Content uploaded by Jordi J. Gimenez
Author content
All content in this area was uploaded by Jordi J. Gimenez on May 06, 2019
Content may be subject to copyright.
Submitted to IEEE Transactions on Broadcasting
1
Abstract—3GPP LTE eMBMS Release 14, also referred to as
FeMBMS (Further evolved Multimedia Broadcast Multicast
Service) or EnTV (Enhanced TV), is the first mobile broadband
technology standard to incorporate a transmission mode
designed to deliver Terrestrial Broadcast services from
conventional High Power High Tower (HPHT) broadcast
infrastructure. With respect to the physical layer, the main
improvements in FeMBMS are the support of larger inter-site
distance for Single Frequency Networks (SFN) and the ability to
allocate 100% of a carrier’s resources to the broadcast payload,
with self-contained signaling in the downlink. From the system
architecture perspective, a receive-only mode enables free-to-air
(FTA) reception with no need for an uplink or SIM card, thus
receiving content without UE registration with a network. These
functionalities are only available in the LTE Advanced Pro
specifications as 5G New Radio (NR), standardized in 3GPP from
Release 15, has so far focused entirely on unicast. This paper
outlines a physical layer design for NR-MBMS, a system derived,
with minor modifications, from the 5G-NR specifications, and
suitable for the transmission of linear TV and radio services in
either single-cell or SFN operation. The paper evaluates the NR-
MBMS proposition and compares it to LTE-based FeMBMS in
terms of flexibility, performance, capacity and coverage.
Index Terms—MBMS, eMBMS, FeMBMS, DTT, MBSFN, SC-
PTM, LTE, 5G, New Radio, NR
I. INTRODUCTION
errestrial Broadcast, as a 3GPP use case, was first
addressed in LTE Advanced Pro 3GPP Release (Rel-) 14
This work was partially supported by the European Commission under the
5G-PPP project 5G-Xcast (H2020-ICT-2016-2 call, grant number 761498).
The views expressed in this contribution are those of the authors and do not
necessarily represent those expressed in the 5G-Xcast project or the author’s
organizations.
Parts of this paper have been published in the Proceedings of the IEEE
BMSB 2018, Valencia, Spain.
J.J. Gimenez and C. Menzel are with the Institut für Rundfunktechnik
GmbH, Floriansmuhlstrasse 60, 80939 Munich, Germany (e-mail:
{jordi.gimenez, christian.menzel}@ irt.de).
J.L. Carcel is with Samsung Electronics R&D UK, Communication House
South Street, Staines-Upon-Thames TW18 4QE, United Kingdom (e-mail:
jose.cervera@samsung.com).
M. Fuentes, E. Garro and D. Gomez-Barquero are with the Univesitat
Politècnica de València, Camí de Vera SN, 46022, Valencia, Spain (e-mail:
{mafuemue, edgarcre, dagobar}@iteam.upv.es).
S. Elliott is with BBC Distribution and Business Development, White City
Place, 201 Wood Lane London, W12 7TQ, England, (e-mail:
simon.elliott@bbc.co.uk).
D. Vargas is with BBC Research and Development, The Lighthouse, BBC
Broadcast Centre, 16 Alliance Court, Alliance Road, Acton, London W3 0RB,
England, (e-mail: david.vargas@bbc.co.uk).
in which the Multimedia Broadcast Multicast Service
(MBMS) system was enhanced to operate in a dedicated mode
for the delivery of linear broadcast services (i.e.radio and TV),
fulfilling a wide set of requirements input by the broadcast
industry [1]. 3GPP’s Enhancements for TV (EnTV) study item
proposed several enhancements resulting in the FeMBMS
(further evolved MBMS) of Rel-14. In order to leverage the
well-established and proven LTE ecosystem, it was decided to
base FeMBMS on the pre-existing LTE Advanced Pro
specifications with enhancements being made as necessary in
order to fulfill the requirements. Enhancements made to the
system architecture comprise: (i) the xMB interface through
which broadcasters can establish the control and data
information of audio-visual services; (ii) a new Application
Programming Interface (API) for developers to simplify access
to eMBMS procedures in the User Equipment (UE); (iii) the
support of multiple media codecs and formats; (iv) a
transparent delivery mode to support native content formats
over IP without transcoding (e.g. reusing existing MPEG-2
Transport Streams and compatible equipment); (v) the support
of shared eMBMS broadcast by aggregating different eMBMS
networks into a common distribution platform; and (vi) the
receive-only mode (ROM), which enables devices to receive
broadcast content with no need for uplink capabilities, SIM
cards or network subscriptions – i.e. free-to-air reception.
From the radio layer point of view the most significant
enhancements are: (i) the possibility to establish dedicated
FeMBMS carriers that allocate up to 100% of the radio
resources to Terrestrial Broadcast (i.e. with no frequency or
time multiplexing with unicast resources in the same frame),
self-contained signaling and system information in the
downlink; (ii) a new, reduced overhead subframe containing no
unicast control region; and (iii) the support of larger inter-site
distances in SFN (Single Frequency Networks) reaching higher
spectral efficiency with a new numerology – 1.25 kHz
subcarrier spacing (SCS) and 200 µs cyclic prefix (CP). The
new numerology changes are the most significant as the longer
OFDM symbol duration, occupying one subframe, made it
necessary to design a new subframe structure, known as the
CAS (Cell Acquisition Subframe), to allocate the
synchronization and control channels, transmitted with much
reduced periodicity (one in every forty subframes) [2].
These changes led to a system like other Digital Terrestrial
Broadcast systems such as DVB-T/T2 [3], ATSC 3.0 [4] or
DAB/DAB+ [5]. In addition to broadcast content, mobile
broadband subscribers who have a SIM card can enjoy
enriched service offerings when combined with independent
unicast for interactivity, in a similar way to conventional
5G New Radio for Terrestrial Broadcast:
A Forward-Looking Approach for NR-MBMS
Jordi Joan Gimenez, Jose Luis Carcel, Manuel Fuentes, Eduardo Garro, Simon Elliott, David Vargas,
Christian Menzel and David Gomez-Barquero
T
Submitted to IEEE Transactions on Broadcasting
2
HbbTV (Hybrid Broadcast Broadband TV) sets [6]. The
introduction of a ROM and the new framing and numerology
options may make FeMBMS suitable for use with conventional
broadcast infrastructure (including high, medium and low
power sites).
A further study item in 3GPP Rel-16 [7] has evaluated the
ability of FeMBMS to support SFN of cells with coverage radii
of up to 100 km (implying even longer CP) and mobile
reception with speeds up to 250 km/h (large SCS). A wider
range of numerologies, supporting multiple network
topologies, capacity improvements from longer symbol
durations (which reduce CP overheads), new reference signals
(RS) and greater bandwidth occupancy were also in the scope
of the study. The benefits of time interleaving [8] and LDM
(Layered Division Multiplexing) [9], also known as MUST
(Multiuser Superposition Transmission) [10], were also taken
into consideration. The signal acquisition and synchronization
procedures were also evaluated as the existing numerology
mismatch between data and control channels for large SFNs
may lead to coverage issues as reported in [11] and [12]. Based
on the findings of the Study Item, a Work Item may then
standardize further improvements in Rel-16 while taking into
account practical considerations such as implementation
complexity and performance.
