAccess (RPMA) scheme for uplink transmissions with amaximum data rate of 78Kbps on 40 different 1MHz widechannels each of which can host up to 1200 orthogonal sig-nals thanks to RPMA Finally theWeightless Special InterestGroup (SIG) proposes three different standards (Weightless-W Weightless-N and Weightless-P) deployed in differentbands (TV white space and subndashGHz ISM bands) offeringdifferent data rates (from 200 bps to 10 Mbps) and employingdifferent modulation schemes (UNB Quadrature Ampli-tude Modulation (QAM) and Phase-Shift Keying (PSK) anddifferent channel access methods (ALOHA TimendashDivisionMultiple Access (TDMA) and FrequencyndashDivision MultipleAccess (FDMA))

Although all these LPWANs can reveal interesting in thispaper we focus exclusively on LoRaWAN since it is a reallypromising and popular network with interesting featuresParticularly (i) it operates in an unlicensed band (ii) itsMAClayer protocol is completely open and (iii) it can handle anunlimited number of packets

Centralserver

Cloud

SinkEnd node

Wireless linkBackhaul link

L P

r

Figure 1 A schematic representation of the indoor industrialmonitoring scenario considered in this paper

3 The Indoor Industrial Monitoring Scenario

The reference scenario considered in this paper is representedin Figure 1 As can be seen it refers to a monitoring networkcomposed by 119873 devices (end nodes) deployed in a buildingwhere an industrial process is taking place The devices aredistributed within a circular area of radius 119903 and periodicallysample different physical quantities that allowmonitoring thestate of the process Each node sends the updated samplevalue as a message of 119871 bytes with a transmission period of 119875seconds to a sink node installed at the center of the buildingwhich will be either a GW in a LoRaWAN network or moregenerally a PANcoordinator in aWPANThe sinkwill in turnsend the data received from the end nodes to a central serverThis configuration resembles that of industrial wireless sensornetworks deployed in monitoring systems [29 30] Clearlymore complex configurations could be addressed using thistype of networks For example different applications inwhichmultiple sinks are used could be envisioned where thenodes can transmit packets of different length with differentperiods However this paper focuses on a basic applicationof LoRaWAN to indoor IIoT to investigate its effectivenessleaving the assessment of more complicated scenarios tofutureworks For the same reasonwe assume that only uplinktransmissions are performed and that they are not confirmedmeaning that acknowledgement packets are not sent by thesink nodes and hence there are no retransmissions

Table 1 summarizes the values of the parameters con-sidered in this paper We assume that the number of nodesin the network ranges from 10 to 1000 while the coverageradius is fixed to 200 meters Since sensor measurementstypically occupy a few bytes the message length in theseapplications is generally low with 50 bytes being a reasonablevalue considering that also other information (timestampnode status battery life etc) may be appended Finallythe transmission period depends on the dynamics of thesampled quantities and in this paper ranges fromoneminuteto 30 minutes It is worth noting that the correspondingsampling rates are considerably lower than those encounteredin other industrial applications (which can be up to somekHz) since this scenario is targeted at the onndashlinemonitoringof industrial processes [31] not aiming at realndashtime control


Table 1 Parameters for the indoor industrial monitoring scenario

Parameter Description Value119873 Number of nodes 10 1000119903 Coverage radius 200m119871 Message length 50 Bytes119875 Transmission period 60 1800 sThe assessment carried out in this paper is based on

three key performance indicators The first one related tothe communication reliability is the probability of success(PoS) that is the percentage of packets sent by the endnodes which are received correctly by the sink A secondimportant metric related to the latency and determinism ofthe communication is the interndashpacket time (IPT) namelythe interval between two consecutively received packets atthe sink pertaining to the same node More formally if weindicate with 119903119894(119896) the time at which the sink receives the 119896thpacket from end node 119894 119894 = 1 119873 the IPT for node 119894 is thefollowing varying quantity

IPT119894 (119896) = 119903119894 (119896 + 1) minus 119903119894 (119896) (1)

Considering an observation period during which the sinkhas received 119870119894 packets from node 119894 the mean IPT can becomputed as

MIPT119894 (119896) = 1119870119894 minus 1119870119894minus1sum119896=1

IPT119894 (119896) (2)

Then the global IPT (GIPT) can be computed averaging theMIPT over all the nodes in the network

The third and last performance metric is the averageenergy consumed (AEC) by each end node which providesinsights on the energy efficiency and allows forecasting thebattery life of end nodes

4 A Realistic LoRaWAN Industrial Model

While it is possible to find accurate models of LoRaWANnetworks in the scientific literature [21] none of them tacklesthe peculiar features of industrial environments In thissection a realistic model for LoRaWAN networks deployedin IIoT applications is hence presented

41 Channel Model In this work indoor LoRaWAN net-works deployed in the 863ndash870MHz ISM band will beconsidered although the channel model is also applicableto the 902ndash928MHz ISM band adopted in US The modelshould take into account all the impairments that can bepresent inside industrial buildings which can be divided intotwo categories namely large-scale effects and smallndashscaleeffects

Large-scale effects include path loss and shadowing sothat the total power loss 119871(119889) is a function of the distance 119889between transmitter and receiver It can be expressed (in dB)as

119871 (119889)dB = PL (119889)dB + 120594dB120590 (3)

where the shadowing term 120594dB120590 is generally modeled as azerondashmean Gaussian random variable with standard devia-tion 120590 The path loss term PL(119889)dB instead is often modeledas a fixed term plus a logarithmic function of the distance 119889multiplied by a coefficient 120578 (path loss exponent) whose valueranges between 2 in ideal conditions (ie linendashofndashsight andfree space) and 4 in non-line-of-sight (NLOS) conditions Acommon approach is to define a breakpoint distance referredto as 1198891 beyond which the propagation becomes NLOSyielding the following path loss model

