Extending LoRaEnergySim Simulator to Support Interference

Management under Multi-Gateway IoT Scenarios

Daniele Stumpo, Floriano De Rango and Francesco Buffone

DIMES, Universit

a della Calabria, Via P. Bucci, Arcavacata di Rende(CS), Italy

Keywords:

Simulator, LoRa, LoRaEnergySim, Multi-Gateway, LPWAN.

Abstract:

Internet of Things (IoT) is gaining more impact on our lives and has been increasingly used. It allows several

wireless devices to be connected, and their distance can range from a few inches to many miles. New IoT

technologies such as LoRa are emerging allowing energy efﬁcient wireless communication over exceptionally

long distances. So, it is particularly important to evaluate its performance through simulations. At this purpose,

it is possible to ﬁnd several tools and simulators for LoRa technology. All of them present distinctive features

and are written with different programming languages with the possibility to enable and disable different

features. In this paper it is analysed in detail the Simulator LoRaEnergySim, which allows the creation of

a network with only one Gateway (GTW). In addition, the simulator is extended considering the case of

Multi-GTW presence with very high IoT node density and considering all interference aspects that can be

related to this new scenario. Besides, the extended simulator, considers now the imperfect orthogonality of

the Spreading Factor (SF) not fully supported in the previous simulator version with the aim to consider more

realistic simulations.

1 INTRODUCTION

In recent years, the IoT paradigm has gaining a lot

of attention, with a continuous rise of new emerg-

ing technologies, especially in the LPWAN domain.

One of these is LoRa, whose characteristics are a

wide range of coverage (even up to 10km), a low bi-

trate, sporadic communication, low duty cycle (DC)

and high scalability. The most important feature is

the low energy consumption; this due to the Chirp

Spread Spectrum (CSS) and the low sensitivity that

has a LoRa receiver. This led the scientiﬁc commu-

nity to create new tools and simulators to evaluate the

performance of this technology. Some tools evaluate

performance at physical level, then, there are several

simulators that implement different features, but cur-

rently there is not one that meets all the needs of a

LoRa network; in fact, no simulator implements all

End Device (ED) classes; most of them implement

only class A devices, and only a few of them imple-

ment class C. In addition, there are simulators that al-

low the creation of networks with only one GTW, and

only a few allows Multi-GTW. Moreover, imperfect

orthogonality of the SF or the energy consumption are

not always considered.

Creating a simulator that implements all the features

of communication and network standards would led

to a greater knowledge about the considered technol-

ogy and provide a much more valuable tool for cre-

ating simulations closer to reality. So, the aim of this

paper is the implementation of additional simulation

modules on an open simulator to allow researchers to

test the LoRa technology in a more effective way.

The paper is organized as follows: Lora technology

are presented in section II; some related works are in-

troduced in section III; in section IV LoRaEnergySim

simulator is explained and in section V some exten-

sions in LoRaEnergySim are explained; ﬁnally, con-

clusions are summarized in section VI.

Table 1: Table of acronyms.

Terms Acronym

Spreading Factor SF

End Device ED

Gateway GTW

Network Server NS

Application Server AS

Duty Cycle DC

Adaptive Data Rate ADR

Packet pkt

Chirp Spread Spectrum CSS

364

Stumpo, D., De Rango, F. and Buffone, F.

Extending LoRaEnergySim Simulator to Support Interference Management under Multi-Gateway IoT Scenarios.

DOI: 10.5220/0011319600003274

In Proceedings of the 12th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2022), pages 364-371

ISBN: 978-989-758-578-4; ISSN: 2184-2841

2 LoRa

LoRa technology is a wireless technology based on

the Industrial-Medical-Scientiﬁc (ISM) band, it is

used for applications that need great coverage, low

power consumption and low bitrate. This technol-

ogy is based on CSS, which has been used for sev-

eral years in military ﬁeld, thanks to the robustness to

interference, but LoRa is the ﬁrst technology that has

implemented it for commercial use. LoRaWAN was

designed and developed by the LoRa Alliance. The

development started from the bottom in order to opti-

mize Low Power Wide Area Networks (LPWANs),

thus, optimizing battery life, wide coverage range,

scalability, reliability and cost (Farrell, 2018). LoRa

and LoRaWAN terms are used interchangeable but,

actually, they have different meanings:

• When talking about ’LoRa’ it refers to the physi-

cal layer or wireless modulation.

