You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327922682

On the Satellite Role in the Era of 5G Massive Machine Type Communications

Article in IEEE Network · September 2018


DOI: 10.1109/MNET.2018.1800024

CITATIONS READS
55 1,309

4 authors, including:

Stefano Cioni Riccardo De Gaudenzi


European Space Agency European Space Agency
99 PUBLICATIONS 1,206 CITATIONS 216 PUBLICATIONS 4,072 CITATIONS

SEE PROFILE SEE PROFILE

Nicolas Girault
European Space Agency
11 PUBLICATIONS 135 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Space communications View project

satnex View project

All content following this page was uploaded by Stefano Cioni on 08 November 2018.

The user has requested enhancement of the downloaded file.


On the Satellite Role in the Era of 5G Massive
Machine Type of Communications
Stefano Cioni, Riccardo De Gaudenzi, Oscar Del Rio Herrero, Nicolas Girault

Abstract

The fifth generation of mobile radio communication systems, dubbed 5G, has the challenge to
cope with tremendous increases in data traffic volume and peak data rates, reduced latencies, along
with improved energy efficient transmissions and new use cases. In fact, besides the traditional mobile
broadband services, future 5G networks will face the opportunity of embodying connections to billions
of objects, the so called Internet of Things or massive machine type communications. The paper is
addressing the question on what is the possible role of satellite systems in massive machine type
communications services. Key satellite massive machine type communications system design trade-offs
are outlined jointly with some examples of network sizing.

Keywords: Multiaccess communication, Satellite communication, Packet radio, Radio com-


munication, Digital communications.

I. I NTRODUCTION

While terrestrial technologies will play a key role in the provision of 5G services, satellite
networks can have a complementary role thanks to their attractive features such as their very wide
coverage area and short service deployment time. Among the variety of 5G standard services
to be supported, the massive machine type communications (mMTC) represents probably the
newest in terms of requirements. Compared to mobile broadband (MBB), required data rates
are extremely low as well the Internet of Things (IoT) devices activity time. Network scalability
and coverage jointly with energy efficiency are going to be essential mMTC features in 5G. In
this paper we will demonstrate that current space segment is able to support 5G mMTC type of
service with appealing cost and performance.

S. Cioni is with Ajilon Tech. Aerospace for European Space Agency, R. De Gaudenzi, O. Del Rio Herrero and N.
Girault are with European Space Agency, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands. e-mail: {Stefano.Cioni@esa.int,
rdegaude@gmail.com, Oscar.Del.Rio.Herrero@esa.int, Nicolas.Girault@esa.int}
2

The rest of the paper is organized as follows: Sect. II presents the main specific mMTC re-
quirements and the current available technologies addressing these peculiar applications; Sect. III
illustrates key mMTC satellite networks system design aspects and trade-offs; Sect. IV is ded-
icated to air interface aspects and provides examples of network sizing; Sect. V deals with
network layer aspects and commonalities with terrestrial 5G solutions. Finally, Sect. VI deals
with the summary of the potential satellite role in the 5G ecosystem.

II. M ASSIVE M ACHINE T YPE C OMMUNICATION N EEDS

So far, the main improvement axis for mobile wireless communication networks has been
the increase of the supported data rate for human type of communications. The 5G suite of
terrestrial wireless standard is also expected to effectively sustain very large amounts of connected
machines. As the name suggests, mMTC is about massive access by a large number of devices,
i.e., about providing wireless connectivity to tens of billions of often miniaturized low-complexity
low-power machine-type devices. mMTC service provision poses new challenging requirements
for the 5G standardization community, such as high spectral and protocol efficiency in a fully
uncoordinated and high density network of devices characterized by very infrequent transmission
of small size packets [1]. The mMTC user equipment (mMTC-UE) shall be characterized by
low power consumption combined with very modest unit price combined with low cost for
installation, operation and service provision.
mMTC applications include a very wide range of use cases for different market segments, from
a massive amount of devices distributed over a wide geographical area providing surveillance,
monitoring critical components and assisting in information sharing, to more locally confined
applications (e.g. smart homes connecting indoor devices in densely populated areas). Table I
shows a classification of mMTC terminals in three categories with specific user needs (Alarm
Devices, Telemetry and Tracking Devices, and Traffic Aggregators), each category addressing a
subset of applications.
Satellites can provide a worldwide coverage for mMTC, for densely populated areas as well
as for remote areas. Table II1 summarizes the key satellite mMTC-UE requirements in two
distinct time frames: 2018 (i.e. today) and 2023 when 5G is expected to be deployed worldwide.

