LPWAN IEEE 802.

11ah and LoRaWAN capacity

simulation analysis comparison using NS-3

Oukessou Yassine, Mohamed Baslam and Mohamed Oukessou

Department of Mathematics and Computer Science, University of Science and Technology,
Beni Mellal, Morocco
Email: oukyassine@gmail.com

Abstract— Low Power Wide Area Networks are the the number of devices, the second by increasing the radius
fundamental keys of the internet of Things (IoT). These networks between the GW and devices. The simulations discussion and
provide long-range coverage to end nodes, exploiting license-free planned future work to our analysis comparison are detailed in
frequency bands and the sub 1Ghz communications. In this paper Section 4. Finally, conclusions are discussed in Section 5.
we evaluate and analyze using NS-3 network simulator, the uplink
throughput of the two promises networks IEEE 802.11ah and
LoRaWAN all by implementing the maximum number of the end II. OVERVIEWS OF IEEE 802.11AH AND LORAWAN
devices served by a single base station. Also in the second scenario
where the nodes are depends on the distance from the base station. A. Overview of IEEE 802.11ah
In the conclusion we brief and compare the obtained results, and The IEEE 802.11ah as a wireless communication Physical
we make actions to be taken in case of making applications or and Mac layer protocol operates in unlicensed sub 1GHz (863–
deployments using these networks. 868 MHz in Europe and 902–928 MHz in North-America),
supporting long range transmission, and a maximum of 8192
devices connected to a single access point. There are two major
configurable channel bandwidths 1 MHz and 2 MHz as well as
Keywords— IEEE 802.11ah, LoRaWAN, NS-3, LPWAN, IoT,
the data rates are ranging from 150Kpbs to 347Mbps. New
sub 1 GHz, ALOHA, Game Theory
features have been designed for the IEEE 802.11ah mac layer,
such as fast association and authentication, RAW, and power
I. INTRODUCTION saving.
The internet of things (IoT) is recently represent a new trend In a dense network where thousands of devices contending
of the evolution of the wireless communications. In which by the channel access the collision probability increase. The RAW
end of this decade will reach 20.8 billion IoT end points [1]. mechanism reduce this collision and limits a set of devices that
Unlike the traditional broadband networks, the low power wide can access the channel at a time and spreads their attempts over
area networks (LPWANs) are the new technologies that are a long period of time to enable the efficient network
designed to bring to the novel IoT applications the new communication of a high number of devices [8]. Devices will be
requirements of low-rate, long-range in the unlicensed sub-GHz divided into Associated Identifier (AID) (a unique value,
frequency bands [2] and delay tolerant at the low energy usage through which the AP assigned to device during association) [9].
and cost. The airtime is split into time slots, then each airtime interval
Today many competing LPWANs solutions are present in assigned to one RAW group. The devices are only allowed to
the market. We focus specifically on: The new IEEE 802.11ah transmit during its RAW group slot and sleep during other RAW
standard, marketed as Wi-Fi HaLow, which was designed to groups slots. Different from the previous IEEE 802.11, the two
provide a much greater coverage at a range of up to 1 kilometer back off function states are used by each device on respectively
while maintaining a data rate of 150 Kbps with a considerably outside and inside RAW. The first outside back off is suspended
higher throughput, and the new Long Range (LoRa) WAN at the start of RAW and resume the previous stored state at the
developed by Semtech and based on LoRa proprietary spread end of the RAW. On the other hand the second inside back off
spectrum technique and Gaussian Frequency Shift Keying used if the device is participating in the RAW by using RPS
(GFSK) [3]. (RAW parameter set) carried by beacons. For a more details of
the standard, the referred existing works should be checked [5,
The rest of this paper is structured as follows. Section 2 6].
provides an overview of the most prominent 802.11ah features
and a briefly description of the principles of Lora technology B. Overview of LoRaWAN
features. In Section 3, we implement the NS-3 [4] simulations
for the both technologies, two scenarios will be performed by The LoRa physical layer uses Chirp Spread Spectrum (CSS)
several devices generates a periodic traffic and served by a modulation, a spread spectrum technique where the signal is
single access gateway, then the throughput and packet loss will modulated by chirp pulses (varies linearly in frequency),
be checked at the gateway point: the first option by increasing parametrized by the Spreading Factor (SF), which can take

978-1-5386-4225-2/18/$31.00 ©2018 IEEE

values SF ∈ {7,..., 12}, we note that the higher the SF is, the 1,2
longer in time the packet and the more reliable its reception will

Collision Packet Loss rate

be [10]. Therefore it’s improving robustness against interference 1
and help to fight against heavy multipath fading characteristic 0,8
for indoor propagation and urban environments [11].
0,6
Typical bandwidth values are 125 and 250 kHz while data
rate ranges from 0.3 kbps to 50 kbps. While the LoRa PHY layer 0,4
is proprietary, the MAC layer, known as LoRaWAN, is open and 0,2
described by the LoRa Alliance specification in [12].
0
0 200 400 600
III. PERFORMANCE EVALUATION
No of devices
Exploiting the NS-3 network simulator [4] and the two
repositories [13,14], we have made the first simulation by
evaluating the throughput of the two LPWAN networks as a Fig. 2 802.11ah MCS10 Packet loss per No of devices
function the traffic generated by N end devices (by increasing
the N value) distributed around a single access gateway in a
fixed radius(no buildings are present). The second simulation we increase the distance as showed
in Figure 3, the MCS10 prove again their best results with
We assume that the device k ∈ {1... N} generates every τ
approximately 980 m where the packet loss due to the under
seconds a packet which occupies the channel for seconds in
order to be transmitted as described in [15]. sensitivity is 50% as described in Figure 4.