• The term ’LoRaWAN’ refers to the speciﬁc proto-

col that was designed based on LoRa technology

by the LoRa Alliance.

The advantages of LoRa are many, but perhaps the

most notable is the wide coverage, in fact, a single

GTW has the ability to cover an entire city or 100km

(10km x 10km). Obviously, the coverage area de-

pends a lot on the environment and the various obsta-

cles present in the area. The LoRa link budget is one

of the biggest in comparison with all other standard-

ized communication technologies, in fact it is about

155dB and thanks to this it is possible to have a wide

range of coverage in an environment.This technology

is suitable for LPWANs, so it is not used in applica-

tions where a large amount of data is required, such

as video streaming, but it is mainly used for transfer-

ring data acquired from a sensor or diagnostic infor-

mation related to networked devices and transmitted

infrequently (Zourmand et al., 2019).

2.1 LoRa PHY

LoRa’s physical layer is a proprietary Spread Spec-

trum modulation scheme from Semtech. This scheme

is derived from the CSS modulation; this type of mod-

ulation manages to change the data rate for sensitivity

within a ﬁxed channel bandwidth.

This is possible due the use of orthogonal SFs, which

can affect the data rate. Applying a speciﬁc SF, it is

possible to change the data rate inﬂuencing the trans-

mission distance or the power. This allows the per-

formance of a network to be optimized at a constant

bandwidth. In addition, Semtech has made only an

implementation of the PHY layer that is independent

from the upper layers; this is a positive factor because

LoRa can coexist and interoperate with existing net-

work architectures. The CSS of LoRa is derived from

Direct Sequence Spread Spectrum (DSSS), in which

the carrier phase of the signal changes according to a

sequence of codes. The signal transmitted by LoRa

occupies a bandwidth typically 125KHz, 250KHz or

500KHz and depending on different geographical re-

gions, a speciﬁc frequency must be used. In addition,

radio technologies that transmit in the ISM band have

regulated limitations, in case of LoRa, in the 868MHz

band, used in Europe, there is a DC of 1%, instead as

regards the transmission power you can use a maxi-

mum power of 14 dBm and only for the uplink Chan-

nel it can be used a DC of 10%.

2.2 LoRaWAN-architecture

LoRaWAN is a low-power, wide area networking pro-

tocol built on top of the LoRa radio modulation tech-

nique. It deals with how the network is organized, its

structure, how to access the medium, what to do when

a message does not arrive. Access to the medium

takes place via the ALOHA protocol, i.e., in a com-

pletely random way, which is in fact the main cause

of the limited DC. LoRa architecture (see Fig. 1) has

four main elements: ED, GTW, Network Server (NS)

and Application Server (AS).

ED: can be a sensors that sends the collected data in

Broadcast towards the GTWs or actuators that wait to

receive a speciﬁc message. In particular, we can iden-

tify three types of EDs, that differ mainly in how they

receive downlink packet (pkt):

• Class A: it has receiver window only after trans-

mitting a pkt; this type of class must be imple-

mented in all LoRa devices and is the least energy-

consuming class.

• Class B: it has a periodic receiver window that is

synchronized based on a beacon sent by the GTW.

• Class C: it is always listening except when send-

ing a pkt.

Figure 1: LoRa architecture.

Extending LoRaEnergySim Simulator to Support Interference Management under Multi-Gateway IoT Scenarios

365

GTW: Can be deﬁned as a concentrator of informa-

tion, it is connected to each ED of the network and

its main task is to forward the received pkts to the NS

through another technology, such as Ethernet, Cellu-

lar Networks, Wi-Fi etc., In addition a GTW to with-

stand a large network must have good reception ca-

pacity. It has, in fact, antennas with a very low sen-

sitivity, and it must be able to handle a huge number

of messages; this is due to the possibility to simulta-

neously listen on multiple channels thanks to the or-

thogonal codes of the SF and then it must also have

very good computational qualities to be fast to decode

messages that are received with different SF.