1
The Table II content is based on internal research work performed by the European Space Agency, taking into account the
evolution of the user needs in the next decade for different mMTC applications by satellite. The targeted improvements in terms
of prices and miniaturization are in line with the trends observed in terrestrial networks.
3

Category Needs/Characteristics Use Cases Examples


Alarm Devices Limited latency Alarms and alerts
High availability and reliability, including indoor Emergency and distress signals
Limited message size Theft prevention
Rare transmissions, event-driven Geofencing
Telemetry and Tracking Devices Support to low and very low end devices Asset tracking and Geo-positioning
From very low to medium device/service cost Environmental monitoring sensors
High service reliability Energy and water management
Miniaturised, preferably self-powered devices Integration with smartphones/tablets
Mobile usage (omni-directional antenna)
Location tracking capability (for most applications)
Scheduled transmissions, time-driven
Traffic Aggregators Backhauling, aggregating traffic from low-end sensors Backhauling of mMTC networks
Relatively high aggregated data rates for remote areas
High service availability

TABLE I
S ATELLITE M MTC PRODUCT CATEGORIES , KEY REQUIREMENTS AND USE CASE EXAMPLES .

Year 2018 2023


Parameter Telemetry & Alarm Traffic Telemetry & Alarm Traffic
Tracking Devices Devices Aggregator Tracking Devices Devices Aggregator
mMTC OEM complete cost [e] 250 130 150 20 20 70
Latency at 95% success [minutes] < 60 <1 < 60 < 60 <1 < 60
mMTC-UE size [cm] 5-15 15 30-100 2-5 2-5 15-50
Message size [bytes] 50 30-50 150-200 50 30-50 150-200
Messages per day (per device) 1-96 1 24-96 1-96 1 24-96
ARPU/month [e] 50 10 30 10 2 12
Mean time between mainten. [years] 5 5 5 15 10 15

TABLE II
S ATELLITE M MTC-UE REQUIREMENTS .

For mass market, it is evident the need for a considerable reduction in the mMTC Original
Equipment Manufacturing (OEM) cost (by a factor 2-12.5) and size (by a factor 2-3) in parallel
to a reduction of the Average Revenue per User (ARPU) (by a factor 2.5-5). Instead, it is assumed
that the message size for the different application categories will remain the same, while the
mean time between mMTC-UE maintenance shall increase by a factor 3.

A. Current mMTC Solutions Landscape

There are a number of technologies currently available to support mMTC services in terrestrial
Low Power Wide Area Networks (LPWAN), and most of them have been captured in [2].
Hereafter, few of them are presented, either because they have been pioneering proprietary
answer to mMTC applications or representing the first standardized 4G technology enabling
low-cost multi-vendor solutions.
4

One of the first solution designed for IoT applications has been LoRa [3], a proprietary
waveform based on spread spectrum techniques operating in the Industrial, Scientific and Medical
(ISM) radio bands, and commercialized by Semtech Corporation. The typical data rate varies
from 300 bps to 30 kbps, and the trade-off is between the device throughput and the transmission
range. Higher spreading factor enables long distance communications at the cost of lower user
data rate, and vice versa.
Another proprietary solution addressing the mMTC market in the same time frame is Sigfox
[4], an ultra narrow band (UNB) signal transmission. In fact, again in ISM radio bands, this
technology has a transmission bandwidth of about 100 Hz, which enables very low power
consumption and inexpensive terminal manufacturing. Consequently, the user data rate is rather
low (i.e., approximatively 100 bps), however the communication range is in the order of 30-40
kilometers in good propagation conditions (e.g., rural environments).
Both technologies rely on a pure unslotted ALOHA protocol. ISM bands are generally un-
licensed, and therefore deployed in an uncoordinated fashion. This makes LPWAN networks
prone to unwanted interference on top of the network self-generated one. Since no interference
cancellation techniques have been adopted to resolve packet collisions, uplink channel collision
avoidance is based on asynchronous and uncoordinated transmission time, on random selection
of the transmission carrier frequency (among a set of tens of allowed values), and on choosing
different spreading factors (only applicable in LoRa networks).
LoRa and Sigfox LPWAN infrastructures are being deployed in many countries. These net-
works are to a certain extent scalable as their capacity can be increased by deploying additional
nodes (network densification). Certain mMTC devices can support multiple LPWAN solutions,
thus allowing for increased coverage and flexibility for the customer.
As far as the first standard in 3GPP aiming at LPWAN is concerned, NB-IoT is a narrow-
band technology made available around mid 2016 [5]. NB-IoT aims at enabling deployment
flexibility, long battery life, low device cost and complexity, and signal coverage extension. NB-
IoT uses single-carrier Frequency Division Multiple Access (FDMA) in uplink and Orthogonal
FDMA (OFDMA) in downlink, with 200 kHz bandwidth. The data rate is limited to 250 kbps
in downlink transmission and to 20 kbps in single-tone uplink communication.
5