The throughput S then is:

30
25
t 𝑆=P ×
,
(Packets/seconds)
Throughput
, 20

The second simulation scenario aimed to evaluate the 15

variation of the throughput by increasing the distance between 10
access gateway and the end devices as well as fixing the number
of devices. 5
0
A. IEEE802.11ah throughput evaluation 0 200 400 600 800 1000
We consider in the first simulation test the only uplink traffic Distance (meters)
and that all the devices are deployed randomly within 150 meters
around the access point, using the same modulation and coding
schemes MCS10 (Ofdm data rate= 150Kbps, BW=1MHz) [16] Fig. 3 802.11ah MCS10 Throughput per distance
as a result that using modes that provide lower data rate
significantly increases the maximum transmission distance , no
duty cycle restriction, all devices are supporting RAW ( only one
1,2
RAW group is used), and finally each device generates 5
packets/s. Figure 1 shows the throughput gathered at the AP, at 1
Packet Loss rate

this stage the MCS10 is resisting to the collision within 0,8

approximately 750 devices where the packet loss due to
congestion is about 56% as shown in Figure 2. 0,6
0,4
30
0,2
(Packets/seconds)

25
Throughput

20 0
0 200 400 600 800 1000
15
10
Distance (meters)
5
Fig. 4 802.11ah MCS10 Packet loss per distance
0
0 200 400 600
B. LoRaWAN throughput evaluation
No of devices
We assume in the first LoRaWAN simulation test the only
uplink traffic generated by devices (each one generates 1 packet
Fig. 1 802.11ah MCS10 Throughput per No of devices
every 10min), and each device 𝑖 use a SF ∈ {6, 7, 8, 9, 10, 11,
Fig. 7 LoRaWan Throughput per distance
12} randomly. The single gateway is emulating 8 windows
receivers working in parallel (1% duty cycle to be shared
between all sub-channels). From Figure 5 we remark that a small 1
throughput fluctuated decrease registered as a function of the

Probability ratio (%)

number of devices with an insignificant interfered and no more 0,8
receivers (If a packet arrives at a certain LoRa channel and there 0,6
are no available receive paths listening, the packet is lost [10])
packets probability ratio as described in Figure 6. 0,4

0,2
600
0
(Packets/minute)

500
0 5000 10000 15000 20000 25000
Throughput

400
Distance (meters)
300
200
Interfered NoMoreReceivers
100
UnderSensitivity
0
0 2000 4000 6000 8000
No of devices Fig. 8 LoRaWan probability packets loss per distance

Fig. 5 LoRaWan Throughput per No of devices IV. SIMULATIONS AND DISCUSSION

As seen in above section the IEEE 802.11ah MCS10
0,25 simulation pose his pillars in terms of throughput (21 packets/s
Probability ratio (%)

0,2 at 750 devices), thanks to OFDM modulation which permits to

achieve easily a data rates up to 150 kbps. However a significant
0,15
decrease registered below 12 packets/s around 1000 meters.
0,1
The second test section shows a high resistant to the number
0,05
of devices in LoRaWan (500 packets/min at 8000 devices),
0 thanks to CSS modulation and duty cycle that improves the
0 2000 4000 6000 8000 robustness against interference and other radio effects.
No of devices Furthermore a high coverage (11 packets/min around 21000m)
plotted above.
Interfered NoMoreReceivers At this point of comparison considering the future things
(smart city, home, industrial, agriculture…) applications which
UnderSensitivity doesn’t need a high data rates, we highlight the LoRaWan as the
best IoT network which permits to achieve a long range distance
Fig. 6 LoRaWan probability packets loss per No of devices with a less important gap of congestion that affects the whole
network.
The final simulation test show us a significant throughput
On the other hand a game theoretic model approach of a
decrease reaching 50% of the under sensitivity probability at
selfish access strategies based on ALOHA random access
approximately 12500 meters as shown in Figure 7, caused by the scheme exploiting Nash and Pareto equilibrium will be analyzed
propagation loss as demonstrated in Figure 8. in the future work in order to optimize the uplink throughput of
nodes.
80 For this reason only uplink transmissions are considered and
(Packets/minute)

60 we assume that all devices try to maximize its uploading rate.