NS: is an element that analyses the data coming from

the network. The data/pkts coming from the different

GTWs are ﬁltered, security checks are performed on

the individual pkts received and then if all goes well

the ACKs are generated. If necessary, the NS applies

the Adaptive Data Rate (ADR) to decrease/increase

transmission power or applying different communi-

cation channel through speciﬁc SF. If the NS receives

the same pkt from multiple GTWs, it can handle the

duplication and discards each copy. Another task of

the NS, which is also the only element that can deﬁne

from whom the data arrives, can check if the received

data is destined for a speciﬁc AS and forwards to it.

AS: has the task of processing application-speciﬁc

data messages received from the EDs. It also gen-

erates all downlink payloads of the application layer

and sends them to the EDs connected through the NS.

A LoRaWAN network can have more than one AS.

3 RELATED WORKS

The simulation of LoRa networks is particularly im-

portant because it can be used for designing and eval-

uating LoRa-based applications without the need of

expensive deployments. Choosing the right system

parameters such as the SF can improve the power con-

sumption of wireless devices which is the main chal-

lenge in LoRa, as well as in IoT (Bouras et al., 2020).

Moreover, using optimized parameters can improve

several network performances as well as energy con-

sumption. Therefore, to extend the network lifetime

the most important goal is to make sure that as less

collisions as possible occur and that the pkts arrive

with enough power towards the Base Station. Thus,

pkts can be correctly interpreted and that there will

be no need to retransmit them wasting energy. This

will be possible by choosing the right combination of

SF, frequency and bandwidth. This will reduce the

transmission time, increase the data rate and already

improve the energy consumption of IoT devices.

3.1 Existing simulators

Currently, it is possible to ﬁnd several LoRa simu-

lators online, using different programming languages

but also implementing various parts of the standard.

The main ones are: FLoRa (FLoRa, nd) (Slab-

icki et al., 2018); PhySimulator (PhySimulator,

nd); LoRaWAN Module for ns-3 (Reynders et al.,

2018); LoRaSim (LoRaSim, 2017); LoRaFREE

(Abdelfadeel et al., 2020); LoRaWAN simulator

(Marini et al., 2021); LoRaEnergySim (LoRaEner-

gySim, nd).

3.1.1 FLoRa

FLoRa is a simulation framework that uses the OM-

Net++ discrete event simulator; it is based on the

Framework of INET and is written in C++. It en-

ables to create LoRa-ED, GTWs and a NS. In addi-

tion, its modules aim to simulate the physical layer

and the LoRaWAN MAC protocol. It also provides

an excellent graphical interface. It offers an accu-

rate model for the LoRa PHY layer and also provides

statistics about the network power consumption. It

also gives the option to simulate a network with 1 or

more GTWs (Slabicki et al., 2018).

3.1.2 PhySimulator

This simulator is written in MATLAB, but it imple-

ments only the link level of LoRa. Through this sim-

ulator it is not possible to build a network of LoRa,

but its purpose is only to evaluate the reception of two

overlapping and interfering LoRa transmissions that

have been modulated with different SF. In each exe-

cution, a reference frame modulated with a reference

ref

is added with a random number of interfering

symbols modulated with a different SF

int

. The resul-

tant overlapped signal is demodulated in absence of

noise, assuming a perfect receiver synchronization.

Finally, it will be generated ﬁgures where you can see

the interference generated due to non-orthogonality of

SF, and, moreover, Packet Error Rate (PER), Symbol

Error Rate (SER) and Bit Error Rate (BER) of the ex-

periment.

3.1.3 LoRaWAN Module for NS-3

This is not a simulator itself, but it is a module to be

integrated in NS-3. The module is written in C++ and

Python and the implementation is compliant with Lo-

RaWAN 1.0 class A. This last one is the most energy

efﬁcient of the LoRa devices. Also, in addition to the

two lower layers, PHY and MAC, the Transport layer

and Application layer have been implemented so that

SIMULTECH 2022 - 12th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

366

as many parameters as possible can be conﬁgured.