B. 3GPP Standardization

The 5G standardization activity happens to a large extent in 3GPP, as it was the case for
previous cellular standard generations. The work on 5G is planned to take place in two releases,
of which the first one is Release 15 for MBB services, scheduled to be stable and all protocols
completed by mid 2018. Additional work will be done in Release 16 for mMTC inclusion and
NB-IoT enhancement, which will be completed by March 2020.
Specifically related to the satellite role in 5G, [6] reports a list of intended use cases and
services, which will benefit from the integration of terrestrial and satellite networks. Of course
concerning mMTC services, there are needs from transport, security, and agriculture players to
require global coverage and service continuity. A preliminary analysis of potential areas of impact
in 5G waveforms via satellite (including also the NB-IoT protocol stack) and the identification
of possible solutions are summarized in [7].

C. IETF Standardization

3GPP standardizes systems, architectures, and protocols specific to their networks and layers.
The Internet Engineering Task Force (IETF) works on Internet technologies, and a significant
number of key 5G protocols will be specified in the IETF, as it was already the case in 3G
(e.g. SIP/IMS, EAP-AKA and Diameter). Three relevant IETF protocols for IoT applications
are the standardized Constrained Application Protocol [8], HTTP/2 [9] and the new QUIC
transport protocol [10]. Additionally, the IETF IPv6 over LPWAN Working Group is currently
defining encapsulation and header compression mechanisms that will allow IPv6 packets to
be sent and received over LPWAN, such as SIGFOX, LoRa, WI-SUN and NB-IoT. This will
enable IoT applications to run transparently over different LPWAN radio access network (RAN)
technologies. The previous protocols have been defined to operate with constrained IoT devices
in constrained networks (e.g. high-latency, low-power, lossy). Their applicability to the satellite
environment is further analyzed in Sect. V.

III. S ATELLITE POTENTIAL AND CHALLENGES FOR M MTC SERVICE PROVISION

A. Global versus Regional Solutions

Typically, a GEO generates multiple beams to cover mMTC-UE service area to increase
the frequency reuse and to maximize the satellite antenna gain. The signals received by the
6

satellite user beams are amplified, filtered and translated in frequency by the satellite payload.
The “transponded” beams signals are then retransmitted towards the terrestrial gateway2 using a
feeder link multiplex and dedicated frequency band. The opposite happens for the signals flowing
from the gateway to the mMTC-UE. In short, the GEO satellite acts as a two-way transponder
linking mMTC-UEs to the gateway. The gateway receive chain is demodulating the mMTC
transmitted information packets and routing them to the terrestrial networks. Additionally, the
gateway modulates and transmits packets directed to the mMTC-UEs for commanding purposes
and takes care of the network signaling and resource management.
A different satellite system architecture for mMTC service provision is represented by Low
Earth Orbiting (LEO) constellations. In this case, unless an inter-satellite link is implemented,
the number of satellites and gateways stations required to provide a real-time global Earth
coverage is much larger and depends on the selected orbital parameters and the spacecraft antenna
covered surface [11], [12]. This is increasing the deployment and operational costs of the system.
Alternatively, if the message delivery latency is not an issue, one can reduce the number of LEO
satellites and select an inclined orbit, which ensures a good coverage of the Earth after a given
number of spacecraft revolutions. Gradual deployment is possible by launching more satellites
as the business grows. Furthermore, using regenerative payload with packet store and forward
capability allows to limit the number of terrestrial gateways to one.
Compared to GEO, LEO constellations provide a reduced transmission latency which, as
discussed in Sect. II, may be of relevance for mMTC services. Moreover, for mMTC-UEs
installed in non GEO line-of-sight (LOS) locations, the time variant geometry between the LEO
spacecraft(s) and the mMTC-UEs may help to provide some temporary LOS conditions allowing
packet delivery. Nevertheless, the LEO temporary link availability may lead to a noticeable
increase in the packet delivery time.
The LEO orbit is often assumed to reduce the mMTC-UE transmitted power requirement
compared to a GEO one. However, this consideration does not hold true in general. In fact,
what matters for sizing the mMTC-UE RF power is the actual bit rate transmitted and the
required signal-to-noise (SNR) for packet demodulation at the gateway site. Assuming that, as
it is typically the case, the feeder link thermal noise contribution is negligible compared to the
user link, it is easy to show that, to keep the same SNR at the gateway side, the satellite antenna

2
Performing the 5G NodeB functionalities.
7

diameter ratio shall be scaled with the inverse of the spacecraft to ground distance. This means
that a GEO orbiting at 35876 km will require an antenna diameter 24 times larger than a LEO
orbiting at 1500 km. This is why 12 to 18 meters deployable reflectors are nowadays used from
GEOs operating at L/S-band instead of a 0.5-0.8 m LEO antenna aperture. Incidentally, as shown
in Fig. 1, following this simple antenna sizing rule of thumb, the beam size on ground will be
the same, independently from the selected orbit.

