A Scalable Slotted Aloha for Massive IoT:

A Throughput Analysis
Soukaina Ihirri12 , Essaid Sabir1 , Ahmed Errami1 , Mohamed Khaldoun1
1 NEST Research Group, LRI , ENSEM, Hassan II University of Casablanca, Morocco
2 LPRI, EMSI, Casablanca, Morocco.

{soukaina.ihirri, e.sabir, a.errami, m.khaldoun}@ensem.ac.ma

Abstract—LoRa (Long Range) is increasingly catching the LoRa represents half of the WAN connections for appli-
attention of researchers, industry leaders and business influencers cations [14]that are not feasible with conventional cell net-
as one of the most promising standard for Internet of Things works, such as: smart meters, Agriculture to control irrigation
(IoT) ecosystem. LoRa uses slotted Aloha protocol allowing
connected IoT devices to send data in a sporadic way and systems, Logistics for equipment monitoring, Environment to
a fully distributed fashion without syncing with other devices. monitor pollution, Climate Change [4], Health to Diagnose
Under a massive IoT environment, LoRa experiences too many Patients in Real Time and Home Automation or Fire Detection
collisions which decreases the average throughput drastically Sensors. Around 250 LoRa base stations are enough to cover
and increases the expected delay. In this paper, we propose the entire country of Switzerland. For Netherlands, approxi-
a scalable slotted Aloha access method for LoRa-empowered
Massive IoT environments. Our scheme uses the option field of mately 500 such base stations are required [5]. MAC protocols
the LoRa frame, and proposes to iteratively reduce the number are widely used, but are too complex and expensive to use
of contending devices till this field expires. The remaining devices for LoRa needs. LoRa networks use the Aloha protocol that
will access the channel using the legacy slotted Aloha. Our allows a connected object to send without having to negotiate
approach does not require any change of the LoRa frame which permission to do so. This protocol is based on the time
is of great interest from implementation perspective. Analytic
results and extensive numerical examples show clearly how our division multiple access (TDMA) MAC scheme, where nodes
scheme outperforms the legacy LoRa. We also provide some are allowed to transmit a packet at any time and scheduled to
insights on how to select the network parameters so as the whole transmit and receive during a concerned Slot(s).
network performance is maximized. LoRa is considered as a viable solution for large scale dis-
tributed measurement systems. A huge amount of information
Index Terms—LoRa, slotted Aloha, MAC layer, IoT devices,
Throughput analysis. is collected from a very large number of smart devices.
However, in some cases its recommended to favor information
I. I NTRODUCTION over others in order to limit the potential for collisions.
Internet of Things (IoT) is one of the key technologies In this paper, we propose a scalable slotted Aloha MAC
following the development of computing and communication, protocol. The solution is an improved version of the slotted
not only at research level but also at application level. It aims Aloha where a logical frame is divided into 2 sub-frames:
at collecting data from billions of devices connected altogether. Contention sequence (CS) code and DATA. The idea is to
Smart and connected things are transforming themselves from resolve the problems related to access to the same shared
stand-alone devices to parts of collaborative systems. There- communication channel. The contribution of the paper is
fore, we are moving towards a more “connected world”, which twofold:(i) a CS code is introduced in the frame option for
did not miss the arrival of a plethora of technologies. So many all devices having a packet to transmit; thus, the number of
data generated and collected from a very large number of smart devices sensing the channel is going to decrease in order
devices. Today, the vision of IoT has expanded to connect to decrease collisions in LoRa transmissions; (ii) we derive
everything for long range, thanks to the combined efforts of a mathematical model to study the performance of Massive
companies and universitys research. The deployment of smart LoRa network. We compute the optimum CS length and the
devices serving various IoT applications is estimated to grow success probability in order to achieve a maximum throughput.
over 30 billion globally by 2020 [1]. The world of IoT did not We demonstrate that the proposed Scalable slotted Aloha
miss the arrival of LoRa (Long Range) technology [2] [3]. present a promising throughput compared to the standard
LoRa is a type of network or more exactly a long distance slotted Aloha.
and low speed technology, operating on free frequencies, This paper is organized as follows. Section II briefly intro-
characterized by triple winner: long range, low flow, and low duces the problem formulation. The average throughput of
energy consumption. The goal of LoRa is to allow the creation the Scalable slotted Aloha is presented in section III, and
of a large network for connected objects, to reduce the costs as section IV provides details about the achievable throughput
well as the electricity consumption of the connected devices. for symmetric case. Numerical Investigations are presented
in section V. Finally, section VI presents some concluding
remarks.