In parallel, 3GPP is standardizing 5G New Radio (NR) and
5G Core (5GC) specifications, with a new and more efficient
radio layer and flexible system architecture. However, NR Rel-
15 and Rel-16 have so far focused on unicast. After the
RAN#79 plenary meeting, the broadcast work was split into
two tracks, one to design a mixed unicast/multicast/broadcast
mode for NR and another for LTE-based dedicated Terrestrial
Broadcast. Although the latter has become known as LTE-
based 5G Terrestrial Broadcast, the system is not based on 5G-
NR.
Several attempts have been made to start a work item for
NR-MBMS, but these were not sufficiently supported to take
them any further. [13], for example, proposed the introduction
of extended CPs based on NR numerology, without success.
Nevertheless, work on NR looks set to continue into 2019,
bringing an opportunity to introduce new functionalities for
Terrestrial Broadcast operation with enhanced flexibility and
scalability over LTE-based specifications which are subject to
legacy constraints.
This paper outlines a physical layer design for a new NR-
MBMS system based on an extension of NR Rel-15. The
design has little impact on the existing unicast mode, achieved
by considering Terrestrial Broadcast services as a configuration
option in which, simply, up to 100% of the 5G mixed mode
carrier resources may be allocated to linear TV/radio services.
The design is based on flexible numerology, framing and
bandwidth configurations in order to cater many different types
of networks ranging from single cell deployments to
nationwide SFN (the latter requiring a more complex design),
and reception environments from fixed rooftop to mobile. The
proposed design has been evaluated in terms of Doppler
performance, coverage, SFN echo tolerance and acquisition of
system information and synchronization. As broadcast is not
yet part of 5G-NR, several assumptions have been made,
considering that other parts of the NR-MBMS system may
enable the expected functionalities.
The rest of the paper is organized as follows: Section II
provides a brief introduction of the physical layer
characteristics of FeMBMS that are relevant for Terrestrial
Broadcast. Section III then introduces the details of 5G-NR
Rel-15 that have been further developed to support Terrestrial
Broadcast in NR-MBMS. Section IV evaluates and compares
FeMBMS and NR-MBMS solutions. Section V concludes the
paper.
II. BACKGROUND: FEMBMS PHYSICAL LAYER DESIGN FOR
5G TERRESTRIAL BROADCAST
Since its introduction in LTE Rel-9, eMBMS has generally
been associated with SFN operation in LPLT cellular networks
with the Multicast Broadcast SFN (MBSFN) mode. From Rel-
13, eMBMS added another type of radio bearer, known as SC-
PTM (single-cell point-to-multipoint), that does not offer SFN
capability but uses the unicast physical data channel with the
regular CP to withstand echo delays due to multipath. Both
types of bearers can be configured for Terrestrial Broadcast in
FeMBMS (i.e. the allocation of up to 100% of the radio
resources) enabling their respective functionalities. Their main
physical layer characteristics are explained below.
A. Numerology options for LTE FeMBMS
Different numerology options are available for FeMBMS
operation.
Within an MBSFN area, all the constituent transmitters
must deliver the same content at the same time, on the same
frequency. The CP appended to the beginning of the OFDM
symbol avoids Inter-Symbol Interference (ISI), provided that
all signals are received with relative delays shorter than the CP
duration. As set out in Table I, FeMBMS has three values of
extended CP (TCP): 16.66 µs, 33.33 µs and 200 µs. The latter
requires a 1 ms OFDM symbol duration – an entire subframe
1
.
The SCS (=1/ TU), and corresponding OFDM symbol
duration (TU) for these three CPs are also shown in Table I.
Importantly, the new, much longer 200 µs CP of Rel-14 has
extended the maximum inter-site distance (ISD) in a network
to around 60 km, thus potentially supporting MBSFN areas in
HPHT networks.
In all cases the overhead due to the CP is 20%. The reason
for the comparatively large overhead comes from a design
based on mobile reception, where a compromise is required
between capacity (that would increase by reducing the ratio
1
In LTE, a 10ms frame is divided into 10 subframes, each of 1 ms duration.
In turn, each subframe is divided into two 0.5 ms slots. Each slot comprises one
Resource Block (RB) in time. In frequency, each RB occupies 180 kHz,
equivalent to 12 consecutive OFDM subcarriers for the normal CP
configuration (15 KHz SCS).
TABLE I. NUMEROLOGY OPTIONS IN FEMBMS
Type
(kHz)
SCRB
OFDM
symbols
per SF
TCP
(µs)
TU (µs)
ISD
(km)
SC-PTM
15
12
14
4.7/5.1
66.7
1.4
MBSFN
15
12
12
16.7
66.7
5
7.5
24
6
33.3
133.3
10
1.25
144
1
200
800
60
Note that a Resource Block (RB) in LTE is 180 kHz wide in frequency and 1 slot long
in time. SCRB=Subcarriers per Resource Block, SF=Subframe, TCP=CP duration,
TU=useful OFDM symbol duration, ISD =SFN Inter-Site Distance.
Submitted to IEEE Transactions on Broadcasting
3
TCP / TU) and resilience to Doppler spread (greater robustness
with higher ).
SC-PTM does not support SFN, and may only be
configured with the conventional ‘unicast’ CPs designed to
cope with multipath. The overhead for the normal CP is
around 7%. Another difference between SC-PTM and
MBSFN is that SC-PTM data is scheduled using the same
physical data channel as for unicast traffic (PDSCH – Physical
Downlink Shared Channel) while MBSFN data is scheduled
using an ad-hoc physical channel that enables the extended CP
(PMCH – Physical Multicast Channel).
B. Reference Signals
The MBSFN modes have a set of reference signal (MBSFN
RS) patterns that are denser in the frequency domain compared
to the standard unicast patterns in order to help receivers cope
with the higher frequency selectivity caused by echoes received
from distant transmitters in SFNs. Each base-station (i.e. cell)
belonging to an MBSFN area transmits the same MBSFN
reference signal pattern at the same time-frequency position in
the PMCH.
As shown in Fig. 1, for the MBSFN 15 kHz SCS, reference
symbols are inserted in every other sub-carrier in OFDM
symbols 3, 7 and 11 of each sub-frame, with a single subcarrier
offset in the OFDM symbol number 7. For 7.5 kHz SCS, one
reference signal is inserted in every four sub-carriers in OFDM
symbol numbers 2, 4 and 6 of each sub-frame, with a two
subcarrier offset in OFDM symbol 4. In the 1.25 kHz variant,
one reference signal is allocated every six subcarriers, with an
offset of 3 subcarriers between odd and even sub-frames.
With respect to multipath or echoes – either artificial or
natural – the frequency spacing between reference signals
determines the length of delay up to which the channel may be
correctly equalized when using time-frequency interpolation.
Delays up to the duration of the equalization interval (EI) may
be tolerated.
The EI is calculated assuming that the receiver is able to
perform time and frequency interpolation. A factor of 57/64 is
considered to account for realistic receiver implementation
[14]. According to the frequency separation between reference
signals Df, the EI (57/64*TU/Df) for MBSFN subframes is 59.3
µs for {15, 7.5} kHz and 237.5 µs for 1.25 kHz.
The overheads due to RS are 12.5% (15 and 7.5 kHz) and
16.6% (1.25 kHz). Note that in the figure, is the length of
the RS pattern in OFDM symbols.