Fig. 1. Example of GEO and LEO Earth coverage.

LEO orbit implies a sizable maximum Doppler frequency offset and rate proportional to the
carrier frequency and inversely proportional to the orbit height. The Doppler effects can be
largely mitigated at the gateway by simply pre-compensating the Doppler effect at the center of
beam (COB). After that, what remains is the differential Doppler between the current mMTC-UE
location and the COB.

B. Frequency Bands

Satellite mMTC systems can exploit a variety of frequency bands ranging from VHF to Ka-
band. The lower frequency bands such as VHF, UHF, L-, and S-band are particularly suitable
for mobile or easy to deploy mMTC-UE. This is because the UE antennas required for these
frequencies are of small size (whip or patch types) and with an omni-directional azimuth
pattern, thus not requiring specific satellite pointing. The use of omni-directional antennas is
8

also very suitable for LEO constellations whereby the satellite position is continuously moving.
Furthermore, the lower frequency bands, and in particular VHF/UHF, the higher level of indoor
penetration. The drawback is that, even in presence of LOS conditions, the omni-directional
antenna pattern collects a non negligible amount of multipath by the surrounding terrain or
buildings, leading to a Ricean type of fading.
Instead, when operating at Ku or Ka-band, to achieve the same SNR at the gateway with the
same mMTC-UE transmit power, the antenna aperture shall be increased proportionally to the
frequency scaling factor3 . This means that omni-directional type of antennas should be abandoned
in favor of more directive type of aerials, with the consequent need for satellite pointing. The
advantage is that the channel becomes very close to a Gaussian one (i.e., very large Rice factor).
The allocated satellite frequency bands are typically proportional to the carrier frequency, thus
the overall throughput supported is scaling with the carrier frequency adopted. In case of LEO
constellations or GEO operating at Ku/Ka-band and serving mobile users, the mMTC-UE shall
track the satellite by means of an electronically or mechanically steerable beam. This is typically
increasing the mMTC-UE cost.

C. Network Scalability Aspects

As discussed in Sect. II, mMTC typical traffic profile is characterized by infrequent small
packets making the mMTC-UE inactive for most of the time. To efficiently exploit the allocated
spectrum, a very large number of mMTC-UEs with low duty cycle shall be able to effectively
share the same radio resource. Therefore, it is fundamental to minimize the amount of signaling
the network is exchanging with the mMTC-UE when it is dormant. Ideally, the mMTC-UE
shall wake-up when a packet has to be transmitted, staying active the minimum time required to
deliver the packet and finally going back to dormant mode. The packet delivery time depends on
the physical layer configuration and protocol time overheads. This calls for an efficient random
access protocol which allows asynchronous (i.e., not time synchronized) transmission of packets
and the minimization of unicast signaling during active mMTC-UE status.

3
The use of a more directional antenna also helps in reducing interference to and from GEO satellites located close to the
serving one.
9

Following [13], the number Numax of mMTC-UEs which can be supported with an acceptable
quality of service is approximated by:
max
TMAC · Rs
Numax ' , (1)
Rb · da
max
where TMAC is the maximum achievable normalized Media Access Control (MAC) Random
Access (RA) throughput at the desired Packet Loss Ratio (PLR) value, Rs is the RA baud rate,
Rb represents the single mMTC-UE transmitted packet data rate, and finally 0 ≤ da ≤ 1 stands
for the average activity factor of the mMTC-UE. Looking at (1), we notice that for given user
bit rate, by increasing the channel baud rate Rs (i.e., the occupied bandwidth) the number of
supported users scales up linearly.
Another key aspect for network scalability is the probability of packet retransmission mini-
mization to avoid network congestion situations in particular when mMTC device transmission
time is correlated. To minimize the channel load fluctuations, it is desirable to achieve a high
level of traffic aggregation. A physical layer with a large processing gain (e.g. using spread
spectrum and very low FEC code rate), allows a higher traffic aggregation resulting in a lower
peak-to-average channel load ratio. This allows operating the channel close to the saturation
point (i.e. no need for large margins) maximizing the channel utilization efficiency [13]. When
the aggregated traffic remains very bursty by nature, e.g. it is time correlated, then it is required
that mMTC-UEs implement a congestion control mechanism to smooth the traffic and maintain
the channel load around a desired operating point, at the expense of an increased end-to-end
packet transmission delay. Typical techniques employed for congestion control are a p-persistent
algorithm, exponential back-off or a combination of the two.