PL (119889)dB

=

0 119889 lt 1198890PL0 + 101205780 log( 1198891198890) 1198890 le 119889 le 1198891PL0 + 101205780 log(11988911198890) + 101205781 log(

1198891198891) 119889 ge 1198891(4)

where the parameters PL0 1198890 and 1205780 together with theshadowing standard deviation 120590 can be extracted from [32]which reports real-world measurements in the 900MHzband from different indoor industrial environments In thiswork the breakpoint distance 1198891 is set to 100m (half ofthe coverage radius) and the path loss exponent in NLOSconditions is set to the typical value 1205781 = 4

The smallndashscale effects mostly refer to fading which istypically modeled through a RayleighRician distributionwhose only parameter is the 119870ndashfactor 119870 The work in [33]reports some 119870ndashfactor measurements in indoor industrialenvironments at 868MHz whose results have been used forthe simulations presented in this paper

42 Link PerformanceModel Taking into account all channelimpairments the received power at the sink in dB can behence modeled as

119875dBmrx = 119875dBm

tx minus 119871dB minus 119865dB + 119866dBtx + 119866dB

rx (5)

where119875dBtx is the transmitted power in dB119866dB

tx and119866dBrx are the

transmitreceive gains in dB respectively 119865dB is a margin thataccounts for fading and 119871dB is the total loss due to large-scaleeffects reported in (3) The signal-to-noise ratio (SNR) canbe immediately derived from (5) by subtracting the thermalnoise power in dB According to LoRa chipset specifications[34] for each selected spreading factor 119896 a minimum SNRlevel119877119896 is requested to achieve a correct demodulationThesevalues are reported in Table 2 together with the data rate ofeach SF and the time required to transmit a 50 bytes message(assuming 125 kHz bandwidth and 45 code rate)

Besides path loss shadowing and fading the other sig-nificant impairment inwireless channels is interference LoRaCSS modulation is quite robust to external interference fromnon-LoRa signals [28] whereas the interference betweendifferent LoRa transmissions strongly depends on their SFIn [28] it is reported the required signal-to-interference(SIR) ratio in dB 119879119896119897 to allow a correct decoding of atransmission with SF 119896 when an another transmission withSF 119897 is interfering However since in typical LoRaWAN


Table 2 Characteristics of LoRa spreading factors

SF Required SNR Data rate TX time for 50 B7 minus75 dB 547Kbps 9958ms8 minus10 dB 313 Kbps 17869ms9 minus125 dB 176Kbps 33690ms10 minus15 dB 098Kbps 63283ms11 minus175 dB 044Kbps 118374ms12 minus20 dB 025 Kbps 22036ms

networks several concurring transmissions with different SFsare present evaluating the impact of each interfering SFsmaybe computationally inefficient Consequently an approachbased on equivalent signal-to-interference plus noise ratio(ESINR) is detailed in the following

Consider a transmission with SF 119896 and an interferingtransmission with SF 119897 where the SIR is SIRdB

119896119897 A correctdecoding is achieved if SIRdB

119896119897 ge 119879119896119897 and also if the SNRis above 119877119896 It can be hence said that a SIR level of 119879119896119897 isldquoequivalentrdquo to a SNR level of 119877119896 from a system performanceperspective This allows transforming any SIR value withrespect to an interfering SF 119897 in an equivalent SIR (ESIR) valueas

ESIRdB119896119897 = SIRdB

119896119897 + 119877119896 minus 119879119896119897 = SIRdB119896119897 + 119864119896119897 (6)

where the matrix E (in dB) is obtained from the values 119877119896 inTable 2 and the matrix 119879 in [28] as

E =[[[[[[[[[[[[

minus135 85 105 115 115 12514 minus16 10 12 12 12145 145 minus185 105 125 12515 15 15 minus21 11 13155 155 155 155 minus235 11516 16 16 16 16 minus26

]]]]]]]]]]]]

(7)

In other words (6) allows ldquonormalizingrdquo the SIR between twointerfering transmissions with arbitrary spreading factorstransforming it in a quantity comparable with other SNRvalues and hence allowing to sum them together This isachieved through the term 119864119896119897 which represents a sort ofweight indicating the impact of the interfering SF 119897 on thetransmission with SF 119896 It is worth observing that this impactis much higher if the two SFs are the same and becomes lessrelevant as the two SFs are distant thus allowing concurrenttransmissions to take place

With the proposed model based on ESI the packeterror rate (PER) of a transmission performed with SF 119896can be evaluated First a single transmission is divided intoldquochunksrdquo each characterized by a specific set of interferersThe bit error rate (BER) during each chunk can be computedthrough the following operations

(1) The received power119875dBmrx is computed according to (5)

and the SNR is derived(2) The SIR for each interfering SF 119897 is derived and the

corresponding ESIR is computed according to (6)

SF 7SF 8SF 9

SF 10SF 11SF 12

minus30 minus25 minus20 minus15 minus5minus10minus35ESINR (dB)

10minus3

10minus2

10minus1

100

BER

Figure 2 BER versus ESINR (in dB) curves for different LoRaspreading factors obtained through Matlab simulations

(3) The ESINR in dB is computed as

ESINRdB119896 = minus10 log10(10minusSNRdB

119896 10 + 12sum119897=7

10minusESIRdB119896119897 10) (8)

(4) The BER corresponding to the ESINR is derivedaccording to the curves in Figure 2 which are ob-tained through Matlab simulations of CSS perfor-mance

Starting from the BER and the number of bits in thechunk which can be approximated from the chunk durationand the data rate the chunk error rate (CER) is obtainedFinally the PER can be derived as