For SC-PTM, cell-specific reference signals (C-RS) are
used, for channel estimation of the PDSCH, with the patterns
shown in Fig.1 It can be seen that they are sparser, and
therefore overheads reduce at 4.7% (15 kHz normal CP) and
5.5% (15 kHz extended CP).
C. Control Channels – Cell Acquisition Subframe (CAS)
Previous releases of LTE eMBMS defined MBSFN frames
with up to 60% broadcast resource allocation (up to 6 out of 10
subframes in each frame could be allocated to broadcast as two
were permanently designated for paging and two more for
synchronization). The capacity not allocated to broadcast could
be allocated to unicast traffic. The latest modifications in
FeMBMS enabled the configuration of up to 80% broadcast
resource allocation and also a dedicated carrier with almost
100% broadcast allocation (97.5%), by minimizing the
signaling required for synchronization, acquisition and system
information and moving it into the newly defined Cell
Acquisition Subframe (CAS) that is transmitted once every 40
subframes (i.e. 2.5% signaling overhead). SC-PTM carriers
are more flexible as they can multiplex broadcast data with
higher time and frequency granularity – as FDM is possible,
there is no need to dedicate complete subframes to broadcast.
The CAS is formed of the following physical signals and
channels:
▪ PSS (Primary Synchronization Signal): symbol timing
and partial physical cell identity (PCI) information
▪ SSS (Secondary Synchronization Signal): frame timing,
transmission mode, CP duration and complete PCI.
▪ CS-RS (Cell-Specific Reference Signal): amplitude and
phase reference for channel estimation.
▪ PBCH (Physical Broadcast Channel): Master
Information Block (MIB), number of TX antennas,
downlink BW and system frame number.
▪ PCFICH (Physical Control Format Indicator Channel):
number of OFDM symbols used for control for each
subframe.
▪ PDCCH (Physical Downlink Control Channel):
downlink control indicator (DCI), transmission
parameters and scheduling.
Fig. 1. Reference signals for MBSFN subframes and unicast subframes with
different numerologies.
SC-PTM Normal 15 kHz SC-PTM Extended 15 kHz
MBSFN 15kHz
MBSFN 7.5kHz MBSFN 1.25kHz
Type
D
f
D
t
SC
-
PTM N
15 3 7
SC-
PTM E
15 3 6
MBSFN 15 1 8
7.5 2 4
1.25 3 2
Submitted to IEEE Transactions on Broadcasting
4
▪ PDSCH (Physical Downlink Shared Channel): System
Information Blocks (SIBs).
The CAS may only be configured with numerologies
defined for unicast (i.e. 15 kHz SCS with or without extended
CP). From the point of view of Terrestrial Broadcast, the CAS
acts as a preamble containing physical layer signaling (similar
to P1 and P2 symbols in DVB-T2 or bootstrap and preambles
in ATSC 3.0). It is important to note that each service
transmitted in MBMS is identified with a corresponding
Temporary Mobile Group Identifier (TMGI)
2
. With the
information obtained from the CAS and the TMGI, the UE can
disclose an on-going MBMS session and receive the service.
When SC-PTM is configured, data is scheduled over the
PDSCH according to a specific Radio Network Temporary
Identifier (RNTI), reusing the unicast mechanism to schedule
user data. For broadcast a Group RNTI (G-RNTI) is used,
which is then mapped to a TMGI.
D. Identified limitations of FeMBMS
Although the LTE FeMBMS improvements standardised in
Rel-14 allow the delivery of linear broadcast services (such as
TV and radio) with a dedicated broadcast mode, the
specifications lack flexibility in parameters such as the CP, TU
duration, and FFT size to adapt the system to the wide range of
broadcasting network topologies deployed around the world:
• For fixed-rooftop reception, HPHT SFN networks with
very large ISD (circa 150 km) may benefit from CPs
greater than 200 µs.
• For mobile environments with high speed requirements
(e.g. 250 km/h) in sparse LPLT networks, such as in rural
areas, numerologies with a CP between 33 and 200 µs
options may be beneficial.
Large inter-site distances and high mobility with speeds up
to 250 km/h are requirements set out in 3GPP [1] for the 5G
physical layer to deliver broadcast and multicast services.
Another limitation is the fixed CP overhead of 20% which
cannot be modified to suit static or mobile scenarios. In fixed
rooftop environments where the Doppler spread does not
impose a significant limitation, narrower SCSs could be used.
These would increase TU in order to lower the CP overhead for
a fixed absolute CP length. Even more, FeMBMS, and in
particular SC-PTM, has not been optimized for single-cell or
MFN networks, where overheads could be reduced.
Another important aspect is the performance of the CAS
subframe due to the aforementioned numerology mismatch (i.e.
the CAS can only use CP lengths that are shorter than the data
subframes). Thus, CAS subframes may limit the final system
performance in networks where significant echoes are received
with delays greater than 16 µs.
Furthermore, multiple input multiple output (MIMO)
techniques with spatial multiplexing are not supported in
FeMBMS [2], which limits the potential maximum capacity of
the system.
Potential inefficiencies are also identified. Service
information in MBMS is carried over multiple control
channels. SIBs are not only transmitted in the PDSCH of the
CAS but also in the logical MCCH (Multicast Control
2
Note that for ROM receivers, a range of TMGIs with MCC = 901 and
MNC = 56 is standardized so that there is no need to register to an operator.
Channel) and SC-MCCH channels. These are multiplexed with
service data in the MTCH (Multicast Traffic Channel) or SC-
MTCH, and transmitted in the PMCH or the PDSCH,
respectively for MBSFN or SC-PTM. In addition to a distinct
downlink channel in MBSFN, service information depends on
the type of carrier and deployment.
III. A NEW PHYSICAL LAYER DESIGN FOR 5G TERRESTRIAL
BROADCAST BASED ON NEW RADIO (NR-MBMS)
5G-NR offers a more flexible and scalable design than LTE
in order to support a wider range of use cases requirements,
including an extensive range of frequency bands and
deployment options. However, NR Rel-15 and Rel-16 have
considered only unicast. Transmission modes and core
functionality do not yet support broadcast or multicast.
Opportunity therefore exists in NR to define transmission
modes suitable for Terrestrial Broadcast with fewer constraints
compared with FeMBMS, which is based on well-established
LTE specifications.
This section explains the main characteristics of 5G-NR
already specified for unicast and explores their potential
extension for Terrestrial Broadcast.
A. 5G-NR physical layer design
5G-NR is based on a CP-OFDM solution similar to LTE.
The main new feature is that the 5G waveform is combined
with a scalable numerology that enables radio resource
allocation over different frequency bands. The SCS is scaled
according to 15×2μ kHz, where 15 kHz is taken as the base
SCS (as for LTE) and 2μ generates additional numerology
options. Five different numerologies are defined with SCS
from 15 kHz to 240 kHz (from μ=0 to 4). Note that the
parameter μ is only defined for positive integer numbers. The
possible values vary with the frequency band in order to cope
with Doppler spread (higher at higher frequencies), and to
extend the bandwidth, greater at high frequencies. This scalable
method permits the different numerology options to be aligned
in the time domain as extensions of basic slots and OFDM
symbols.
The most significant change with respect to LTE is that one
slot always contains 14 OFDM symbols for all different SCS
values. Therefore, the number of slots per subframe (and
frame) increases for wider SCS and an RB is defined as 12
subcarriers in only one OFDM symbol in the time domain
(rather than one slot as per LTE). With respect to the system
bandwidth, 5G-NR brings the possibility to configure larger
bandwidth carriers (e.g., 100 MHz or 400 MHz) than LTE.