In order to keep track of the energy consumption an

additional class has been implemented, whose name

is LoRaRadioEnergyModel (Reynders et al., 2018).

3.1.4 LoRaSim

LoRaSim is a discrete event simulator written in

python2.7 that uses several libraries, such as SimPy,

which is a library that provides an environment for

making simulations. This simulator offers the chance

to run four different network conﬁgurations. Two of

them are done with a single GTW and two with up

to four GTWs. It is possible to use either omnidi-

rectional or directional antenna modules. Then, it is

allowed to set different parameters as the number of

EDs, the number of GTWs, the distance between two

GTWs, the number of LoRa networks and the simula-

tion time. Other two features that can be set is the gen-

eration of a graph that shows the conﬁguration of the

network and how to control the collisions, in a sim-

pliﬁed way, then controlling only SF, Freq and time

or in a complete way also controlling the power and

then using the capture effect (Bouras et al., 2020) (Bor

et al., 2016). The log-distance path loss model is im-

plemented as radio propagation model. This simula-

tor does not track the energy consumption of individ-

ual devices, but the energy consumption of the entire

network, so deriving the amount of energy spent by

individual devices using different SFs would be difﬁ-

cult. In order to start a simulation, one has to create

a conﬁguration ﬁle, in which the different simulation

parameters are speciﬁed, such as, the number of EDs,

how the pkt generation takes place, the pkt size, etc.

Regarding the generation of pkts, it can be set in three

different ways, i. e., periodic generation, random gen-

eration, or following a exponential model. At the end

of the simulation there will be created a ﬁle in which

the PDR for each application and the total energy con-

sumption will be reported.

3.1.5 LoRaFREE

In this simulator, besides the uplink communication

(ED-GTW), the downlink communication has also

been implemented (to respond with an acknowledge-

ment (ACK) from the GTW to the ED). In addition to

this, it also implements imperfect orthogonality of the

SF. The simulator allows only periodic communica-

tion by the EDs and the propagation model used is the

log-distance path loss model as in LoRaSim; more-

over, in comparison to LoRaSim, the energy con-

sumption module has been extended, as it also con-

siders the energy consumed when receiving feedback

from the GTW to the ED (Abdelfadeel et al., 2020).

3.1.6 LoRaWAN Simulator

LoRaWAN simulator is entirely written in MatLab

and implements a complete LoRa transceiver for the

PHY layer which is responsible for signal generation,

modulation and demodulation. A MAC layer that

is responsible for multiple channel access and takes

care of mutual interference and supporting more than

one GTW is also implemented. The simulator sup-

ports a number of features, such as both uplink and

downlink communications, imperfect SF orthogonal-

ity, capture effect, both uplink and downlink interfer-

ence, DC limitations, A and C class devices. At the

end of the simulation, there are several performance

metrics such as the delivery rate for both uplink and

downlink and the energy consumption values for each

device and for the entire network (Marini et al., 2021)

3.1.7 LoRaEnergySim

LoRaEnergySim is a simulator developed in Python3

and uses the SimPy library, and it is a cross-layer sim-

ulation framework to analyse the energy consump-

tion of devices in a realistic way. In addition, the

ADR mechanism is implemented in the simulator

and downlink messages have also been implemented.

This has a signiﬁcant impact on the number of data

extraction rate. The payload size, transmission fre-

quency can be set easily. Power consumption will be

done and any state changes of the ED, such as, idle,

active, sleep are represented (Callebaut et al., 2019).

4 LoRaEnergySim

After analyzing all the different simulators that were

found online, it is chosen to work with the Lo-

RaEnergySim simulator (the code can be found in

https://github.com/GillesC/LoRaEnergySim). This

simulator is chosen because:

• It uses Python as its Programming languages, with

the following beneﬁts:

– Portable, therefore it is easy to install on differ-

ent platforms such as Linux, Windows, etc

– Easy to use because the syntax and the differ-

ent modules and functions that are already in-

cluded in the language are consistent, intuitive,

and easy to learn.

– Rich in Libraries, so if it is necessary to imple-

ment new objects it can always take a look on

the Python Package Index.

– Efﬁcient, so it guarantees high performance.