PER119896 = 1 minus119873cksprod119899=1

(1 minus CER119896119899) (9)

where 119873cks is the number of chunks that make up the trans-mission

43 Strategies for the Choice of the Spreading Factors One ofthe degrees of freedom in configuring a LoRaWAN networkis to assign to each node its SF In this paper we addresssome techniques that allow a static assignment of the SFs asdescribed below It is worth mentioning however that thestandard also allows for a dynamic strategy called adaptivedata rate (ADR) which is not considered in this papersince it requires downlink transmissions Nonetheless thetechniques that are going to be presented can be easilyextended to the dynamic case

The simplest approaches are either to assign the same SFto all the nodes in the network or to randomly distributeall the available SFs among them A more refined approach


often adopted in other studies [21] is to assign to each nodethe lowest SF for which the SNR at the sink is higher thanthe threshold defined in Table 2 However as discussed in[35] when the maximum distance between nodes and sink islimited (as in the case of indoor environments) this strategywill always lead to all the nodes being assigned the lowestSF that is 7 (which ensures the highest data rate) thusincurring in fairness problems and not fully exploiting theorthogonality of the different LoRa SFs

To overcome this issue an innovative strategy for theselection of SFs is proposed in this paper based on aconstrained optimization procedure Let S be the set ofavailable SFs and 119904119894 the SF assigned to node 119894 This procedureneeds to fulfill two constraints

(a) The aforementioned constraint on the SNR that is

SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)

(b) An additional constraint on the transmit period 119875that is

119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)

where TX(119904119894) is the transmission time for SF 119894 asreported in Table 2 and DC119894 is the duty cycle lim-itation of node 119894 which depends on the operatingfrequency band [36]

After ensuring that these constraints are observed theSFs are distributed in the most uniform possible way amongthe nodes to maximize the orthogonality of transmissionsMore formally let N119896 be the set of nodes for which SF 119896is assigned and let 119873min and 119873max be the minimum andmaximum cardinality of the setsN119896 that is

119873min = min119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)

For each node a SF is chosen among the ones that respectthe constraints of both (10) and (11) so that the differencebetween the minimum and maximum cardinality is mini-mized

argmin 1003816100381610038161003816119873max minus 119873min1003816100381610038161003816 (13)

This strategy which is indicated as ldquofairrdquo in the follow-ing section ensures that orthogonality of transmissions ismaximized and that the nodes never exceed the duty cyclelimitations

44 Energy Model A fundamental aspect to address in aLoRaWANnetwork is the energy consumedby the endnodesA LoRa ED can be in four possible states [34] idle standbytransmitting and receiving Specifically a node is always inidle state except when it transmits a packet and during theautomatically opened receive windows where it can be either

Table 3 Supply current for LoRa EDs in different states

State Symbol ValueIdle 119868idle 15 120583AStandby 119868standby 14mATx 119868tx 28mARx 119868rx 112mA

standby or receiving (not in the case of this paper since thereare no downlink transmissions)

Table 3 reports the current consumed by a LoRa ED in thefour different states as reported in [34] In order to computethe energy consumption of each ED the current of each statehas been multiplied by the supply voltage (119881DC = 33V)and by the time that the ED passes in that specific stateFor example the energy consumed by a single node duringa transmission period to transmit a payload of 50 B withspreading factor 119904 is

EC = 119881DC sdot 119868idle sdot [119875 minus TX (119904) minus 119879RX1 minus 119879RX2] + 119881DC

sdot 119868tx sdot TX (119904) + 119881DC sdot 119868rx (119879RX1 + 119879RX2) (14)

where119879RX1 and119879RX2 are the durations of the first and secondreceive windows respectively

5 Performance Evaluation

The LoRaWAN industrial model presented in Section 4 hasbeen implemented in the popular ns3 network simulator [37]in order to assess the performance of this network in the IIoTscenario

51 Simulations Setup The starting point for the develop-ment of ns3 LoRaWAN simulations was the work in [21]in which an original LoRa module for ns3 was presentedallowing accurate simulation of uplink transmission in aLoRaWAN network Several features have been integratedto this module to allow realistic simulations of LoRaWANnetworks employed in IIoT applications Specifically

(i) An accurate channel model for indoor industrialbuildings has been introduced that accounts for pathloss shadowing and fading Realistic channel modelparameters were selected considering the experimen-tal measurements reported in [32 33]

(ii) The simplified link performance model of [21] hasbeen expanded adding the ESINRndashbased approachdetailed in Section 42 as well as the MatlabndashbasedBER versus ESINR curves

(iii) A different interference model with respect to [21]has been used in the old model the power of apartially interfering transmission was ldquoequalizedrdquo onthe entire packet duration whereas as discussed inSection 4 in the adopted model a packet is dividedinto chunks each one with a specific set of interferersand corresponding error rate


SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess

Figure 3 PoS versus number of EDs in a LoRaWAN networkemployed in IIoT applications for different SF choice strategies Thetransmission period is fixed to 10 minutes

(iv) An energy model has been added to [21] allowing aprecise computation of the energy consumed by theend nodes

The performance figures of LoRaWAN have been alsocompared with those of aWPAN which have been simulatedexploiting the lr-wpan module included in the standardns3 distribution upgraded with some modifications Inparticular the 868MHz binary phase-shift keying (BPSK)PHY layer introduced in the IEEE 802154-2006 standard[38] was considered in order to provide a fair comparisonwith LoRaWAN deployed in the same frequency band Thischoice limited the achievable data rate of IEEE 802154to 20Kbps Moreover the modifications proposed in [39]mostly concerning the introduction of an energymodel havebeen considered

In all the simulations the reference scenario is thatreported in Figure 1 with the parameters shown in Table 1For each choice of parameters the results were averaged over10 different runs each simulating the network performancefor 2 hours