D. Terminal Aspects

1) Cost: Cost reduction and miniaturization are two important requirements for mMTC-UEs.
For achieving these targets the use of integrated circuits for signal processing, digital/analogue
conversion, and possibly RF functions is pivotal. Software Defined Radio (SDR) can be useful
for improving functionality and performance of the devices during their lifetime through over
the air updates. Local oscillator needs to be cheap while providing acceptable performances in
terms of phase noise and frequency stability. Power amplifiers range typically spans from few
mW to less than 1 W of peak radio frequency power depending on the frequency band and
10

type of service. Finally, when possible, omni-directional patch antenna are used for lowering
production and installation cost.
2) Energy Efficiency: Achieving a high level of energy efficiency for mMTC-UEs is of
paramount importance when electrical power resources are scarce. This happens for systems
deployed in isolated regions where no electrical power grid is available or in mobile applica-
tions. The energy efficiency maximization can be accomplished operating along three resource
minimization axes: a) mMTC-UE activity time; b) energy consumption during the activity period
(i.e. transmission and reception phases); c) “sleep” mode energy consumption minimization. The
first target is obtained by designing an mMTC system whereby, ideally, the device should only
be activated for the minimum time required to send the information bits minimizing packet
retransmissions i.e. low RA PLR. Requirements b) and c) imply that signaling information
during dormant mMTC-UE status is minimized. When switched to active status, the mMTC-
UE should have a fast (re-)acquisition time and exploit a protocol that minimizes the energy
required for transmitting the information bits. This is a challenging task as the use of low-
cost oscillators and the mMTC-UE exposure to large temperature variations will extend the
range of downlink satellite carrier frequency search required by the mMTC-UE. A reduction
of acquisition time through more powerful signal processing will imply a power consumption
increase. The latter in turn lead to an increase of the modem energy consumption. An overall
system design optimization accounting for these contrasting requirements is a key mMTC-
UE design aspect. The low-quality mMTC-UE frequency reference together with the wanted
avoidance of frequent downlink reception to keep cost and energy consumption bounded is also
implying an increased gateway frequency search range thus higher modem complexity. The air
interface energy efficiency aspects are dealt with in Sect. IV.
3) Antenna: mMTC antennas can take very different shapes depending on their applications
and the frequency band adopted for the service provision. The most directive antennas are
parabolic dishes typically chosen for Ku- and Ka- band for the reasons explained in Sect. III-B.
Their main drawbacks are the occupied volume and the need for accurate pointing. At these
frequencies, when the mMTC-UE or the satellite is moving, dynamic antenna pointing is required,
and this is usually achieved by means of mechanical or electronic steering, which adds complexity
and cost. Another possibility is to use small antennas, such as patches, with low gain and low
directivity which represent a very suitable solution for L- or S- band. These aerials can be
low cost and do not require pointing. In any case, using circularly polarized antennas (both on
11

mMTC-UE and satellite) will avoid losses due to misalignment in the linear polarization case.

IV. A IR I NTERFACE A SPECTS

A. NOMA trends in 5G

To address the 5G mMTC challenging requirements, New Radio (NR) is moving from Or-
thogonal Multiple Access (OMA) to Non Orthogonal MA (NOMA) solutions whereby multiple
users efficiently share the same radio resource and exploit multi user detection (MUD) to
resolve collisions. A very good overview, categorization and comparative performance analysis
of currently proposed NOMA schemes for 5G NR is reported in [14]. In particular, the different
proposals from the key wireless industry players can be grouped in three main multiple access
(MA) categories: a) Codebook-based MA; b) Sequence-based MA; c) Interleaver/scrambler-based
MA. Codebook-based MA maps the user data packet stream in a multi-dimensional codeword
belonging to a codebook. The mapping is done in a way to achieve signal spreading and
introducing zero elements to mitigate inter user interference while keeping MUD complexity
bounded. Decoding is obtained through relatively complex iterative message passing algorithm.
Sequence-based MA exploits nonorthogonal complex number sequences to separate users sharing
the same spectrum and easing the MUD process. Affordable complexity linear Minimum Mean
Square Error (MMSE) combined with Successive or Parallel Interference Cancellation (SIC/PIC)
are proposed for the packet detection. Interleaver/Scrambler-based MA is exploiting interleave-
division to separate users sharing the same bandwidth. Some repetition/scrambling is also adopted
to spread the signals and achieve some interference averaging effect. Depending on the size of
the interleaved bit stream, simpler MMSE-SIC or more complex soft SIC decoding techniques
will be used.
Results reported in [14] show that depending on the propagation environment (outdoor/indoor)
and the MA loading factor the gain of the different proposals for 5G NR NOMA is varying
but always performing better than the OMA. The most complex MUD structures are typically
outperforming simpler linear MMSE-SIC type of detectors at least for high loading factors.