• It has already implemented a module on the en-

ergy consumption of the single device.

Extending LoRaEnergySim Simulator to Support Interference Management under Multi-Gateway IoT Scenarios

367

• Uses a log-distance path loss model, that can

be easily changed by modifying the simulation

model.

• It has an easily modiﬁable structure because it is

structured in different classes.

The simulator provides a generic framework for eval-

uating and co-optimizing PHY and MAC parameters

under realistic network conditions. Each component

of the simulator is built on the basis of individual

components. It works in a very simple way: there

is a class A ED that sends its pkts to the AirInterface;

over this interface some models for the signal prop-

agation were implemented and it is possible to eval-

uate the SNR and the collision; if the pkt passes all

the physical requirements it arrives at the GTW that

processes it, but only if it arrives with enough power

to be interpreted. After this step, if the ED waits for a

conﬁrmation, the GTW will send an ACK message to

conﬁrm the received message.

4.1 Simulator Structure

The Simulator-Structure is summarized in Figure 2.

4.1.1 LoRaWAN ED Class A

Each ED is characterized by an energy proﬁle, a set

of LoRa parameters, such as SF and frequency, and a

location, given by the class Location. The energy pro-

ﬁle used in the simulator is as in the Figure 3 (Calle-

baut et al., 2019).

From the ﬁgure it can be seen that there are mainly

three peaks of energy consumption. The ﬁrst corre-

sponds to the transmission of a pkt, which has a du-

ration equal to the Time on Air (ToA). The other two

peaks correspond to the two reception windows, the

ﬁrst, RX1, is when the ED listens on the same fre-

quency of the uplink message for a duration that de-

pends on the SF used during the transmission of the

pkt, while in the second, RX2, it listens on the down-

link frequency. If the message should be received al-

ready in the ﬁrst reception window, the ED will not

use the second slot.

Figure 2: LoRaEnergySim Structure.

Figure 3: Energy Proﬁle of a LoRaWAN ED Class A.

4.1.2 AirInterface

This module has mainly three tasks: calculating the

path loss, translating the received signal strength

(RSSI) into an SNR value and ﬁnally determining a

pkt collision. These tasks are managed through:

• Propagation Model: uses the log-distance path

loss model with shadowing, and it is used to

model the signal when it is transmitted from the

ED to the GTW.

• SNR model: adds background noise to the trans-

mitted signal.

• Collision Detection Task: pkts collide if they will

be transmitted with same frequency, same SF, are

time-overlapped. Finally, the power difference

between the two pkts is evaluated and if the dif-

ference is less than power threshold (e.g. 6dB)

pkts collide, otherwise the most power pkt will be

correctly interpreted.

4.1.3 Gateway

The GTW implemented on the Simulator is capable of

simultaneously reading 8 pkts with different parame-

ters (SF and Freq). Furthermore, the signal must not

have collided, and the signal strength must be higher

than the Antenna sensitivity. If a communication has

been set up where messages must be acknowledged,

then the GTW itself creates an ACK message. Cur-

rently it is assumed that the downlink message always

reaches the ED as the GTWs have a higher allowed

transmission power. Also, there is the possibility of

only creating networks with a single GTW.

5 Extended LoRaEnergySim

Based on the LoRaEnergySim Simulator, new exten-

sions that allow the simulator to increase its ﬂexibility,

therefore increasing the number of simulation scenar-

ios have been led out. Another important thing is the

introduction of the imperfect orthogonality of the SFs.

5.1 Changes Made to Simulator

In order to bring the simulator closer to the real world,

several changes have been made:

SIMULTECH 2022 - 12th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

368

Table 2: SIR Threshold between SFs.

int

→ 7 8 9 10 11 12

ref

↓

7 1 -8 -9 -9 -9 -9

8 -11 1 -11 -12 -13 -13

9 -15 -13 1 -13 -14 -15

10 -19 -18 -17 1 -17 -18

11 -22 -22 -21 -20 1 -20

12 -25 -25 -25 -24 -23 1

• From Single-GTW to Multi-GTW.

• ACK Management with ideal NS.

• Different pkt rates on different EDs.

• Piggybacking on ACK, to communicate parame-

ters to ED, like ADR.