52 Tuning of a LoRaWAN Network A first assessment isconcerned with the comparison of different strategies forthe assignment of SFs in a LoRaWAN network used forindoor industrial monitoring applications Four differentstrategies defined in agreement with the analysis carried outin Section 43 have been considered namely two constantstrategies in which all the nodes are assigned either the lowestspreading factor (SF 7) or the highest (SF 12) a randomstrategy in which SFs are randomly assigned to the nodes theinnovative strategy described in Section 43 called ldquofairrdquo

Figure 3 reports the PoS computed over all the trans-missions in the network for the different SF assignmentstrategies and for different networks sizes ranging from 10

SF 7FairFair modified

400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess

Figure 4 Detailed comparison of different SF choice strategies in aLoRaWAN constant to SF 7 fair selection on the complete list of SFsand fair selection on a restricted list (SF 7 8 and 9)

to 1000 nodes The transmission period for all nodes isfixed to 10 minutes It can be observed that first the PoSis generally high confirming the good robustness of LoRamodulation The worst behavior is obtained when all nodesare assigned SF 12 especially as the network size increasessince the transmission time with this SF is quite long (morethan 2 seconds per packet) and causes severe interferenceproblems Conversely when all nodes are assigned SF 7 thePoS is almost constant to a very high value (about 97)performing even better (when the number of nodes in thenetwork is high) than the ldquofairrdquo strategy which instead yieldsa slightly higher PoS when there are less than 200 nodesThisresult can be explained observing that SF 7 guarantees thelowest transmission time thus minimizing the interferenceconversely the ldquofairrdquo strategy leverages the orthogonalityof different SFs but distributes all the available SF values(including the highest ones) among the nodes increasingthe transmission times and hence ultimately increasing theprobability a transmission is subjected to interference

To overcome this issue a slightly modified version of thefair strategy is proposed in which the set of available SFsS islimited to the three lowest ones (7 8 and 9)The performanceof this upgraded scheme can be observed in Figure 4 alwaysrelevant to PoS versus number of EDs with 119875 = 10 minutesThe modified fair strategy now outperforms the strategy inwhich all nodes are assigned SF 7 both when the number ofnodes is low and when it is high never falling below a PoSof 965 As a result of these considerations in the followingof this paper the only considered schemes for the selectionof SFs in LoRaWAN are the ldquofair modifiedrdquo strategy and theldquoconstant to SF 7rdquo one

53 Comparison of LoRaWAN and IEEE 802154 To supportthe claim that LoRaWANcan represent a good choice for IIoT


LoRaWAN SF 7LoRaWAN Fair ModIEEE 802154

400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess

Figure 5 PoS versus number of nodes for LoRaWAN and IEEE802154 employed in IIoT applications The transmission period isfixed to 10 minutes

monitoring applications the performance of this network iscompared with that of an IEEE 802154ndashbased one

The configuration of the IEEE 802154 network is equiva-lent to that of LoRaWANMoreover the nonndashbeacon versionof the IEEE 802154 protocol has been considered in whichnodes (that generate new packets with period 119875) accessthe channel in a random fashion following a carrier sensemultiple access (CSMA) algorithm

The first comparison shown in Figure 5 reports thePoS for different network sizes and a transmission periodfixed to 10 minutes It can be observed that both LoRaWANstrategies (ldquoconstant to SF 7rdquo and ldquofair modifiedrdquo) slightlyoutperform IEEE 802154This is due to the better robustnessof LoRa modulation with respect to the IEEE 802154 oneindeed comparing the derivations reported in the standard[38] with the curves in Figure 2 IEEE 802154 needs almosta 10 dB higher SNR to offer the same BER as the less robustLoRa SF (the BER of IEEE 802154 with the 868MHz BPSKPHY has been derived from the standard [38]) Moreover inLoRaWAN the collision probability is reduced with respect toIEEE 802154 because of the orthogonality between differentSFs

A second metric considered in the comparison betweenLoRaWAN and IEEE 802154 is the GIPT namely the timeelapsed between two consecutive correct packets receivedat the sink averaged over all the nodes in the networkA performance assessment under the same conditions ofFigure 5 (119873 ranging from 10 to 1000 and 119875 = 10 minutes)is provided in Figure 6 It can be observed that all valuesare quite close to the transmission period as expectedeven if again IEEE 802154 performs slightly worse thanLoRaWAN for every network size This is due to both thelower robustness of IEEE 802154 modulation and the higherrandomness of its channel access scheme Among the two


400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (

s)Figure 6 GIPT versus number of nodes for LoRaWAN and IEEE802154 employed in IIoT applications The transmission period isfixed to 10 minutes

0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3

400 600 800 1000200Transmission period (s)

Figure 7 Average energy consumed in a 2-hours simulation versustransmission period for LoRaWAN and IEEE 802154 employed inIIoT applications The number of nodes is fixed to 500 The detailedplot allows to better distinguish the performance of the two SFchoice strategies adopted in LoRaWAN

LoRaWAN strategies the ldquofair modifiedrdquo strategy confirms tobe the best one offering a slightly lower GIPT

Finally the performance figures of LoRaWAN and IEEE802154 are compared in terms of the average energy con-sumed by each node in Figure 7 In this case the numberof nodes is fixed to 500 and the transmission period isvaried from 1 to 30 minutes First it can be observed thatthe average energy consumed by IEEE 802154 nodes is


roughly 100 times higher than that of LoRaWAN nodeswhich stay most of the time in deep sleep state and hencehave an extremely low-power consumption To this regardit is worth observing that the IEEE 802154 protocol hasbeen implemented according to its legacy version withoutadopting any specify power saving strategy Clearly theintroduction of (purposely defined) strategies of this typecould definitely lead to better performance in terms of powerconsumption [40]