B. NOMA-like Satellite Solutions

Because of the inefficiency of terrestrial channel sense RA protocols due to the large prop-
agation delay [13], satellite systems have been pioneering the introduction of efficient NOMA
RA schemes minimizing signaling overhead and packet retransmissions. The introduction of
12

SIC to resolve packet collisions has been one of the main recent innovations leading to a
major breakthrough in satellite RA solutions. An exhaustive review of SIC-based RA techniques
discussed in the following is reported in [13].
Considering the satellite mMTC requirement for signaling overhead and mMTC-UE energy
consumption minimization, appealing MA solutions are Enhanced Spread-spectrum ALOHA
with SIC (E-SSA) and Asynchronous Contention Resolution Diversity ALOHA (ACRDA) [16].
As explained in [13], ACRDA requires the transmission of multiple packet replicas (typically one
packet repetition) in random slots of the same frame to solve the packet collisions. Instead, E-SSA
does not require packet to transmit replicas as the packet collisions can be resolved exploiting
the spreading sequence random time and carrier frequency offset. Both RA schemes allow time
asynchronous packet transmission achieving normalized throughput in excess of 1 bit/symbol
with PLR lower than 10−3 and three order of magnitudes better than classical ALOHA. As
shown in the following, in terms of energy efficiency, the E-SSA one is double compared to
ACRDA because it does not require systematic packet replica transmission to solve collisions4 .
Let define Nslot as the number of RA contention slots in a frame. Intrinsically, slotted RA schemes
such as ACRDA, are affected by a Nslot -times higher RF peak power demand compared to E-
SSA for which Nslot = 1 [13]. This drawback, potentially affecting the mMTC-UE cost, can be
alleviated using a Multi-Frequency version of the ACRDA scheme (MF-ACRDA) [13]. Using
multiple randomly selected transmission frequencies instead of time slots (i.e. using several lower
baud rate carriers rather than one higher baud rate carrier), allows to reduce the transmission
packet baud rate extending its time duration. In this way the mMTC-UE peak RF power can
be reduced to a level comparable to E-SSA. It is often assumed that spread-spectrum multiple
access systems occupy more bandwidth than non spread-spectrum systems. While this is true
for a single user transmission, this is definitely not the case for efficient RA protocols like
E-SSA, whereby the spectrum is shared by a multitude of mMTC devices. Furthermore, even
non SS RA schemes such as ACRDA or MF-ACRDA, require an increase of the occupied
bandwidth to share the spectrum among mMTC-UEs. This is achieved in a different manner
compared to E-SSA. For ACRDA each mMTC-UE transmits only for a fraction (Nrep /Nslot ) of
the frame duration with a baud rate Nslot -times higher than a single packet transmission will
require. Instead for MF-ACRDA, the number of available time slots is reduced to mitigate the

4
More precisely, Table I in [13] shows results for CRDSA and E-SSA. Following (13) it is easy to find that the energy
efficiency of ACRDA is the same as CRDSA.
13

packet baud rate. However, the number of sub-bands in which transmission can take place is
increased correspondingly. Summarizing, all RA protocols need bandwidth expansion to achieve
an efficient media sharing no matter which multiple access scheme is chosen.
More recently a Codebook-based MA solution exploiting soft SIC and specially designed Low
Density Parity Check Codes named Spread Asynchronous Scrambled Coded Multiple Access
(SA-SCMA) appeared in literature [15]. This technique achieves higher throughputs than E-SSA
and ACRDA at the expense of a higher complexity gateway demodulator. A drawback is that the
sparse LDPC code designed to ease the MUD convergence shows a large performance reduction
in terms of AWGN performance, thus reducing the RA power efficiency.

C. Energy Efficiency Considerations

In [16] it has been found that, operating the RA protocol at the maximum MAC load point
for which the PLR is negligible, the normalized energy efficiency5 ψen can be approximated by

ηPHY
ψen ≈ , (2)
N rep Nslot
where ηPHY is the physical layer stand-alone spectral efficiency measured in bits/symbol or
bits/chip, N rep represents the average number of packet replicas. It can be deduced that the
transmission energy efficiency is maximized reducing the number of packet replicas transmitted
and maximizing the single RA packet physical layer spectral efficiency divided by the number
of contention frame slots. Following Table I in [16], the highest energy efficiency is achieved
by E-SSA.
An important parameter affecting the mMTC-UE Direct Current (DC) to RF power conversion
efficiency ηe is the High Power Amplifier (HPA) operating point in terms of drive level. The
achievable ηe value is somewhat proportional to the HPA compression point at which one can
operate. Using (quasi)-constant envelope modulations like Quadrature Phase Shift Keying or
Continuous Phase Modulations are certainly beneficial and often adopted in practice.