• Display the network topology based on parame-

ters at the end of the simulation.

• Addition of new stats such as pkts latency, col-

lided pkts, the number of totals sent and created

pkts and so on.

• New parameters for Antenna Sensitivity (Antenna

Sensitivity, 2019)

• Introduction of imperfect orthogonality (Croce

et al., 2018)

5.1.1 From Single-gateway to Multi-gateway

The ﬁrst change was the extension into a Multi-GTW

scenario, supporting the creation of a network with

1 to 9 GTWs. They were positioned more or less

balanced and equidistant, but could be repositioned

differently easily. The important consideration at this

stage was how to handle the interference created on all

GTWs. So, for each GTW, it will be created a propri-

etary AirInterface where the communication channel

of the single GTW is managed. However, it is very

important that every time an ED wants to communi-

cate a pkt, it must insert the pkt on all the AirInter-

faces, so it is possible to calculate on all the various

GTWs how the pkt arrives, if it does not collide and

if it arrives with enough power. In addition, as the

LoRa standard is deﬁned, the pkt can be read by sev-

eral GTWs at the same time, since the communication

is in Broadcast.

5.1.2 Imperfect SF-orthogonality

Another added feature is the control of imperfect SF

orthogonality, in fact, collisions between pkts modu-

lated with different SFs can also occur if the received

interference power is strong enough. This contradicts

the common belief that SFs can be considered orthog-

onal. The results of (Croce et al., 2018) show that

intra-SF collisions are indeed a problem in LoRa net-

works. Also, Table 2 was extracted from (Croce et al.,

2018), and inserted into the simulator.

Originally, the simulator checked for four conditions

to see if there was a collision, as mentioned earlier.

After applying this upgrade, the behavior of the simu-

lator now is ﬁrstly the check of two conditions: if two

EDs use the same frequency and if there is a temporal

overlap in the communication; then it will be consid-

ered the Table 2, it will check simultaneously SF and

power level.

To make clearer the whole behavior an example is as

follows: let us suppose there are two pkts in the com-

munication channel with the same frequency and a

time overlap, with the following SF and power val-

ues:

• PKT

→ SF=8, P

=-102dBm

• PKT

→ SF=12, P

=-129dBm

Now it will be seen what happens when these 2

pkts arrive at the GTW. Starting to verify PKT

from table 2 we taking to account the cell referring

to SF

ref

=8 and SF

int

=12, which indicates a power

threshold of 13dB. This means that if the interfering

PKT

presents a power greater than 13dB compared

to PKT

, then the last one can not be interpreted. Here

below we can see how it works:

• SF

REF

→ 8, SF

INT

→ 12

Inter f erence > −13 > NoInter f erence

Tx(SF

REF

)

− P

Tx(SF

INT

)

= −102 − (−129) = +27

The calculation just performed implies that PKT

can

be interpreted. The same process has to be taken into

account for PKT

. Referring again to table 2, it must

be considered SF

ref

=12 and SF

int

=8. Now, if the in-

terfering PKT

presents a power greater then 25dB

compared to PKT

, then the last one can not be inter-

preted, in fact, in this case the difference is:

• SF

REF

→ 12, SF

INT

→ 8

Inter f erence > −25 > NoInter f erence

Tx(SF

REF

)

− P

Tx(SF

INT

)

= −129 − (−102) = −27

Since in this case the interfering signal is 27dB greater

than the referenced Signal, it can be concluded that

PKT

can not be interpreted.

5.2 ACK Management with NS

Each time an ED must receive an ACK it has been as-

sumed that it can only be sent by the nearest GTW;

if the pkt does not arrive at the nearest GTW, because

it experienced a collision or because it arrived at the

nearest GTW with too low power, but the pkt arrives

at a more distant and visible GTWs, then the NS will

Extending LoRaEnergySim Simulator to Support Interference Management under Multi-Gateway IoT Scenarios

369

inform all GTW about the packer reception and send-

ing back the ACK to the sender ED.