Note that different numerologies can be multiplexed within the
same NR carrier bandwidth both using Time (TDM) and
Frequency Division Multiplexing (FDM) thanks to a new
concept called Carrier Bandwidth Part (CBP). Using CBP it is
possible to define groups of consecutive RBs, including
different numerologies, over the same carrier. A maximum of 4
BW parts can be specified, which may be enough for the
purpose of multiplexing different numerologies from a single
wideband transmitter.
5G NR also introduces some variations with respect to
control channels. In particular, the frame structure avoids the
mapping of control channels across the full carrier bandwidth,
Submitted to IEEE Transactions on Broadcasting
5
giving more flexibility than LTE and enabling the selection of
the position of control channels across the frequency domain.
In this case, PSS, SSS and PBCH are transmitted in a
synchronization signal block (SS Block) occupying a concrete
number of RBs per CBP. The PDCCH carries DCI signalling
mapped in the CORESETs.
The reference signals in 5G NR has suffered changes
compared to LTE. For instance, there is not Cell specific
Reference Signal (C-RS) and a new Reference Signal PTRS
has been introduced for Time/Frequency tracking. The signals
used for channel estimation are the Demodulation Reference
Signals (DMRS).
Another aspect to consider is channel coding. For user
plane data, Low-Density Parity Check (LDPC) codes are
adopted. Control channels are coded by means of a new
channel coding technique based on the channel polarization
concept [15]. Channel polarization brings a method for
constructing capacity-achieving codes for binary input
symmetric memoryless channels, as opposed to capacity
approaching. It has been demonstrated that 5G-NR Polar codes
outperform the Tail-Biting Convolutional Codes (TBCC) used
in LTE control channels.
A. Design of NR-MBMS for Terrestrial Broadcast
The starting point taken in this document to development an
NR-based Terrestrial Broadcast system was to begin by
defining a number of new transmission modes in order to
provide flexibility to cater for a wide range of deployment
scenarios. With this starting point in mind, the following main
design principles were then taken into account:
▪ Minimization of the footprint with respect to unicast
transmission and scheduling processes. Reuse of unicast
scheduling mechanism with one RNTI per service.
▪ Numerology options adequate for diverse scenarios
including Multiple Frequency Network (MFN) and SFN
configurations and topologies from low-power low-tower
(LPLT) up to high-power high-tower (HPHT), with
different inter-site distances.
▪ Transmission modes adequate for mobile (improving
Doppler performance for high speed reception) and static
reception (reducing capacity overheads at the expense of
Doppler tolerance).
▪ Leveraging of 5G-NR bandwidth configuration and
spectrum utilization efficiency to transmit large
bandwidth signals instead of bandwidth-limited (to a few
MHz) carriers, if desired.
▪ Efficient multiplexing in time and frequency domain of
local, regional and nation-wide services targeting mobile
and fixed reception.
1) Waveform and Numerology options
The diverse nature of networks that may be used for
transmitting Terrestrial Broadcast services (with network
topologies ranging from LPLT to HPHT and from single-cell,
MFN to nationwide SFN coverage areas) make it highly
desirable to provide a range of new numerologies to better
cater for the different types of transmission networks that could
be used. Table II provides a set of numerologies that may be
considered for Terrestrial Broadcast operation with different
SCS, overhead and CP duration.
The reference numerology with μ=0 (A) is already suitable
for single-cell or MFN Terrestrial Broadcast operation,
particularly from LPLT networks.
An extensive set of numerologies can be derived for SFN
operation. Reference [13] introduced the concept of NR-
MBMS with extended CP by using a negative μ factor. This
leads to SCS of 7.5 kHz (μ=-1), 3.75 kHz (μ=-2), etc.
Examples derived with this method are B, C, D or F,
highlighted in bold. However, although this mechanism may be
useful to derive appropriate extended numerologies for LPLT
deployments (note the similar numbers as for LTE), it is
impractical to derive numerologies for SFN deployments with
large-ISD (e.g. HPHT). These would require of longer OFDM
symbol durations that may complicate implementation due to
leading to non-integer multiples of current NR subframes.
Moreover, following the same principle as in LTE, i.e.
targeting mobile reception, the options are again limited to a
few modes with 20% CP overhead.
Other numerologies may be derived following the method
in [16] by re-defining the number of SCs per RB. Although
initially proposed for LTE, the same mechanism can be applied
to 5G-NR as 15 kHz SCS is also the standard numerology as
well as 25 RBs per 5 MHz carrier (4.5 MHz effective
bandwidth). Equation 1 provides the means to derive the new
number of SC per RB as a function of the OFDM symbol
duration and a CP fraction, which needs to be selected so that
the number of SC per RB is an integer.
(1)
By this method it is possible to derive multiple
combinations of TCP and TU. Note that OFDM symbol
durations of 0.5, 1, 2, 2.5, 5 and 10 ms are particularly
interesting as an integer number of them fit into a 10 ms frame.
Therefore, it is possible to select the appropriate combinations
according to the deployment scenario and the receiving
environment.
For instance, it is possible to select numerologies with
acceptable mobility performance (i.e. wide SCS) and low
capacity overhead for single-cell or MFN configuration (no
TABLE II. NUMEROLOGY OPTIONS FOR 5G TERRESTRIAL BROADCAST
µ
(Hz)
TU (µs)
CP
Fraction
TCP
(µs)
TS
(ms)
SCRB
ISD
(km)
A
0
15000
66.67
~7%
4.7/5.1
0.07
12
1.4
B
0
15000
66.67
20%
16.67
0.08
12
5
C
-1
7500
133.33
20%
33.33
0.17
24
10
D
-2
3750
266.67
20%
66.67
0.33
48
20
E
-
2500
400.00
20%
100.00
0.50
72
30
F
-3
1875
533.33
20%
133.33
0.67
96
40
G
-
1250
800.00
20%
200.00
1.0
144
60
H
-
625
1600.00
20%
400.00
2.0
288
120
I
-
3333
300.00
10%
33.33
0.33
54
10
J
-
2045.45
488.88
2.22%
11.11
0.50
88
3.3
K
-
1022.72
977.78
2.22%
22.22
1.0
176
6.6
L
-
511.36
1955.56
2.22%
44.44
2.0
352
13.2
M
-
416.67
2400
4%
100
2.5
432
30
N
-
208.33
4800
4%
200
5.0
864
60
O
-
104.67
9600
4%
400
10.0
1728
120
p
-
217.39
4600
8%
400
5.0
828
120
SCRB=Subcarriers per Resource Block
Submitted to IEEE Transactions on Broadcasting
6
need for a very large CP to cope with SFN echoes),
numerologies with a large CP for moderate speed (e.g. those
with 20% CP overhead), or numerologies with a large CP for
SFN and static reception (with e.g. 4% CP overhead).
Numerology E could be useful for LPLT SFNs with a
compromise between SFN gain in short/medium ISD
deployments, while retaining high mobility. Numerology G
corresponds to the one specified in FeMBMS.
It should be noted that although decreasing the TCP/TU ratio
reduces CP overhead, increasing the FFT size may cause
significant impact in receiver complexity, which would need to
be considered, along with other practical aspects of the receiver
design.