978-1-5386-7747-6/19/$31.00 ©2019 IEEE 508

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.
 II. P ROBLEM FORMULATION 4 Bytes 1 Byte 2 Bytes 0..15 Bytes

Frame Device Frame Frame Frame

A. LoRa network Basics Header Address Control counter Option

The IoT market has been witnessing the rapid spread of

new commercial technologies, based on another networking MAC
1 Byte

Frame Frame
paradigm referred to as Low Power Wide Area Network Payoad Frame Header
Port Payload
(LPWAN). LPWAN is a very popular networking technology
that supports a large amount of underground, underwater or
1 Byte 4 Bytes
even deep inside building devices for an interesting life time. PHY
Payoad MAC
MAC Payload MIC
It is an excellent fit to the IoT traffic coming from different Header

applications. LPWAN employ low power communication

20 Bytes
technologies that connect thousands of IoT devices. Several (Header and H CRC) 16 Bytes

LPWAN protocols has been presented in the literature [6] Radio

H
PHY Preambule Header PHY Payload CRC
CRC
such as Sigfox [7], NB-IoT, symphony link and LoRaWAN
[8].
It is important to note that Long Range (LoRa) and its
network architecture LoRaWAN, is currently the most Fig. 1. Lora Frame
adopted technology defines as one of the greatest among
LPWAN protocols [15]. Many advantages can be cited in
term of low cost, excellent battery performance and also the Ethernet, 3G/4G, or other WAN technologies. The LoRaWAN
deployment with less cost. We should highlight that LoRa and components can be classified as shown in figure 2:
LoRaWAN are two separate things that are often conflated: • Class A: network end node such as sensors for low power
LoRa is the physical layer, while LoRaWAN (Aloha based) uplink communication. End devices can only communi-
is the MAC layer, in other words, the software thats put on cate with gateways via single-hop LoRa communication.
the chip (physical layer) to enable networking. • Class B: Gateways which are connected to the central
LORA is an open technology; it means that any improvement server via the IP protocol suite. It is for delay guarantted
is desirable. A typical LoRaWAN network consists of downlink communication.
end-devices (sensors) gateways and a server which collects • Class C: Data server which is for end devices without
and analyzes information mined by the devices(See figure 2). energy con.
The PHY and MAC layers of LoRa were recently described
in [9]. The MAC layer and, according to the specifications in
[10], it is basically an Aloha protocol controlled primarily by
the LoRa Net Server.

B. LoRa (Physical Layer)

LoRa is a physical layer radio modulation technique, De-
veloped by Semtech. At the PHY layer, a LoRaWAN frame
(see fig.1) starts with a preamble. Apart from the synchro-
nization function, the preamble defines the packet modulation
scheme, being modulated with the same spreading factor as
the rest of the packet. Typically, the preamble duration can be
configured to be between 10.25 and 65,539.25 symbols. After
the preamble, there is an optional Phy Header that includes
information concerning the payload length and the strategy of
encoding. This header is followed by its own CRC to discont
packets with invalid headers. The payload present the H-CRC.
Fig. 2. Exemple of LoRa Network
At the end we found the Optional CRC. Figure 1 defines also
the sub-layers of LoRa frame, such as the the PHY Payload,
MAC payload and Frame Header with its commponenets. C. Background and main idea of the proposed protocol
A LoRaWAN network consists of three components: end The topology of LoRa network is a star of stars, which
devices, gateway(s), and a network server. The communication means that groups of IoT devices are connected to gateways
between an end device and gateway is ensured by LoRa via LoRa wireless links. In very dense deployment scenarios,
and the LoRaWAN MAC. The gateway forwards the end the potentially huge number of IoT devices that shall gain
device data to the server. To connect a gateway to server we connectivity through a single base station will raise a number
use local area or wide-area networking technologies, such as of issues related to the signaling and control traffic, which may