To manage this, it is necessary to analyse what is hap-

pening with the pkt on all the AirInterfaces, check-

ing for interference with other pkts in the communi-

cation channel and analysing whether there are any

collisions and how strongly the pkt reaches the vari-

ous GTWs. After that, it is checked whether the com-

munication with the closest GTW was successful and

if it was, then the GTW will directly send the ACK to

the ED in the correct reception window, ignoring what

happens on the other GTWs; thus, excluding pkt du-

plication, on the other hand, if the closest GTW does

not receive the pkt, then it will be checked on the other

GTWs and it will be seen if, at least one of them, has

received it; if this is the case, then, it will be commu-

nicated to the closest GTW of the ED that must com-

municate an ACK to the ED; instead, if even the other

GTWs do not receive it, then the communication has

failed and a retransmission must be attempted. More-

over, when sending the ACK, there is the possibility

of inserting information to be given to the EDs, so as

in the case of the ADR.

5.3 Other Simulator Changes

Considering that in a network it is possible that there

are different types of devices related to different IoT

applications, it is provided now in the simulator the

possibility to select for each ED different pkts rate. It

can be created denser and sparser zones supporting

homogeneous and heterogeneous distributions; this

feature has been introduced to consider GTWs under

different trafﬁc loads. In ﬁgure 4 it is possible to see

as the additional behaviors of heterogeneous ED den-

sity is supported.

There is now also the possibility of inserting some in-

formation on the acknowledgment pkt, so in piggy-

backing as in the case of ADR; of course, there are

limited resources so there is no chance to put large

information, but it is possible to include some param-

eters such as SF or transmission power or the number

of pkts that a ED must transmit in ACK.

At the end of the simulation now, activating a func-

Figure 4: Unbalanced network vs. Balanced Network.

Figure 5: Coloured ED according to Frequency or SF.

tion, there is the possibility to show a plot, where

there will be inserted GTW and EDs and these can

be distinguished through colors. In ﬁgure5 are show

colors applied to ED with different SFs or different

frequencies. In the simulator another important fea-

ture that has been added is the collection of new in-

formation such as the energy consumed by each ED

and the whole network, pkts latency, collided pkts and

the number of totals sent and created pkts. Moreover,

it will be created a ﬁle containing besides those previ-

ously mentioned some new statistics, like:

• Weak pkts, collisions due to imperfect orthogo-

nality, times that it was not possible to receive an

ACK because of DC limitation and retransmitted

pkts;

• How many times there have been pkts that have

completely failed the communication; therefore,

they have exceeded the maximum retransmissions

value:

• How many collisions, distinguishing if they hap-

pened on the same SF or between different SF, on

the single GTWs;

• How many times each GTW has reached the

DC Limit and has not been able to send an ACK;

• What happens when a pkt does not arrive towards

the nearest GTW but towards other GTWs?

These statistics are supported also on the single ED,

so it can be seen how many times it was used a spe-

ciﬁc SF, how many times a speciﬁc Frequency, and

then will be computed the average value of the afore-

mentioned metrics for both the single and the whole

network.

5.4 Final Structure of the Simulator

The new structure of the simulator can be seen in ﬁg-

ure 6. Every ED has its own parameters like the Lo-

cation, the Energy Proﬁle and the LoRa Parameters;

moreover, every ED, to communicate with the GTW,

uses the AirInterface, and every time that a ED will

send a pkt, a link to each AirInterface is provided

to account for the total interference introduced in the

SIMULTECH 2022 - 12th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

370

network. Every AirInterface compute if the pkts col-

lided or not and if they arrive with enough power to

the GTW. So, its task is to virtualize the communi-

cation channel of a GTW, then the NS is created in

an abstract mode, and it decides when a GTW must

send an ACK and when every GTW must respect the

DC Limit before it will send the ACK message.

Finally it worth noting that upon this kind of simula-

tor could be applied any kind of suited protocol and

an example of this is presented in (De Rango et al.,

2021)

Figure 6: Final Structure of LoRaEnergySim.