Regarding reference signals, DMRS defined for unicast
may be used for single-cell or MFN deployments, therefore
with no changes. For SFN the most demanding channel may
require the definition of new RS, in particular denser in the
frequency domain. Also for SFN, a common DMRS
scrambling sequence for the stations involved in an SFN area is
required.
2) Bandwidth, Multiplexing and Spectrum Utilization
For Terrestrial Broadcast deployments in frequencies below
1 GHz, NR frequency range FR1 provides the following
options regarding bandwidth (frequency range FR1): 5, 10, 15,
20, 25, 40, 50, 60, 80 or 100 MHz. The bandwidth utilization in
LTE was set to 90 percent in order to limit out-of-band
emission produced by the slow decay of OFDM spectrum. This
means that only 100 PRBs (18 MHz) can be transmitted per 20
MHz bandwidth carrier. In NR, for the same case 106 PRBs
(19.08 MHz) can be configured with 95.4% bandwidth
utilization. Note that with numerology 15 kHz a 50 MHz
carrier can be configured with 270 RBs (97.2% bandwidth
utilization). For 10 MHz the bandwidth utilization is 93.6% (52
RBs).
Note that the total allocated bandwidth can be extended via
Carrier Aggregation, which in NR supports the bundling of up
to 16 carriers. The combination of large bandwidth
transmission with robust modulation and coding could also
enable the introduction of Wideband Broadcasting transmission
as presented in [17].
The potential application of the CBP concept to Terrestrial
Broadcast distribution is proposed next. In this case, a single
wideband carrier can multiplex not only services addressing
different reception conditions, but also different coverage areas
and network deployments. Fig. 2 shows a wideband carrier that
allocates three different CBPs, each one with different
numerology (see OFDM A, B and C). For instance, part of the
Fig. 2. The figure presents how the concepts of multiplexing numerologies within a given NR carrier to allocate services addressing different coverage
requirements. Carrier Bandwidth Parts (FDM) or frames with different numerologies (TDM) could be used for the provision of Terrestrial Broadcast services. In
the figure each service is transmitted in a different resource region according to the desired numerology. Note that for single-cell or MFN transmissions there is
only one numerology (15 kHz) employed. It is assumed that each service can be assigned a distinct DCI (G-RNTI, resource mapping and MCS index) therefore
treating each TV/radio service in a similar way as user data in regular unicast frames. Note that over time services can be re-scheduled, activated and deactivated,
according to broadcaster’s demand.
Tsubframe
... ...
SSBlock Coreset SI
...
...
#OFDM B
#OFDM A
Total Carrier BW
Regional SFN CBP
National SFN CBP
SSBlock Coreset SI
... ...
SSBlock Coreset SI
...
...
#OFDM C
SSBlock Coreset SI
SSBlock Coreset SI
Local Services CBP
...
...
PDCCH
DCI per TV/radio service
Group RNTI (G-RNTI)
T/F Resources in PDSCH
MCS Index
PDSCH
Service Information
Service Area Information
Service Continuity
+
time
frequency
Submitted to IEEE Transactions on Broadcasting
7
BW can be reserved to schedule services for nationwide SFN
coverage (OFDM A) alongside services intended for regional
(OFDM B) or local (OFDM C) areas. A receiver would only
demodulate the specific resources containing a desired service.
Following this approach, a single 5G-NR carrier can transmit
services intended for different reception conditions and
coverage targets without the need of transmitting isolated
multiplexes as currently done in Terrestrial Broadcast systems.
The use of CBPs is considered as an option for the FDM
multiplexing of single-cell or MFN services with those
intended for SFN areas within the same carrier. Note that
different numerologies can also be multiplexed in TDM.
3) Control Channel, Synchronization, Acquisition and
Scheduling mechanism
Signalling and synchronization mechanisms may be
simplified with respect to FeMBMS, i.e. by reducing the
amount of signalling to only that necessary for the correct
reception and discovery of services. As system preambles,
control channels are assumed to convey SS/PBCH Block, MIB
signalling via the PDCCH and a series of SIBs via the PDSCH.
Additional service announcement information may be allocated
over the PDSCH so that physical channels and procedures do
not require major modifications.
The characteristics of the SS/PBCH block and CORESET
in NR not only provide higher flexibility but may also reduce
the associated overhead.
On the one hand, a SS/PBCH block consists of 4 OFDM
symbols in the time domain and 240 carriers in the frequency
domain. A set of SS/PBCH block (namely SS burst) is
transmitted according to the numerology, the frequency band
and a periodicity of 5, 10, 20, 40, 80, or 160 ms [18]. Hence,
the largest capacity overhead for SCS = 15 kHz, and
frequency band below 6 GHz is 2 SS/PBCH Blocks every
frame (assuming a periodicity of 5 ms), making a total of
2x4x240=1920 REs per frame. These could be multiplexed in
TDM/FDM with the necessary periodicity according to the
nature of the services and the reception conditions (e.g. a high
repetition rate is useful to minimize acquisition time when
reception faces challenging conditions).
On the other hand, PDCCH flexibility of NR is enhanced
with the introduction of CORESETs, which can be considered
as smaller control units of former LTE Control Region. While
LTE PDCCH is extended through the whole carrier bandwidth
of OFDM symbols 1-3 of every subframe, NR PDDCH uses a
reduced part of the carrier bandwidth used. One CORESET
occupies from 1 to 16 Control Channel Elements (CCEs),
comprising 6 Resource Element Groups (REGs), which in turn
are composed by 12 REs. Hence, CORESET sizes go from
1x6x12 = 72 REs to 16x6x12 = 1152 REs per frame. 4 NR-
MBMS services with independent CORESETs, in a 50 MHz
bandwidth, would lead to a 2.78% overhead
3
. Since longer
periodicity values, e.g. 40 ms, as well as larger carrier
bandwidths are contemplated in NR, the associated overhead
can be reduced.
3
50 MHz = 270 RB*12 RE/RB*14 OFDM symb/subframe*10
subframes/frame = 453600 REs in a frame.
4*(1152 REs/CORESET + 1920 REs/SS-BCH burst) *100 / 453600 =
2.78%
Therefore, the existing mechanism in NR would have
similar overhead to that in FeMBMS.
These control channels are proposed to be transmitted using
regular unicast numerology as they may contain information
which is intended to be transmitter-specific. By means of
network planning the control channel resources for each
transmitter can be scheduled so that co-channel interferences
are avoided.
For single-cell or MFN transmissions as well as for SFN,
coordinated frame scheduling is required in order to avoid
overlaps between adjacent time/frequency resources that may
create interference (in the first case) and to provide
synchronous SFN transmission (in the latter case).
Although the mismatch between numerologies of the control
channels and SFN data may still exist, it may be possible to
detect the signal at the expense of larger detection time by
means of a large aggregation level of e.g. PDCCH.
Regarding data scheduling, a reuse of the procedures
existing for unicast are desired in order to minimize
implementation impact and to exploit flexible resource
allocation. A mechanism like the one used by SC-PTM in LTE
could therefore be used. A G-RNTI could identify the
resources allocated to broadcast in a similar way as user-
specific content is identified with a C-RNTI. Furthermore,
treating each TV/radio service in a similar way to unicast
traffic, may make it possible to define a series of parameters
for each specific TV/radio service. The transmitted service
information would enable identification of time/frequency
resource allocation in the frame, the corresponding MCS (with
a corresponding DCI format), as well as service area
information or service continuity information (by means of
SIBs).