509
Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.
become the bottleneck of the system [4]. At the same time, highlight that the CS’s bits can’t be all zeros or ones, because
the probability that devices can have a successful transmission these two combinations;if they are present in the battle; will
tends towards zero. Therefore, LoRa’s limitations are unclear represent always a winner and a loser .
because there is still no deployment of a massive LoRaWAN
network. In this paper we introduce a hybrid MAC protocol E. Protocol design (Frame option field)
for Massive LoRa network to increase the chances that a As it was described in sub section B that LoRa Frame
device having a packet to transmit has access to the channel. contain respectively the Preamble, Header, H-CRC, the PHY
The proposed protocol is based on both the ZFA(Zero Failure payload and finally the CRC. In order to improve anything
Access) access protocol [11], which is used to resolve the related to LoRa Standards, no modification in pre-existing
problem of conflicts in a shared communication channel LoRaWAN Libraries is required.
environment, and the slotted Aloha Used for LoRa Networks. In a different way from what has been presented in the
The concept of the proposed scalable slotted Aloha is to literature, our approach is to improve the slotted Aloha
eliminate stations and let at list a set of stations having a for LoRa Network by exploring the Frame option of the
packet to transmit to access the channel. We assume that the frame Header to implement our technique. The frame
set of the elective users depend on a contention sequence Options contains commands used to validate connection,
code in the frame option of LoRa frame Header. for the transmission power and to change the data rate, etc.
Throughout this, it can be seen that we will not change the
basics of LoRaWAN.
D. Divide and rule paradigm
As we mentioned previously, the proposed solution is based It is known that Aloha present high collision probability in
on ZFA [11]. The ZFA is based on putting all stations in a massive networks. Our Approach rely on allowing several
competition where all of them sensed the channel to be station to access the shared channel and to decrease the
unoccupied and try to access the medium at the same time. collision by appropriately selecting an optimal Code sequence.
To limits the number of collisions, the ZFA pick out at the This CS code used in order to achieve an optimal transmission
end one station to access the channel via competition process. probability as well as an optimal throughput. Each frame
The frame of each station is divided into 2 sub-frames; the CS station contains a unique sequence CS that satisfy 0 < m <
code which is including m bits used in the completion process log2(N ) where m represent the size of the CS sequence in the
and data sub frame. We should highlight that each station must frame option, and N is the number of station in the concerned
pass through three successive phases: network.
1 Phase of unoccupied channel, when all stations must If we take for example m=1, Half of the users will access the
sense the channel before going through combat process. channel(see fig.3). With m = 2,Split the number of users into
2 The combat between stations, in which, all stations trans- four halves. N4 are going to share the channel. In a general
mit there CS code in order to choose one station as a case, with m as length of the CS code, the number of users is
winner going to be 2Nm . Our proposed protocol define a kind of battle
3 The phase of transmitting, which is the last phase. The where all station try to access the shared channel in order to
chosen station can transmit in the shared channel. give the opportunity to active stations that have at least one
packet ready to transmit their packets.
the throughput in ZFA [11] is written as follow:

LoRa Frame
X Y Y
θZF A = λi (1 − λj ) (1)
Z∈M i∈Z j∈M \Z N/2
Network Size

m=1
ZFA allows only one station to access the medium. However, N/4
when we talk about LoRa network, we ought to think about
m=2
the massive amount of information coming from different N/8
stations. The problem of collision can be easily generated. Our m=3
proposed solution is an improvement of the standard slotted
Aloha for Massive IoT network. Based on divide and conquer
technique, the goal is to identify the set of active nodes having
a packet to transmit to access the medium and transmit data
Network Size

successfully via competition process.

In our analysis, we rely on the assumption that The CS N/(2^m)
sequence is unique and must be changed after each battle. Centention Sequence Data
under this assumption, and in order to ensure the balance CS(m Length)
between stations, we use the cycle strategy [8]. This technique
approximately ensure an equitable channel access by a rotation Fig. 3. Scalable slotted Aloha Frame
of the bits of the chosen CS code of each station.We should