The detailed plot in Figure 7 allows also comparing thetwo LoRaWAN strategies for SF choice ldquoconstant to SF 7rdquoand ldquofair modifiedrdquo showing that the former scheme allowsconsuming less energy This is motivated by the fact thatthe energy consumed is directly proportional to the amountof time spent in TX state which drains the highest currentas reported in Table 3 and SF 7 offers the lowest possibletransmission time However it has to be considered that asshown in Figure 4 setting the spreading factors of all nodesto 7 also yields a higher probability of error Consequently ifretransmissions were activated the additional attempts nec-essary to transmit corrupted packets would certainly increasethe energy consumption In any case it is worth observingthat from the simulations carried out the estimation of thebattery lifetime for an ED results in about 2 years for a simple1000mAh battery

It has to be noted that the nonbeacon version of IEEE802154 considered in this section is only one of themany options provided by this standard which recentlydefined new operating modes specifically tailored for themonitoring and control of industrial processes such asthe Time-Slotted Channel Hopping (TSCH) mode of IEEE802154e [41] These enhancements allow achieving a muchimproved timeliness reliability and energy consumptionwith respect to the nonbeacon version of the standard andare definitely to be preferred when performing control ofcritical processes Nonetheless inmany noncritical industrialmonitoring applications such as the one discussed in thispaper where sampling rates are not very high LoRaWAN canprovide a cost-effective and easy-to-deploy alternative whichstill guarantees very high reliability and ultralow energyconsumption as proven by the simulation results

6 Conclusions

This paper addressed the adoption of LoRaWAN for indoorindustrial monitoring systems which represent an interest-ing IIoT field of application After an accurate theoreticalanalysis a realistic simulation model of LoRaWAN has beendeveloped that allowed investigating the behavior of networkconfigurations typically deployed for industrial monitoringThe obtained results showed very good performance interms of reliability timeliness and energy consumptionParticularly a newly introduced technique for the selectionof the spreading factor revealed able to outperform othertraditionally adopted techniques A comparison with a con-figuration of an equivalentWPAN namely IEEE 802154 hasalso been carried out and provided encouraging results

Several future activities can be envisioned as follow-upsof this work First the occurrence of downlink transmissions

has to be appropriately investigated in order to evaluate theperformance of LoRaWAN in presence of retransmissionsand adaptive data rate strategies It would be also importantto model Class B LoRa devices as they support synchroniza-tion through beacons and hence may allow developing ascheduled channel access method (eg TDMA) which canimprove timeliness Moreover in LoRaWAN networks theEDs do not have direct Internet connectivity so that they canbe only remotely accessed throughGWs To this regard someproposals to integrate IPv6 over LoRaWANhave been alreadydeveloped [42] their feasibility and impact on networkperformance however need to be carefully investigated

Finally a further step of analysis is represented by the exe-cution of experimental sessions on real LoRaWAN testbedsthat would allow tuning the accuracy of the theoretical andsimulation analyses presented in this work

Data Availability

We have not yet made the data publicly available becausesome parts of the code are protected

Conflicts of Interest

The authors declare that they have no conflicts of interest

References

[1] K Ashton ldquoThat rsquoInternet of Thingsrsquo Thingrdquo httpwwwrfid-journalcomarticlesview4986

[2] C Perera CH Liu and S Jayawardena ldquoThe emerging internetof things marketplace from an industrial perspective a surveyrdquoIEEE Transactions on Emerging Topics in Computing vol 3 no4 pp 585ndash598 2015

[3] F Tao Y Zuo L D Xu and L Zhang ldquoIoT-based intelligentperception and access of manufacturing resource toward cloudmanufacturingrdquo IEEE Transactions on Industrial Informaticsvol 10 no 2 pp 1547ndash1557 2014

[4] S Jacobsson and A Bergek ldquoTransforming the energy sectorThe evolution of technological systems in renewable energytechnologyrdquo Industrial and Corporate Change vol 13 no 5 pp815ndash849 2004

[5] L L Bello ldquoNovel trends in automotive networks A perspectiveonEthernet and the IEEEAudioVideoBridgingrdquo inProceedingsof the 2014 IEEE Emerging Technology and Factory Automation(ETFA) pp 1ndash8 Barcelona Spain September 2014

[6] F TongKe ldquoSmart Agriculture Based on Cloud Computing andIOTrdquo Journal of Convergence Information Technology vol 8 no2 pp 210ndash216 2013

[7] A Hassanzadeh S Modi and S Mulchandani ldquoTowardseffective security control assignment in the Industrial Internetof Thingsrdquo in Proceedings of the 2015 IEEE 2nd World Forumon Internet of Things (WF-IoT) pp 795ndash800 Milan ItalyDecember 2015

[8] L D Xu W He and S Li ldquoInternet of things in industries asurveyrdquo IEEE Transactions on Industrial Informatics vol 10 no4 pp 2233ndash2243 2014

[9] T Sauter ldquoThe three generations of field-level networks - Evolu-tion and compatibility issuesrdquo IEEE Transactions on IndustrialElectronics vol 57 no 11 pp 3585ndash3595 2010


[10] E Toscano and L Lo Bello ldquoA topology management protocolwith bounded delay for Wireless Sensor Networksrdquo in Pro-ceedings of the Factory Automation (ETFA 2008) pp 942ndash951Hamburg Germany September 2008

[11] S Mumtaz A Alsohaily Z Pang A Rayes K F Tsang andJ Rodriguez ldquoMassive Internet of Things for Industrial Appli-cations AddressingWireless IIoT Connectivity Challenges andEcosystem Fragmentationrdquo IEEE Industrial Electronics Maga-zine vol 11 no 1 pp 28ndash33 2017