With dynamic scheduling, TV/radio services can be
transmitted according to operator demands, exploiting better
statistical multiplexing and spectral efficiency. Services could
also be switched on and off or created (e.g. introduction of
local services) over time, therefore only consuming the
resources of the 5G NR carrier when required.
Fig. 2 shows an illustrative example of a potential framing
configuration. Note that some of the contents addressing
different coverage areas (nationwide SFN, regional SFNs and
local areas) are multiplexed within the same carrier.
IV. PERFORMANCE EVALUATION
This section presents performance evaluation results of the
proposed NR-MBMS system in comparison to LTE
FeMBMS, focusing on the modes relevant for Terrestrial
Broadcast and considering both single-cell or MFN modes and
SFN modes. Special attention is given to the latter as these
would require a different performance than for unicast modes.
The analysis is focused on spectral efficiency, robustness
against frequency selectivity (e.g caused by natural multipath
in single-cell or MFN configurations and artificial multipath
due to echoes in an SFN) and time selectivity (due to Doppler
shift for moving users). The analysis focusses on the channels
conveying payload data (user plane). The link-level
performance of the physical layer signaling (control plane) is
also evaluated for different reception environments. Note that
the results are obtained assuming realistic channel estimation
Submitted to IEEE Transactions on Broadcasting
8
with Least Square (LS) equalization and interpolating linearly
in time domain and with an FFT-based method in frequency
domain [20]. The selected Quality of Service (QoS) is a block
error rate, BLER < 0.1%.
For reference with LTE FeMBMS, performance is
evaluated assuming that all resources in a frame are allocated
to a given service (i.e. the maximum transport block size is
selected according to the maximum number of resource blocks
in a 5 MHz bandwidth carrier). The frequency band
considered is UHF (700 MHz) as a traditional band of
Terrestrial Broadcast systems.
Link-level simulations are complemented by coverage
performance studies, which are mainly focused on new
numerologies for SFN.
A. Link-level Performance under Rician and Rayleigh fading
Link-level performance is influenced by the channel and
reception conditions that may hamper the correct reception of
the signal. An adequate selection of the modulation and
coding scheme (MCS) in both LTE FeMBMS and NR-MBMS
systems implies a compromise between maximum spectral
efficiency and SNR operation point.
This paper has considered three different types of reception
conditions for Terrestrial Broadcast services according to [2],
with their corresponding minimum SNR requirements. These
are the following:
▪ Fixed roof-top reception, modelled by a Rician fading
channel as per [19]. The minimum SNR requirement is
considered to be 20 dB.
▪ Portable reception, modelled by Rayleigh fading [19],
with a minimum SNR requirement of 10 dB.
▪ Mobile reception at a reference user speed of 60 km/h,
modeled by a Typical Urban 6 (TU-6) channel model
[19], with a minimum SNR requirement of 6 dB.
Fixed reception benefits from Line-of-Sight (LoS) conditions,
while portable reception introduces some loss due to the lack
of it. Regarding mobile reception, different circumstances
affect performance. Medium speed is generally the best case
whereas low-speed reception lacks time selectivity and at high
speed performance is degraded due to high Doppler shift.
Table III presents the SNR thresholds of the MCS indexes
that are close to 6, 10 and 20 dB in FeMBMS and NR-MBMS
for the three reception conditions. Note that, for comparison a
SCS of 1.25 kHz is considered for both systems.
As it can be observed, NR-MBMS provides a performance
gain in all cases, although the improvement is below 1 dB.
This gain comes from an advanced physical layer with LDPC
decoding.
Single-cell or MFN configuration, which can be
implemented using the regular 15 KHz SCS mode, may
experience similar gains. In this case, SC-PTM can be
compared to the proposed NR-MBMS MFN modes. In terms
of performance, there are differences in the order of 1 dB for
robust MCS modes whereas the difference decreases for high
MCS indexes. As an example, the required CNR threshold in
AWGN conditions for SC-PTM is -3.2 dB (MCS2), 5 dB
(MCS12) and 15.5 dB (MCS24) whereas for NR-MBMS with
a single-cell/MFN mode the values are -4.1 dB, 5.1 dB and
15.7 dB, respectively [22].
B. Link-Level Performance with echoes in SFN networks
NR-MBMS performance is evaluated in frequency selective
fading environments, where SFN multipath may represent a
degradation. In this case, the performance is represented as the
variation of the minimum required CNR depending on the
relative echo delay.
The SFN scenario is modelled by extending the 0 dB echo
channel in [19]. This channel is defined as a channel model
formed by two paths with same amplitude and a time delay
between them equivalent to 90% of the CP duration. The
second path is moved in order to set different delays inside
and outside the CP.
Fig. 3, focused on ISD from 30 km to 120 km (e.g. HPHT
networks), shows the performance for different echo delays
for the three MCS indexes defined in Table III. As it can be
observed, the required CNR remains constant when artificial
echoes arrive inside the CP region. When the echoes arrive out
of the CP region, a performance degradation begins until the
Nyquist limit (Tp), which limits the system operation.
Comparing the different configurations, it can be observed that
NR-MBMS outperforms LTE FeMBMS in terms of SFN
performance. Special focus is given to the performance of the
system for echoes arriving after CP and before Tp (i.e. within
the equalization interval). Robust enough MCS indexes permit
to extend system performance beyond CP. In such case, a
proper design of the reference signals for SFN networks is
critical as low CPs not stressing the framing design of the
system could still be used in SFN deployments. These results
are aligned with the conclusions presented in [21].
For SFN deployments with short ISD (e.g. LPLT networks),
Fig. 3. Required CNR performance for different MCS and echo delays in
Terrestrial Broadcast and LTE FeMBMS.
TABLE III. SNR (DB) ACHIEVED IN THE CONSIDERED SCENARIOS.
TECHNOLOGY
Mobile
(MCS2)
Portable
(MCS12)
Fixed
(MCS24)
SFN
Modes
FeMBMS
7.0
11.4
20.5
NR-MBMS
6.6
9.8
19.9
Fig. 3. Required CNR performance for different MCS and echo delays in
Terrestrial Broadcast and LTE FeMBMS.
Submitted to IEEE Transactions on Broadcasting
9
reference [22] provides link level performance results for
numerologies with extended CP and negative factor µ (µ=0,
µ=-1 and µ=-2). In general, robust MCS can provide better
resilience against SFN echoes outside the CP.
C. Resilience to Doppler shift at high speed and SFN
deployments
This section studies the tolerance to Doppler spread with
practical receiving algorithms in both FeMBMS and NR-
MBMS following the procedures in [22] and [23] as a
reference.
Doppler shift is a natural effect of the relative speed
between users and transmitters. OFDM systems are affected
by this phenomenon which is directly related to SCS.
Therefore, the effect is critical for mobile reception in SFN
deployments, in particular when large ISD need to be covered.
Single-cell or MFN modes may also suffer from this effect
when low overhead configurations are used (i.e for low CP to
TU ratios).
Channel estimation assisted by reference signals in OFDM
systems relies on measurements made on those subcarriers
which are reference-bearing signals. This can be performed
every -th symbol. Since symbols occur at the rate
fs=1/(TU+TCP), it follows that the Nyquist limit for temporal
channel variation (i.e. Doppler limit) that can be measured is
[19]:
(2)
Hence, the performance depends on the subcarrier spacing,
system bandwidth, the operational frequency band and the
accuracy of channel estimation method used.