510
Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.
 III. AVERAGE T HROUGHPUT OF S CALABLE SLOTTED The throughput can be simplified by replacing the following
A LOHA expression from equation (4) with that simplification:
In general, the throughput is the fraction of slots that contain  
PN −1 N −1
a successful transmission. For a very low transmission rate (a A= z=0 (λ)z (1 − λ)N −1−z (1 − p)z
few packets/day), throughput is impacted by packets collisions. z
However, at higher transmission rates duty cycling impacts A = (1 − λp)N −1
the throughput. We consider an N IoT devices transmitting Then the throughput in the symmetrical case can be simpli-
to the same base station over a shared channel. The idea is fied to a good extent:
to put station trying to access the channel into a combat in M −m
θ(p, λ) = N.p.λ(1 − λp)N −1 (5)
order to have a set of stations that can access the channel. M
Our proposed technique aims to keep the set of station having We consider pλ = q then the formula (5) can be presented as
a packet to transmit beyond the active ones. We assume that follow:
time is slotted and the frame is divided into the CS code and
data frame. We should highlight that the CS code must be M −m
θ= N.q(1 − q)N −1 (6)
unique. We summarize principle notations of the paper in the M
following list: We highlight that the concept of the proposed protocol is to
• N is the total number of the station in the concerned eliminate stations and let at list a set of users to access the
network. channel. According to what has been presented in section II-
• K set of active nodes having a packet to transmit. c, the set of the elective users can be expressed as: 2Nm .Then,
• P = (p1 , p2 , ..., pN ) : denote the probability of transmis- formula (7) can be written as follow:
sion on the Channel on a given time slot. M −m N N

• λ = (λ1 , λ2 , ....λN ): The probability that a station has a

θ= .q(1 − q) 2m −1 (7)
M 2m
packet to transmit. In this section we will study the way throughput scales in
A user is said to be active if it has at least one packet Lora network when the nodes are fixed. Firstly, we have to
ready to be transmitted. Let ξ(i) denote the probability that look for an optimal m and q, which represent respectively
the station i will be active. So the normalized throughput of the size of the CS code and the transmission probability that
a station i, which is defined as the fraction of time, can be maximize the throughput.
presented in our case as: Since the goal of this study was to find the maximum through-
X X Y put, 3 propositions were presented. First proposition in case
θ(P, λ) = pi .ξ(i). ξ(Z). (1 − Pi ) (2)
we have the optimum transmission probability. Proposition 2,
i∈N Z∈N \i j∈Z
achieved when we use the optimum CS code length, while to
gain the best throughput option, the third proposition is given
Q Q
where: ξ(Z) = k∈z λk . k0 ∈N \Z (1 − λk0 ) and ξ(i) = λi .
by taken into consideration both the optimal CS and optimal
Now, the normalize throughput can be written as: transmission probability.
X X Y Y 1) Proposition 1: The optimal transmission probability can
θ(P, λ) = Pi λi λk (1−Pk ) (1 − λl ). be presented as:
2m
i∈N Z∈N \i k∈Z l∈N \{i,Z}
q∗ = (8)
(3) N
P ROOF
Where:
P Q To maximize the throughput we derived formula (7) with
• k∈Z λk (1 − Pk )): The probability that active respect to the transmission probability dΘ
Z∈N \i dq
nodes Z keep silent.
Q N
−2
• l∈N \i\Z (1 − λl ): The probability of being inactive. dθ N q(M − m)(2m − 2)((1 − q) 2m
=−
dq M (9)
IV. ACHIEVABLE T HROUGHPUT FOR SYMMETRIC I OT N
−1
D EVICES N (M − m)((1 − q) 2m
+ .
M
In a symmetric case wherepk = pj = p and λi = λj = λ.
The optimal transmission throughput is obtained as follow: By solving dΘdq = 0 we found the expression (8). The
achievable Throughput according to q ∗ is given by :
M −m 1
−1 θ(q ∗ ) = (1 − q ∗ ) q∗ −1
N
!
X N −1 (10)
θ(P, λ) = z
(λ) (1 − λ) N −1−z
(1 − p) z M
z (4)
z=0 2) Proposition 2: The optimal CS code length is:
M −m
. N.p.λ m∗ = log2 (λN ) (11)
m

511
Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.
 P ROOF
1
In order to compute the optimal CS length; that maximize N1=10
N2=100
the total throughput; the first order derivation is computation- 0.9
N3=1000
ally intractable, which could not help to drive a good form of N4=10000