6 CONCLUSIONS

The aim of this paper is to provide an overview about

LoRa and the existing Simulator engines, which al-

low to evaluate the performance of LoRa network

in different ways. The overviewed Simulators are

Flora, PhySimulator, LoRaWAN Module for NS-3,

LoRaSim, LoRaFREE & LoRaEnergySim. We fo-

cused our attention on the last one providing a set of

custom modiﬁcations to extend the basic behaviour

implemented by LoRaEnergySim. The most impor-

tance additional features introduced in the simulator

are: Multi-GTW-Extension with interference man-

agement, the imperfect SFs orthogonality, ACK man-

agement and new statistics to improve the network

analysis.

REFERENCES

Abdelfadeel, K. Q., Zorbas, D., Cionca, V., and Pesch, D.

(2020). Free - ﬁne-grained scheduling for reliable and

energy-efﬁcient data collection in lorawan. IEEE In-

ternet of Things Journal, 7(1):669–683.

Antenna Sensitivity (2019). Understanding adr. https://lora-

developers.semtech.com/documentation/tech-papers-

and-guides/understanding-adr.

Bor, M. C., Roedig, U., Voigt, T., and Alonso, J. M. (2016).

Do lora low-power wide-area networks scale? In Pro-

ceedings of the 19th ACM International Conference

on Modeling, Analysis and Simulation of Wireless and

Mobile Systems, MSWiM ’16, page 59–67, New York,

NY, USA. Association for Computing Machinery.

Bouras, C., Gkamas, A., Katsampiris Salgado, S. A., and

Kokkinos, V. (2020). Comparison of lora simula-

tion environments. In Barolli, L., Hellinckx, P., and

Enokido, T., editors, Advances on Broad-Band Wire-

less Computing, Communication and Applications,

pages 374–385, Cham. Springer International Pub-

lishing.

Callebaut, G., Ottoy, G., and van der Perre, L. (2019).

Cross-layer framework and optimization for efﬁcient

use of the energy budget of iot nodes. In 2019

IEEE Wireless Communications and Networking Con-

ference (WCNC), pages 1–6.

Croce, D., Gucciardo, M., Mangione, S., Santaromita, G.,

and Tinnirello, I. (2018). Impact of lora imperfect

orthogonality: Analysis of link-level performance.

IEEE Communications Letters, 22(4):796–799.

De Rango, F., Lipari, A., Stumpo, D., and Iera, A. (2021).

Dynamic switching in lorawan under multiple gate-

ways and heavy trafﬁc load. In 2021 IEEE Global

Communications Conference (GLOBECOM), pages

1–6.

Farrell, S. (2018). Low-power wide area network (lpwan)

overview. Rfc, RFC Editor.

FLoRa (n.d.). Home — ﬂora - a framework for lora simu-

lations. https://ﬂora.aalto.ﬁ/.

LoRaEnergySim (n.d.). Github - gillesc/loraenergysim:

Lora network simulator to monitor energy consump-

tion. https://github.com/GillesC/LoRaEnergySim.

LoRaSim (2017). Lorasim.

https://www.lancaster.ac.uk/scc/sites/lora/lorasim.html.

Marini, R., Mikhaylov, K., Pasolini, G., and Buratti, C.

(2021). Lorawansim: A ﬂexible simulator for lorawan

networks. Sensors, 21(3).

PhySimulator (n.d.). Lora – simulation & experimentation

tools for lora networks. http://lora.tti.unipa.it/.

Reynders, B., Wang, Q., and Pollin, S. (2018). A lorawan

module for ns-3: Implementation and evaluation. In

Proceedings of the 10th Workshop on Ns-3, WNS3

’18, page 61–68, New York, NY, USA. Association

for Computing Machinery.

Slabicki, M., Premsankar, G., and Di Francesco, M. (2018).

Adaptive conﬁguration of lora networks for dense iot

deployments. In NOMS 2018 - 2018 IEEE/IFIP Net-

work Operations and Management Symposium, pages

1–9.

Zourmand, A., Kun Hing, A. L., Wai Hung, C., and Ab-

dulRehman, M. (2019). Internet of things (iot) us-

ing lora technology. In 2019 IEEE International Con-

ference on Automatic Control and Intelligent Systems

(I2CACIS), pages 324–330.

Extending LoRaEnergySim Simulator to Support Interference Management under Multi-Gateway IoT Scenarios

371