A wide range of Doppler shifts is evaluated, which can later
be mapped to a given receiver speed according to the assumed
frequency. Note that for single-cell or MFN configurations
using a wide subcarrier spacing (e.g. 15 kHz) Doppler may not
impose a critical limit. Therefore, theoretical Doppler limits
for different SFN configurations are shown in Table IV.
Reference signals with Dt values 3 (15 kHz) and 2 (extending
1.25 kHz to 2.5 kHz and 625 kHz) are assumed.
These theoretical values are evaluated considering realistic
scenarios modelled by a TU-6 channel with variable speed.
The Doppler shift limit is calculated as the value that entails a
CNR performance loss of 3 dB compared to the lowest CNR
achieved in the whole range [24].
Fig. 4 shows how the use of a different numerology has a
great impact on the mobility tolerance. However, in general
the results practically do not change between LTE FeMBMS
and NR-MBMS if this parameter is the same. For instance,
both LTE and NR with SCS 15 kHz allow Doppler shifts up to
1550 Hz, equivalent to 2390 km/h at 700 MHz. The common
SCS that can be selected in both LTE FeMBMS and NR-
MBMS, i.e. 1.25 kHz, reduces the maximum Doppler shift to
180 Hz approximately, equivalent to 280 km/h when MCS 2 is
selected. A higher MCS may provide lower correction
capabilities, thus lowering this value. It should also be noticed
that compared to FeMBMS, the user speeds permitted with
SCS 625 Hz (CP 400 µs) are considerably lower. This mode
has been designed to support high-demanding coverage
requirements in fixed reception scenarios, which implies the
use of long CPs that, in turn and in order to minimize
overheads, require narrow SCS. This mode permits 75 Hz
Doppler shift, which represents a user speed of 115 km/h with
MCS 2.
The use of a more robust MCS (e.g. MCS index 0) reduces
the SNR at low user speeds. However, the maximum Doppler
shift permitted is still 75 Hz. Overall, improved performance
is achieved at the expense of capacity when using robust
transmission modes.
One important drawback of LTE FeMBMS and NR-MBMS
is the lack of time interleaving which will improve
performance by increasing time diveristy. With time
TABLE IV. THEORETICAL DOPPLER LIMIT AND MAXIMUM USER SPEED AT
700 MHZ.
SCS
625 Hz
1.25 kHz
2.5 kHz
15 kHz
(µs)
400
200
100
16,67
(µs)
1600
800
400
66,67
2
2
2
3
Doppler limit (Hz)
125
250
500
2000
Max. user speed @
700 MHz (km/h)
192
384
771
3085
Fig. 4. Doppler shift (Hz) vs. SNR (dB) when using the different SCS
options in both LTE FeMBMS and NR-MBMS. Complete range of Doppler
Shift from 0 to 1800 Hz (top) and zoom from 0 to 500 Hz (bottom).
Submitted to IEEE Transactions on Broadcasting
10
interleaving, the system could withstand user speeds higher
than 250 km/h, as shown in [25]. The SCS of 2.5 kHz
represents a compromise between both options. In this case,
SFN coverage may suffer with large ISD, but mobility is still
relatively high, i.e. 390 Hz equivalent to 600 km/h at 700
MHz.
D. Link-level performance of Control Channels
This section analyses the performance of the coding
schemes employed in LTE and 5G-NR PDCCH, as this is
detected as one of the most limiting channels from those
constituting the FeMBMS CAS [11]. 3GPP has adopted Polar
coding for control channel in 5G-NR, which reduce decoding
complexity while almost closing the gap to Shannon. LTE
employs TBCC (Tail Biting Convolutional Code) instead.
In order to increase the decoding probability of the PDCCH
(i.e. decoding of DCI formats), LTE and NR define different
Aggregation Levels (AL). According to the selected DCI
format, it is possible to define an AL so that the scheduler will
define an appropriate robustness increase. There is a trade-off
between robustness and data rate. The higher the AL, the
higher the number of subcarriers used to transmit a DCI
which, at the same time, limits the number of DCIs that can be
allocated into a given subframe. While LTE permits to use an
AL up to 8, 5G-NR has increased this value up to 16
repetitions. This was adopted for allowing the correct
demodulation even at very high noisy conditions.
Fig. 5 presents the performance of both PDCCH coding
schemes for a wide range of DCI lengths and AL over fixed,
portable and mobile reception conditions. This validates that
5G-NR Polar codes outperform LTE TBCC. In particular, the
performance gains for DCI = 12 bits and AL = 8 are between
1.6 dB and 2 dB for the different reception scenarios under
evaluation. It can be observed that doubling the AL reduces
approximately in 3 dB the SNR requirements. The shortest
DCI length (i.e. DCI = 12) is the most robust one, thanks to
having the lower effective coding rate. The possibility to
increase AL up to 16 brings an advantage which may
eliminate the potential problems for detecting the CAS in
LTE.
E. Coverage evaluation for SFN networks
An overview on the suitability of enhancing numerologies
for SFN networks is provided in this section. Wide area SFNs
have been modelled using an hexagonal network layout with
five rings of sites around a central transmitter. The available
SINR, incorporating the effects of SFN self-interference, has
been computed at an apex of the central hexagon.
In the LPLT networks the effective radiated power (ERP)
was set to 40 W at an effective antenna height of 30 m while 50
kW and 250 m were used for the HPHT network.
Table V sets out the receiving environment parameters used
in the simulations; all values are in-line with [26]. ITU-R
P.1546-5 has been used to calculate the mean signal strengths
of the wanted and interfering signals in 100m x 100m ‘pixels’
comprising the coverage area. Within a pixel these signals vary
from one location to another according to a log-normal
distribution with standard deviation of 5.5 dB and has thus
been modelled as random variables. The Schwartz and Yeh
method has been used to calculate the combined wanted and
interfering signal powers so that the probability of reception at
any point within the pixel can be determined.
A generic analysis of the coverage for fixed roof-top
reception has been conducted as a function of ISD for various
different CP lengths (33, 100, 200, 300 and 400 µs) where the
Fig. 5. 5G-NR polar codes and LTE TBCC codes performance for different DCI lengths (12, 48, 96, 132) and Aggregation Levels (1, 2, 4, 8, 16 for Polar and
1, 2, 4, 8 for TBCC) in Fixed (left), Portable (center) and Mobile (right) reception conditions.
TABLE V. COVERAGE SIMULATION PARAMETERS
Parameter
Roof-Top Reception
Receiving Antenna Height
10 m
Receiver Noise Figure
6 dB
Rx Antenna Pattern
ITU-R BT.419
Rx Antenna Gain
13.15 dBi
Antenna Cable Loss
4 dB
Implementation Margin
1 dB
Noise Bandwidth
4.5 MHz
Frequency
700 MHz
Propagation Model
ITU-R P.1546-5 over land
Wanted Signal Time Value
50% time
Interfering Signal Time Value
1% time
Location Variation
5.5dB (log-normal distribution)
Signal Summation
Schwartz & Yeh power sum
Pixel size
100m x 100m
Submitted to IEEE Transactions on Broadcasting
11
Fig. 6. Available SINR at the worst pixel of the LPLT and HPHT
networks as a function of the ISD and different CP duration (SFN).
two latter CPs have been hypothecated in order to determine
whether there would be any benefit in further extending the
CP. For these two modes the OFDM symbol period has been
extended accordingly so that the CP always represents ¼ of the
symbol duration – in line with the standardized eMBMS modes
and those initially proposed for NR-MBMS. The achievable
SINR, at the apex of the central hexagon in the network was
then computed for reception qualities of 70% and 95%
locations, as two common metrics for coverage estimation.