Optimal tranmission probability q*

0.8
the optimal CS length .So we proceeded otherwise.
A theoretical limit on the CS length under a saturated consid- 0.7
eration for ZFA protocol [11] has been evaluated and proof
that m∗ equal to log2 (N ), where N is the network size.Our 0.6

approach reach log2 (λN ) in unsaturated network where λ is 0.5

the probability of having a packet to transmit.
According to the optimal m, the throughput formula (6) can 0.4

be presented as follow: 0.3

∗
M −m 1 1
Θ(m∗ ) = .q(1 − q) λ −1 (12) 0.2

M λ
0.1
It’s obvious that 0 < p∗ < 1 and as it has been already
m
mentioned that q∗ = λp∗ then 0 < 2λ N < 1 to finally find 0
0 2 4 6 8 10 12 14
that 0 < m∗ < log2 (λN ). Contention Code Length
we consider 2 throughput function, one depend on the the m
and the other in view of m∗ , then we compute the sign of the
Fig. 4. The Optimal transmission probability
difference.We found that Θ(m) < Θ(m∗ ).
3) Proposition 3: : Substituting equation (7), (8) and (11),
we obtain the achievable throughput proposition depending on
14
both the optimal CS length and transmission probabilitym∗
and q∗. 12
M − log2 (λN ) 1
Θ∗ = (1 − λ) λ − 1 (13)
Optimal Contention sequence length m*

M 10

V. N UMERICAL I NVESTIGATIONS
8
Different studies have sought to evaluate the performance
of Lora Network [12] [13]. It’s known that the network 6

throughput of a system is N.p.(1 − p)N −1 and the maximum

4
throughput is achieved when p=1/N.
In order to determine the system performance, we compute 2
the optimum m and q (q∗ and m∗ ). Our approach to optimize
the system performance is to select the optimal m∗ in order 0

to achieve an optimal transmission probability (q∗ ). λ1=0.1

−2
λ2=0.3
λ3=0.6
At a first glance, we represent the variation of q∗ for λ=0.9
−4
different m values. We take into consideration the frame length 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Number Of Device N
in our analysis. 4 cases was presented,very small network,
small network, medium and extended one.
Fig. 4 represents the optimal transmission probability (q∗ Fig. 5. The optimal CS according to Network Size
eq.(8)) with respect to the CS portion of the frame. As the size
of the CS increases, the probability of transmission increases
for different network sizes. Note that the optimum probability which make the fight. We should highlight that we consider
of transmission is an increasing function. This probability not only the case where the CS is generated via a cyclic allocation
only depend on the size of the CS, but also on the size of the technique [11].
network. Fig. 5 represents the variation of CS with respect to the size
As the probability of transmission increases, the optimal size of the network, for different case of transmission probability. It
of the CS code becomes larger, which gives us the possibility can be seen that the larger the size of the network, the number
that several elements can transmit at the same time on the of elements that will transmit is very numerous, therefore the
shared channel. optimal size of the CS becomes important with a very high
As expected, results for q∗ and m∗ are depending on the probability of transmission.
Network size. For the second optimal parameter m∗, we seek Now, we will evaluate the optimum throughput according
to see its variation with respect to the number of the stations to the two optimal parameters q∗ and m∗. Additionally, simu-

512
Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.
 legacy slotted Aloha. We currently deal with the derivation
0.502 the delay distribution of our scheme in order to get insights
λ=0.5; M=800
λ=0.5; M=1200 on how to support delay sensitive applications over LoRa.
0.501 λ=0.5; M=1700
λ=0.5; M=2000
R EFERENCES
0.5
[1] G.Jayavardhana, et al. “Internet of Things (IoT): A vision, architectural
0.499 elements, and future directions“. Future generation computer systems
Achievable Throughput

29.7 1645-1660, 2013.