[12] S Tayeb S Latifi and Y Kim ldquoA survey on IoT communicationand computation frameworks An industrial perspectiverdquo inProceedings of the 7th IEEE Annual Computing and Communi-cation Workshop and Conference CCWC 2017 January 2017

[13] V C Gungor and G P Hancke ldquoIndustrial wireless sensor net-works challenges design principles and technical approachesrdquoIEEE Transactions on Industrial Electronics vol 56 no 10 pp4258ndash4265 2009

[14] K Beom-Su v HoSung K K Hoon D Godfrey and K Ki-Il ldquoA survey on realndashtime communications in wireless sensornetworksrdquo Wireless Communications and Mobile Computingvol 2017 Article ID 1864847 14 pages 2017

[15] T Watteyne V Handziski X Vilajosana et al ldquoIndustrialWireless IP-Based Cyber-Physical Systemsrdquo Proceedings of theIEEE vol 104 no 5 pp 1025ndash1038 2016

[16] U Raza P Kulkarni andM Sooriyabandara ldquoLow PowerWideArea Networks An Overviewrdquo IEEE Communications Surveysamp Tutorials vol 19 no 2 pp 855ndash873 2017

[17] W Guibene J Nowack N Chalikias K Fitzgibbon M Kellyand D Prendergast ldquoEvaluation of LPWAN Technologies forSmart Cities River Monitoring Use-Caserdquo in Proceedings of the2017 IEEEWireless Communications andNetworking ConferenceWorkshops WCNCW 2017 March 2017

[18] N Varsier and J Schwoerer ldquoCapacity limits of LoRaWANtechnology for smart metering applicationsrdquo in Proceedings ofthe 2017 IEEE International Conference on CommunicationsICC 2017 fra May 2017

[19] R Sanchez-Iborra and M-D Cano ldquoState of the art in LP-WAN solutions for industrial IoT servicesrdquo Sensors vol 16 no5 article no 708 2016

[20] A Willig K Matheus and A Wolisz ldquoWireless technology inindustrial networksrdquo Proceedings of the IEEE vol 93 no 6 pp1130ndash1151 2005

[21] DMagrinM Centenaro and L Vangelista ldquoPerformance eval-uation of LoRa networks in a smart city scenariordquo inProceedingsof the 2017 IEEE International Conference on CommunicationsICC 2017 fra May 2017

[22] J Haxhibeqiri A Karaagac F Van den Abeele W JosephI Moerman and J Hoebeke ldquoLoRa indoor coverage andperformance in an industrial environment Case studyrdquo inProceedings of the 2017 22nd IEEE International Conference onEmerging Technologies and Factory Automation (ETFA) pp 1ndash8 Limassol September 2017

[23] G Margelis R Piechocki D Kaleshi and P Thomas ldquoLowThroughput Networks for the IoT Lessons learned from indus-trial implementationsrdquo in Proceedings of the 2nd IEEE WorldForum on Internet of Things WF-IoT 2015 pp 181ndash186 itaDecember 2015

[24] P Neumann J Montavont and T Noel ldquoIndoor deploymentof low-power wide area networks (LPWAN) A LoRaWANcase studyrdquo in Proceedings of the 2016 IEEE 12th InternationalConference on Wireless and Mobile Computing Networking and

Communications (WiMob) pp 1ndash8 New York NY October2016

[25] D M Hernandez G Peralta L Manero R Gomez J Bilbaoand C Zubia ldquoEnergy and coverage study of LPWAN schemesfor Industry 40rdquo in Proceedings of the 2017 IEEE InternationalWorkshop of Electronics ControlMeasurement Signals and theirApplication to Mechatronics ECMSM 2017 esp May 2017

[26] J Petajajarvi K Mikhaylov R Yasmin M Hamalainen andJ Iinatti ldquoEvaluation of LoRa LPWAN Technology for IndoorRemote Health and Wellbeing Monitoringrdquo International Jour-nal of Wireless Information Networks vol 24 no 2 pp 153ndash1652017

[27] N Sornin M Luis T Eirich T Kramp and O Hersent ldquoLoRaAlliance LoRaWAN specificationrdquo in LoRaWAN SpecificationRelease v10 2015

[28] C Goursaud and J M Gorce ldquoDedicated networks for IoTPHY MAC state of the art and challengesrdquo EAI EndorsedTransactions on Internet of Things vol 1 no 1 p 150597 2015

[29] G P Hancke and V C Gungor ldquoGuest editorial special sectionon industrial wireless sensor networksrdquo IEEE Transactions onIndustrial Informatics vol 10 no 1 pp 762ndash765 2014

[30] L Lo Bello J Akerberg M Gidlund and E Uhlemann ldquoGuestEditorial Special Section on New Perspectives on WirelessCommunications in Automation From Industrial Monitoringand Control to Cyber-Physical Systemsrdquo IEEE Transactions onIndustrial Informatics vol 13 no 3 pp 1393ndash1396 2017

[31] Z Iqbal K Kim and H-N Lee ldquoA cooperative wireless sensornetwork for indoor industrial monitoringrdquo IEEE Transactionson Industrial Informatics vol 13 no 2 pp 482ndash491 2017

[32] Y Ai M Cheffena and Q Li ldquoRadio frequency measurementsand capacity analysis for industrial indoor environmentsrdquo inProceedings of the 9th European Conference on Antennas andPropagation (EuCAP rsquo15) April 2015

[33] J Ferrer-Coll P Angskog C Elofsson J Chilo and P Stenum-gaard ldquoAntenna cross correlation and ricean K -factormeasure-ments in indoor industrial environments at 433 and 868 MHzrdquoWireless Personal Communications vol 73 no 3 pp 587ndash5932013

[34] ldquoSX127273 860 MHz to 1020 MHz Low Power Long RangeTransceiverrdquo httpwwwsemtechcomimagesdatasheetsx1272pdf