D. Example of mMTC Satellite Networks Sizing

Table III provides examples of GEO regional (or multi-regional) and LEO global satellite
systems return link sizing exploiting existing or planned space segment capabilities. The GEO

5
Assuming the same mMTC-UE RF power and occupied bandwidth for all schemes considered.
14

Orbit GEO LEO


Coverage Regional Global
Frequency band Ku Ka L Ku L
Number of satellite beams 1 (global) 82 200 16 16
Number of beams colors 1 4 4 4 1
Number of satellites 1 1 1 648 48
Number of active satellites 1 1 1 194 14
mMTC-UEOEM price [e] 150 160 90 250 20
(includ. installation)
mMTC-UE Antenna size [m] 0.80 0.60 0.20 0.5 0.10
mMTC-UE EIRP[dBW] 36.2 34.0 -16.0 21.0 -23.0
Satellite slant range [km] 39000 39000 39000 1500 1500
Channel bandwidth [MHz] 5.0 5.0 0.2 1.0 1.25
mMTC-UE Peak bit rate [kbps] 42.7 42.7 0.47 8.5 0.33
Multiple access scheme E-SSA E-SSA E-SSA E-SSA E-SSA
Detection scheme Coherent Coherent Coherent Coherent Non-Coherent
Multiple access spectral efficiency [bps/Hz] 2.5 2.0 0.6 2.0 0.375
Aggregated throughput/beam/sat [Mbps] 12.5 10.0 0.12 2.0 0.46875
Max number of devices per channel/beam/sat
270,000,000 216,000,000 2,592,000 43,200,000 10,125,000
Max number of devices per MHz/beam/sat 54,000,000 43,200,000 12,960,000 43,200,000 8,100,000
Overall max number of devices per MHz/sat
54,000,000 885,600,000 648,000,000 172,800,000 129,600,000
Overall max number of devices per MHz 54,000,000 885,600,000 648,000,000 33,523,200,000 1,813,400,000
TABLE III
E XAMPLES OF M MTC RETURN LINK ( M MTC-UE- TO -G ATEWAY ) SATELLITE SYSTEM SIZING . F OR CONSTELLATIONS THE
NUMBER OF ACTIVE SATELLITES HAS BEEN ASSUMED TO BE IN AVERAGE 30% THE OVERALL NOMINAL NUMBER , N/A
STANDS FOR NOT APPLICABLE .

systems are operated at Ka, Ku, or L-band while the LEO systems are operated at Ku or L-band.
The table specifies key mMTC-UE characteristics and an indication of the maximum number
of devices per channel or MHz which can be supported by the system. In these calculations, it
is assumed that each mMTC-UE transmits a message of 50 bytes 10 times per day in average,
corresponding to a duty cycle of approximately 10-6 . It is interesting to observe that, as discussed
before, both GEO and LEO can satisfactorily support mMTC services with a reasonable OEM
cost and mMTC-UE power consumption. With a relatively small channel bandwidth, up to several
billions of mMTC devices can be served in just a MHz of band reusing existing GEO Ka-band or
LEO L-band space segment. The channel bandwidth can be gradually scaled up as the business
develops, to offer a larger capacity.

E. Terrestrial and Satellite mMTC Air Interface Convergence

From Sect. IV, it is clear that the NOMA solutions currently envisaged for 5G mMTC may
also be adapted for satellite networks to achieve economies of scale. One possible issue is the
overhead associated to maintain the mMTC-UE network synchronization which is required for
15

slotted RA, and which appears to be the preferred 5G option. Some adaptation of the mMTC
protocol may therefore be required although key NOMA technologies may be reused.

V. N ETWORK A SPECTS

The satellite network architecture is similar to the terrestrial one with the main difference
being the long distance between mMTC-UEs and the RAN. This is particularly true in the case
of GEO satellites, which at an altitude of 36000 km introduce a minimum one-way latency
between mMTC-UEs and RAN of 238 ms.
The direct use of unmodified mainstream protocols, such as HTTP/1 and TCP, is challenging
in constrained IoT devices (e.g. CPU, memory, power) and constrained networks (e.g. high-
latency, low-power, lossy). To this end, the IETF has developed lightweight application transport
protocols that are better suited for the transmission of small messages with low duty cycle in
constrained environments.
It is worth mentioning the Constrained Application Protocol (CoAP) [8], which is used in
LPWANs for IoT services. CoAP endpoints exchange messages according to a request/response
mode and the resources are accessed through a Uniform Resource Identifier (URI) similarly to
HTTP, but CoAP packets are much smaller (between 10 and 20 bytes) than HTTP/TCP flows
and run over UDP. It offers a reliable transmission mode with Stop-and-Wait Automatic Repeat
reQuest (ARQ) and an exponential back-off mechanism for retransmissions, but the protocol is
open to the implementation of a Go-Back-N ARQ to make a more efficient use of the available
resources and reduce the delivery delay in high-latency networks such as sattelite [17].
Additionally, the IETF has standardized HTTP/2 [9] which primary goal is to reduce latency
by enabling full request and response multiplexing, minimize protocol overhead via efficient
compression of HTTP header fields, and adds support for server pushed content. HTTP/2 runs
over TCP, but the IETF is currently standardizing HTTP/2 support over QUIC [10], a new
UDP-based, stream-multiplexing, always-encrypted transport protocol focused on minimizing
application latency. Besides, QUIC also provides data reliability and a pluggable congestion
control, which can be optimized for IoT scenarios.