0.498 [2] O.Georgiou, U.Raza. “Low power wide area network analysis: Can LoRa
scale¿‘. IEEE Wireless Communications Letters, vol. 6, no 2, p. 162-165
0.497 , 2017.
[3] A.Augustin, J.Yi, T.Clausen,, W. Townsley. “A study of LoRa: Long
0.496 range low power networks for the internet of things“. Sensors, 16(9),
1466. 2016.
0.495 [4] J.M.TALAVERA, L.E. TOBN, GMEZ,A. Jairo et al.“ Review of IoT
applications in agro-industrial and environmental fields“. Computers and
0.494
Electronics in Agriculture, vol. 142, p. 283-297, 2017,.
[5] W. J.Sung, H. G.Ahn, J. B. Kim, S. G. Choi . “Protecting end-
0.493
device from replay attack on LoRaWAN“. In Advanced Communication
Technology (ICACT), 2018 20th International Conference IEEE..pp.
0.492
167-171? , February 2018,
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 [6] J.Petajajarvi, K.Mikhaylov, A.Roivainen,T.Hanninen, M. Pettissalo .
Number of devices N “On the coverage of LPWANs: range evaluation and channel attenuation
model for LoRa technology“. In ITS Telecommunications (ITST), 2015
14th International Conference on . IEEE..pp. 55-59, December 2015,
Fig. 6. The achievable throughput according to m* and q* for different [7] G. G. Ribeiro, L. F. de Lima,L. Oliveira, J. J.Rodrigues, C. N. Marins,
Network size (Proposition 3) G. A. Marcondes. “An Outdoor Localization System Based on SigFox“.
IEEE 87th Vehicular Technology Conference (VTC Spring) . IEEE.pp.
1-5; In 2018.
[8] K.Mikhaylov, J.Petaejaejaervi, T. Haenninen, “Analysis of capacity and
lations are used also to analyze the variation of the throughput scalability of the LoRa low power wide area network technology“. In
according to both optimal CS length and optimal transmission European Wireless 2016; 22th European Wireless Conference; Proceed-
probability .Fig. 6 shows the variation of the throughput in ings of VDE.pp. 1-6, May 2016.
[9] C. Goursaud, , J. M. Gorce, “Dedicated networks for IoT: PHY/MAC
view of size of networks for different frame length and λ. It state of the art and challenges“. EAI endorsed transactions on Internet
is noted that the optimal throughput is approximately the same of Things, 2015.
for very small size of the network. In addition, the more the [10] L.Vangelista, A. zanella, M. zorzi. “Long-range IoT technologies: The
dawn of LoRa“. In Future Access Enablers of Ubiquitous and Intelligent
network becomes extended, stable the throughput is became Infrastructures . Springer, Cham..pp. 51-58, September2015.
for different frame size. [11] S. Arabi, A. Errami,M.Khaldoun, E. Sabir, J.El Mesbahi.“ Towards
We observe that the achievable throughput of the proposed a zero-Failure Distributed Access for Wireless Collision Channels“.
In Advances in Ubiquitous Networking . Springer, Singapore.pp. 63-
solution is promising compared to the standard slotted Aloha 76,2016.
that has approximately 0.368 as maximum throughput and tend [12] M.Centenaro, L. Vangelista, R. Kohno. “On the impact of downlink
toward zero for a large traffic. However, the throughput in our feedback on LoRa performance. In Personal, Indoor, and Mobile Ra-
dio Communications (PIMRC)“, 2017 IEEE 28th Annual International
proposed solution remains constant for a very dense traffic. It Symposium on . IEEE.pp. 1-6, October 2017.
is known that slotted Aloha is used in low-data-rate traffic.The [13] D.Bankov,E. Khorov, A. Lyakhov, .“ On the limits of LoRaWAN
Scalable slotted Aloha is a good contender for a massive IoT channel access“. In Proceedings of the 2016 International Conference on
Engineering and Telecommunication (EnT), Moscow, Russia .pp. 29-30,
networks. November 2016.
[14] P, Samuela, C. Carciofi, and M. Faccioli. ”NB-IoT and LoRA con-
VI. C ONCLUSION nectivity analysis for M2M/IoT smart grids applications.” 2017 AEIT
In this work, we present scalable slotted Aloha as a new International Annual Conference. IEEE, 2017.
[15] V. Lorenzo, A. Zanella, and M. Zorzi. ”Long-range IoT technologies:
access method for LoRa under Massive number of IoT devices. The dawn of LoRa.” Future access enablers of ubiquitous and intelligent
Scalable slotted Aloha uses a contention sequence (CS) to be infrastructures. Springer, Cham, 2015.
implemented in the option field of LoRa frame header. This
CS field is used to elect, in a distributed fashion, a smaller
group of nodes who might transmit within a given frame.
The main idea is to reduce by half the number of contending
IoT devices at each CS bit. This way, after i bits of CS, the
number of contending devices is at most N/2i (N being the
total number of IoT devices). We mainly aim to investigate
the proposed solution in terms of average throughput under
different network size and frame lengths. Next, we derive the
optimal CS and the optimal transmit probability allowing to
maximize the average throughput. Further, We illustrate the
behavior of our scalable slotted Aloha and compare it to the

513
Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on March 02,2024 at 18:27:31 UTC from IEEE Xplore. Restrictions apply.