[35] B Reynders W Meert and S Pollin ldquoPower and spreadingfactor control in low power wide area networksrdquo in Proceedingsof the 2017 IEEE International Conference on CommunicationsICC 2017 fra May 2017

[36] ldquoE ETSI ldquo300 220-1 (v2 41)rdquo Electromagnetic compatibilityand Radio spectrumMatters (ERM)rdquo 2012

[37] ldquoThe Network Simulator - ns-3rdquo httpwwwnsnamorg[38] ldquoIEEE Standard for Information Technology - Telecommuni-

cations and Information Exchange Between Systems - LocalandMetropolitan Area Networks - Specific Requirements - Part154rdquo in Wireless Medium Access Control (MAC) and PhysicalLayer (PHY) Specifications for Low Rate Wireless Personal AreaNetworks (LR-WPANs) IEEE Std 2006

[39] V Rege and T Pecorella ldquoA Realistic MAC and Energy Modelfor 802154rdquo in Proceedings of the the Workshop pp 79ndash84Seattle WA USA June 2016

[40] A El-Hoiydi and J Decotignie ldquoWiseMAC an ultra low powerMAC protocol for the downlink of infrastructureWireless Sen-sor networksrdquo inProceedings of the Proceedings 9th InternationalSymposium on Computers and Communications (ISCC rsquo04) pp244ndash251 July 2004


[41] D De Guglielmo S Brienza and G Anastasi ldquoIEEE 802154eA surveyrdquo Computer Communications vol 88 pp 1ndash24 2016

[42] P Weber D Jackle D Rahusen and A Sikora ldquoIPv6 overLoRaWANrdquo in Proceedings of the 3rd IEEE InternationalSymposium on Wireless Systems within the IEEE InternationalConferences on Intelligent Data Acquisition and Advanced Com-puting Systems IDAACS-SWS 2016 pp 75ndash79 deu September2016

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


terms of robustness reliability latency determinism energyefficiency and security [9 10] Therefore a careful selectionof the most appropriate network for a specific application isnecessary in order to meet those requirements and provideeffective IIoT solutions Also according to [11] the truepotential of the IIoT paradigm can be unlocked only when awireless communication architecture is envisioned As a con-sequence it is necessary to analyze the suitability of differentwireless networks in view of their deployment in IndustrialIoT applications To this regard several wireless communica-tion systems have been considered for IoT applications [12]They range from very shortndashrange solutions such as nearfield communication (NFC) to extremely longndashrange onessuch as WiMAX and from low-power technologies such asBluetooth Low Energy (BLE) to highndashpower ones as cellularnetworks (2G3G4G) The several amendments to the IEEE80211 standard for Wireless LANs (WLANs) and to theIEEE 802154 standard for Wireless Personal Area Networks(WPANs) are also highly regarded Moreover industrialwireless sensor networks [13 14] and more specificallydedicated wireless networks for industrial applications existsuch as WirelessHART and ISA 10011a and are currentlybeing expanded to guarantee IP support under the 6TiSCHfamily of standards [15]

In addition to the aforementioned solutions Low-PowerWide-Area Networks (LPWANs) have recently emerged inthe IoT scenario [16] themost popular beingNB-IoT SigFoxIngenu Weightless and LoRaWAN These networks that areavailable on licensed as well as unlicensed bands combinea very long communication range (up to several kms) withextremely long battery life at the cost of a limited throughputNowadays LPWANs aremostly used for outdoormonitoringapplications such as environmental monitoring [17] andsmart metering [18] However their features are appealingfor IIoT applications as well and hence they have beenvery recently started to be considered also in this scenario[11 19] Indeed the significantly high energy efficiency ofLPWAN devices can reveal truly interesting for cost-effectiveIIoTdeploymentsMoreover the remarkable communicationrobustness that allows LPWANs to achieve longndashrange com-munications can be useful in industrial applications wherethe wireless channel is often impaired by multipath andfading [20] thus giving them an edge against other low-power wireless technologies

The above considerations represent the main motivationof this paper which basically addresses the use of LPWANsand in particular of LoRaWAN for indoor industrial moni-toring applications and supports this claim with a thoroughperformance assessment To this aim an accurate simulationmodel for LoRaWANnetworks is developed starting from thework in [21] including a realistic channel model for indoorindustrial environments

To the best of the authorsrsquo knowledge this is one ofthe first works that widely investigates the performance ofLoRaWAN for indoor industrial monitoring applicationsIndeed in [22] a case study is presented that addresses aspecific indoor LoRaWAN industrial application whereasin [23] some possible industrial applications of LPWANsare described in general In [24] indoor applications of

LoRaWAN are considered but not for industrial scenariosIn [25] other LPWAN solutions are addressed for IIoT andfinally [17 21 26] deal with the LoRaWAN performance inoutdoor scenarios

In detail the rest of this paper is organized as followsSection 2 introduces the most widespread LPWAN solutionswith a particular focus on LoRaWAN Section 3 describesthe scenario considered in this work namely indoor indus-trial monitoring Section 4 presents a realistic model forLoRaWAN networks employed in industrial environmentsfollowed by a detailed performance analysis in Section 5Finally Section 6 concludes the paper and outlines somefuture directions of research

2 Low-Power Wide-Area Networks

LPWANs are designed to offer affordable connectivity to ahigh number of low-power devices distributed over largegeographical areas In this section the most widespreadLPWAN solutions are discussed with a particular focus onLoRaWAN