VI. S UMMARY AND C ONCLUSIONS

In this paper we reviewed the possible role of satellite in the provision of mMTC 5G services,
from requirements at service and terminal level to examples of network dimensioning. We
16

addressed the key system design aspects and associated trade-offs. Finally, we have been showing
that current space segment is able to provide mMTC services complementing terrestrial networks’
coverage.

R EFERENCES

[1] C. Bockelmann, N. Pratas, H. Nikopour, K. Au, T. Svensson, C. Stefanovic, P. Popovski, and A. Dekorsy, “Massive
Machine-type Communications in 5G: Physical and MAC-layer solutions”, IEEE Communications Magazine, Vol. 54, Issue
9, pp.59-65, Sep. 2016.
[2] U. Raza, P. Kulkarni, and M. Sooriyabandara, “Low Power Wide Area Networks: An Overview”, IEEE Communications
Survey and Tutorials, 2nd Quarter 2017, vol. 19, pp. 855-873, 2017.
[3] F. Sforza, “Communications system”, US Patent 8’406’275, Mar. 2013. [available online:
https://www.google.com/patents/US8406275]
[4] C. Fourtet, and T. Bailleul, “Method for using a shared frequency resource, method for manufacturing terminals, terminals
and telecommunication system”, 2013. [available online: https://www.google.com/patents/US20130142191]
[5] Y.P.E. Wang, and et. al, “A Primer on 3GPP Narrowband Internet of Things”, IEEE Communications Magazine, vol. 55,
pp. 117-123, Mar. 2017.
[6] 3GPP TR 22.822, “Study on using Satellite Access in 5G; Stage 1 (Release 16)”, June 2018.
[7] 3GPP TR 38.811, “Study on New Radio (NR) to support Non-Terrestrial Networks; (Release 15)”, June 2018.
[8] IETF RFC 7252, “The Constrained Application Protocol (CoAP)”, June 2014.
[9] IETF RFC 7540, “Hypertext Transfer Protocol Version 2 (HTTP/2)”, May 2015.
[10] IETF draft-ietf-quic-transport-08, “QUIC: A UDP-Based Multiplexed and Secure Transport”, December 2017
[11] R. A. Wiedeman, and A. J. Viterbi. “The Globalstar Mobile Satellite System for Worldwide Personal Communications”,
Proc. 3rd Int. Mobile Sat. Conf., (IMSC), vol. 93, pp. 291-296. 1993.
[12] “Iridium M2M”, https://www.iridium.com/solutions/m2m.
[13] R. De Gaudenzi, O. del Rio Herrero, G. Gallinaro, S. Cioni, and D.P. Arapoglou, “Random access schemes for satellite
‘networks, from VSAT to M2M: a Survey”, International Journal of Satellite Communications and Networking, Wiley
International Journal of Satellite Communications and Networking, Dec. 2016, doi:10.1002/sat.1024
[14] H. Kim, Y.-G. Lim, C.-B. Chae, and D. Hong,“Multiple Access for 5G New Radio: Categorization, Evaluation, and
Challenges”, https://arxiv.org/pdf/1703.09042.pdf
[15] N. Becker, S. Kay, L. Lee,and M. Eroz, “Spread Asynchronous Scrambled Coded Multiple Access (SA-SCMA) A New
Efficient Random Access Method”, IEEE 2016 Global Communications Conference (GLOBECOM),Washington, DC, USA,
December 4-8 2016.
[16] A. Mengali, R. De Gaudenzi, C. Stefanovic, “On the Modeling and Performance Assessment of Random Access with
SIC”, IEEE Jour. on Sel. Areas in Communications, Vol. 36, Issue 2, Feb. 2018, pp. 292-303.
[17] M. Bacco, M. Colucci, A. Gotta, “Application Protocols enabling Internet of Remote Things via Random Access Satellite
Channels”, 2017 IEEE International Conference on Communications (ICC), May 21-25 2017, Paris, France.

View publication stats

You might also like