Throughput

Each device 𝑖 ∈ [0-n] can arbitrarily set its channel access

40 probability τ . The outcome of the game is a set of strategies
20
taken by the players (devices): τ = (τ , τ , … , τ ) ∈ [0,1]𝑛 , we
can readily express the uplink throughput per channel of the i-th
0 device as:
0 5000 10000 15000 20000 25000
𝑆 =τ ×p_ ×𝐿
Distance (meters)
 Where L is the packet length which is assumed to be fixed,
p_ the probability that at least one confirmation received
from the server (considering the class A [12]).
[4] Available on: https://www.nsnam.org/. [Accessed on 2017-10-07]
We note that the pure ALOHA impose the Poisson process [5] E. Khorov, A. Lyakhov, A. Krotov, and A. Guschin. A survey on IEEE
on the two access channel and ACK probabilities [17]. 802.11ah: An enabling networking technology for smart cities. Computer
Communications, 58:53–69, 2015.
We assume also that the channel and all SFs are orthogonal,
[6] T. Adame, A. Bel, B. Bellalta, J. Barcelo, and M. Oliver. IEEE 802.11ah:
so when two or more transmissions occurs at the same time a the WiFi approach for M2M communications. IEEE Wireless
collision will happen. Communications, 21(6):144–152, 2014.
[7] Le Tian, Jeroen Famaey and Steven Latré. Evaluation of the IEEE
V. CONCLUSION 802.11ah Restricted Access Window Mechanism for dense IoT networks.
2016 IEEE 17th International Symposium on A World of Wireless,
In this paper, we have made a comparison using the famous Mobile and Multimedia Networks (WoWMoM).
network simulator NS3 between two promising low power [8] Yanru Wang , Kok Keong Chai , Yue Chen , John Schormans and
range networks. Jonathan Loo. Energy-aware Restricted Access Window Control with
Retransmission Scheme for IEEE 802.11ah (Wi-Fi HaLow) based
Simulation results reveal that the IEEE 802.11ah using Networks. 2017 13th Annual Conference on Wireless On-demand
MCS10 modulation has the congestion rate of 56 % with 750 Network Systems and Services (WONS).
sensors nodes, where the propagation packet loss of 50 % [9] E Khorov, A Krotov, A Lyakhov, Modelling Machine Type
around 1 Km from the access point. On the other hand Communication in IEEE 802.11ah Networks, 2015 IEEE International
LoRaWan resulting below 37 % in congestion packet loss for Conference on Communication Workshop (ICCW), IEEE, May, pp.
1150–1154.
an access gateway serving approximately 8000 nodes, where a
[10] D. Magrin, M. Centenaro and L. Vangelista, "Performance evaluation of
propagation loss of 50 % around approximately 12500 meters. LoRa networks in a smart city scenario," 2017 IEEE International
Finally we highlighted LoRaWan as the best network which Conference On Communications (ICC), Paris, 2017.
can serving the new things applications. Therefore we plan to [11] “LoRa™ Modulation Basics,” Semtech Co., Camarillo, CA, AN1200.22,
optimize it using the Mathematic models of the game theory. Rev. 2, May. 2015.
[12] Available on: https://www.lora-alliance.org/for-developers. [Accessed
on 2017-12-08]
REFERENCES [13] Available on:https://github.com/MOSAIC-UA/802.11ah-ns3. [Accessed
on 2017-12-11]
[14] Available on: https://github.com/DvdMgr/lorawan. [Accessed on 2017-
[1] P. Middleton, T. Tully, J. F. Hines, T. Koslowski, B. Tratz-Ryan, K. F. 11-11]
Brant, E. Goodness, A. McIntyre, and A. Gupta, “Forecast: Internet of
Things - Endpoints and Associated Services, Worldwide, 2015,”p. 57, [15] N. Benvenuto and M. Zorzi, Principles of communications Networks and
2015. [Online]. Available: https://www.gartner.com/doc/3159717/ Systems. Wiley Online Library, 2011.
forecast-internet-things—endpoints [16] Le Tian , Sébastien Deronne , Steven Latré , Jeroen Famaey.
[2] M. Centenaro, L. Vangelista, A. Zanella, and M. Zorzi, “Long-Range Implementation and validation of an IEEE 802.11ah module for NS-3.
Communications in Unlicensed Bands: the Rising Stars in the IoT and WNS3 '16 Proceedings of the Workshop on ns-3, pages 49-56, 2016.
Smart City Scenarios,” IEEE Wireless Commun., vol. 23, no. 5, pp. 60– [17] Dmitry Bankov, Evgeny Khorov, and Andrey Lyakhov. Mathematical
67, Oct. 2016. Model of LoRaWAN Channel Access. 2017 IEEE 18th International
[3] N. Sornin et al., “LoRaWAN Specification,” LoRa Alliance Inc., San Symposium on A World of Wireless, Mobile and Multimedia Networks
Ramon, CA, Ver. 1.0., Jan. 2015 (WoWMoM), China, 12-15 June 2017.