21 LoRa and LoRaWAN LoRaWAN is an open networkstandard [27] developed by the LoRa Alliance which mainlydefines the Medium Access Control (MAC) layer and mes-sage formats It is based on LoRa a proprietary physical(PHY) layer developed by Semtech Corporation and derivedfrom chirp spread spectrum (CSS) modulation In this tech-nology each symbol is spread in a fixed bandwidth119861 and thetime duration of the symbol is varied according to an indexcalled spreading factor (SF) which can range between 7 and12 Consequently the duration of a symbol varies from 1119861 times27 and 1119861 times 212 This spreading technique allows recoveringdata even when the received power is very low (also underthe noise level) thus offering very robust communication atthe cost of a reduced data rate [28] which does not exceed547 Kbps Moreover transmissions with different spreadingfactors are somewhat orthogonal to each other increasingnetwork capacity as better discussed in Section 4

LoRaWAN networks are deployed in the unlicensed in-dustrial scientific andmedical (ISM) bands of 863ndash870MHzin Europe and 902ndash928MHz in the US According to theregulations in these bands the transmitting devices mustlimit their maximum power to 14 dBm (27 dBm in the8694ndash86965MHz subndashband) and adopt either a dutyndashcycledtransmission (01 1 or 10 percent according to the subndashband)or a listenndashbeforendashtalkndashadaptivendashfrequencyndashagility (LBTAFA) behavior

A LoRaWAN network includes three types of entitiesnamely End Devices (EDs) Gateways (GWs) and NetworkServer (NS) EDs are typical field devices that collect sensorinformation from the field and possibly send commandsThey are connected (via wireless links) to one or moreGWs that in turn are connected (either through a wiredor cellular backhaul link) to a single NS which managesthe entire network and originates downlink transmissions (ifany)There is no exclusive association between EDs andGWsand the same uplink message can be received by several GWswith different signal qualities







Centralserver

Cloud

SinkEnd node


L P

r









IPT119894 (119896) = 119903119894 (119896 + 1) minus 119903119894 (119896) (1)



IPT119894 (119896) (2)







119871 (119889)dB = PL (119889)dB + 120594dB120590 (3)


PL (119889)dB

=


1198891198891) 119889 ge 1198891(4)




119875dBmrx = 119875dBm


rx (5)













119896119897 + 119877119896 minus 119879119896119897 = SIRdB119896119897 + 119864119896119897 (6)


E =[[[[[[[[[[[[


]]]]]]]]]]]]

(7)






SF 7SF 8SF 9

SF 10SF 11SF 12


10minus3

10minus2

10minus1

100

BER




119896 10 + 12sum119897=7

10minusESIRdB119896119897 10) (8)




(1 minus CER119896119899) (9)








SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)


119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)




1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)



















SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia








Centralserver

Cloud

SinkEnd node


L P

r









IPT119894 (119896) = 119903119894 (119896 + 1) minus 119903119894 (119896) (1)



IPT119894 (119896) (2)







119871 (119889)dB = PL (119889)dB + 120594dB120590 (3)


PL (119889)dB

=


1198891198891) 119889 ge 1198891(4)




119875dBmrx = 119875dBm


rx (5)













119896119897 + 119877119896 minus 119879119896119897 = SIRdB119896119897 + 119864119896119897 (6)


E =[[[[[[[[[[[[


]]]]]]]]]]]]

(7)






SF 7SF 8SF 9

SF 10SF 11SF 12


10minus3

10minus2

10minus1

100

BER




119896 10 + 12sum119897=7

10minusESIRdB119896119897 10) (8)




(1 minus CER119896119899) (9)








SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)


119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)




1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)



















SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






IPT119894 (119896) = 119903119894 (119896 + 1) minus 119903119894 (119896) (1)



IPT119894 (119896) (2)







119871 (119889)dB = PL (119889)dB + 120594dB120590 (3)


PL (119889)dB

=


1198891198891) 119889 ge 1198891(4)




119875dBmrx = 119875dBm


rx (5)













119896119897 + 119877119896 minus 119879119896119897 = SIRdB119896119897 + 119864119896119897 (6)


E =[[[[[[[[[[[[


]]]]]]]]]]]]

(7)






SF 7SF 8SF 9

SF 10SF 11SF 12


10minus3

10minus2

10minus1

100

BER




119896 10 + 12sum119897=7

10minusESIRdB119896119897 10) (8)




(1 minus CER119896119899) (9)








SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)


119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)




1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)



















SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia










119896119897 + 119877119896 minus 119879119896119897 = SIRdB119896119897 + 119864119896119897 (6)


E =[[[[[[[[[[[[


]]]]]]]]]]]]

(7)






SF 7SF 8SF 9

SF 10SF 11SF 12


10minus3

10minus2

10minus1

100

BER




119896 10 + 12sum119897=7

10minusESIRdB119896119897 10) (8)




(1 minus CER119896119899) (9)








SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)


119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)




1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)



















SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






SNRdB119894 ge 119877119904119894 119894 = 1 119873 (10)


119875 ge TX (119904119894)DC119894

119894 = 1 119873 (11)




1003816100381610038161003816N1198961003816100381610038161003816 119873max = max

119896isinS

1003816100381610038161003816N1198961003816100381610038161003816 (12)



















SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



SF 7SF 12

RandomFair

400 600 800 1000200Number of EDs

082

084

086

088

09

092

094

096

098

Prob

abili

ty o

f Suc

cess








400 600 800 1000200Number of EDs

0945

095

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




400 600 800 1000200Number of EDs

0955

096

0965

097

0975

098

Prob

abili

ty o

f Suc

cess







400 600 800 1000200Number of EDs

610

612

614

616

618

620

622

624

Glo

bal i

nter

pack

et ti

me (


0

50

100

150

200

250

300

Aver

age e

nerg

y co

nsum

ed (J

)


200 400 600 800 10000

05

1

15

2

25

3









6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






6 Conclusions





Data Availability




References
















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia








































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


on the use of lorawan for indoor industrial iot...

Documents