Coverage quality is then expressed as the percentage of
locations exceeding a given SINR threshold within a pixel for
99% of the time.
Fig. 6 presents the results for LPLT (top) and HPHT
(bottom) networks. It is found that for all the LPLT ISDs
studied, the 200 µs CP would be sufficiently long. Extending it
further would provide no additional benefit against SFN self-
interference – the achievable SINR would not increase.
Conversely it can be seen that the 200 µs CP significantly
improves the SINR for all the LPLT ISDs studied compared
with the 33 µs option while a 100 µs variant may be a good
addition for networks with ISDs of 5 to 10km.
For HPHT networks, it can be seen that the 200 µs CP for
ISDs greater than 70 km – i.e. ISDs typical of existing
Terrestrial Broadcast networks – is too short. The introduction
of longer CPs would improve the coverage of the system. Here
a 300 µs and 400 µs CPs have been computed, being the latter
the one providing the highest improvement.
According to the results, wide area coverage in existing
Terrestrial Broadcast networks – where ISDs of 60 km or
more are common - may be limited to modes with SINR
thresholds below 12-13 dB for 95% coverage availability, or
below 19 dB for 70% coverage availability.
The coverage of eMBMS in a national SFN is now
assessed in the UK Terrestrial Broadcast network in order see
what may happen in a more practical setting.
In this example the UK Prediction Model (UKPM) was
used – a prediction model jointly developed by ITC, BBC,
Crown Castle and NTL for planning Terrestrial Broadcast
services in the UK [27]. All 1100+ UK Terrestrial Broadcast
transmitters were modelled with the eMBMS parameters
shown in Table V. All other physical characteristics of the
network, such as antenna patterns, ERPs, transmitter locations
and antenna heights were otherwise unchanged.
It is clear from Table VI that the 200 µs CP would be too
short to achieve near-universal coverage with a national SFN.
Although this result is somewhat different to the hexagonal
network simulations, it may be explained by observing that
practical networks are much less regular. For example, they
contain real terrain and ISDs of various lengths, some greater
than 60 km. Sea paths over convex sections of coast also lead
to higher interference than is found in the land based regular
hexagon networks. A longer CP, in the order of 400 µs, would
therefore be beneficial.
Simulation for additional modes with 200 and 400 µs CP
but with significantly larger OFDM symbol durations have also
been carried out. A clear benefit is achieved in both cases
where 11.1% and 1.2% more coverage is reached in
comparison with the respective modes with 20% overhead.
Overall, the network may benefit from an SINR increase
together with larger capacity thanks to lower overheads.
V. CONCLUSIONS
The design of the air interface of an MBMS system based
on 5G New Radio (NR), NR-MBMS, has been outlined. The
design extends the recent 5G-NR Release 15 and Release 16,
initially only focused on unicast transmissions, to Terrestrial
Broadcast services. The design does not necessarily require a
split between a mixed mode carrier containing unicast/
multicast/broadcast or a dedicated carrier as the latter is
simply derived from the allocation of 100% of resources to
Terrestrial Broadcast services.
For the single-cell or MFN configurations, the physical
layer design that has been outline has a minimal impact with
respect to unicast. Existing synchronization and acquisition
mechanisms are could be reused with only minor changes.
Linear TV/radio services can be allocated by means of a group
identifier (G-RNTI) in a similar fashion as unicast data is
TABLE VI. PERCENTAGE OF UK HOUSEHOLDS AT PERCENTAGE LOCATIONS
(Hz)
TU (µs)
TCP
(µs)
TS
(ms)
ISD
(km)
Coverage
%
1250
800.00
200.00
1.0
60
86.5
625
1600.00
400.00
2.0
120
98.1
208.33
4800
200
5.0
60
97.6
217.39
4600
400
5.0
120
99.3
Submitted to IEEE Transactions on Broadcasting
12
scheduled. LPLT (small cells) as well as HPHT (large cells)
stations can be employed. The NR carrier may be used to
allocate up to 100% broadcast data multiplexed in both time
and frequency domains with high granularity and without
major constraints (by reusing the existing procedures for
unicast).
SFN may be enabled by extending the single-cell mode
which may require a more complex design in terms of
numerologies and a corresponding trade-off between mobility
and SFN coverage. Note also that MFN numerologies may
also be optimized to reduce capacity overheads. It is also
important to note that although it is desirable from a
deployment perspective to have as much flexibility as possible,
consideration should also be given on the potential receiver
complexity and testing that may impose limitations on the
maximum number of numerology options finally included in
the specifications.
Based on 5G NR, the system outlined herein may
outperform the existing FeMBMS system (based on LTE).
The design takes into account different reception scenarios
targeting high speed (at expense of capacity overhead) and
static reception (maximizing SFN efficiency and capacity).
The use of the new physical layer features of 5G-NR such as
new LDPC and Polar codes, increased bandwidth efficiency or
efficient numerology multiplexing permits the configuration
of new transmission mechanisms that outperform FeMBMS.
5G-NR may have up to 7.2% higher bandwidth utilization
compared with FeMBMS. With the use of bandwidth parts, a
single wideband carrier can multiplex services intended for
different reception conditions and also different coverage
areas, including local, regional SFN and nation-wide SFN.
Data channels can benefit from approximately 0.5 dB gain in
CNR threshold whereas the gain in term of control channels is
more noticeable thanks to the possibility of increasing
aggregation levels. The existing control channels for unicast
may already enable reduced signaling overhead with respect to
the CAS in FeMBMS and may not require any modification as
they are more flexible in terms of resource allocation and
periodicity. In terms of overheads, a skillful design may be
possible to maximize capacity by an adequate CP and useful
OFDM symbol duration.
Common techniques used in other standards, such as
physical layer time interleaving for improved robustness in
mobile environments, would also be of benefit, should they be
adopted by 5G-NR Terrestrial Broadcast.
REFERENCES
[1] 3GPP TR 38.913 v14.3.0, “Study on scenarios and requirements for next
generation access technologies,” June 2017.
[2] D. Vargas and D. Mi, Eds., “LTE-Advanced Pro Broadcast Radio
Access Network Benchmark,” Deliverable D3.1, 5G-PPP 5G-Xcast
project, Nov. 2017.
[3] I. Eizmendi et al. "DVB-T2: The second generation of terrestrial digital
video broadcasting system" IEEE Trans. Broadcast., vol. 60 no. 2 pp.
258-271, June 2014.
[4] L. Fay, L. Michael, D. Gomez-Barquero, N. Ammar, and M. W.
Caldwell "An overview of the ATSC 3.0 physical layer specification",
IEEE Trans. Broadcast., vol. 62 no. 1 pp. 159-171, March 2016.
[5] ETSI EN 300 401 V1.4.1" "Radio Broadcasting Systems; Digital Audio
Broadcasting (DAB) to mobile portable and fixed receivers", June 2006.
[6] HbbTV 2.0.2 Specification, HbbTV Association, December 2018.
[7] 3GPP TR 36.776 V1.1.0, “Study on LTE-based 5G Terrestrial Broadcast
(Release 16)”, March 2019.
[8] P. Klenner, J. Baek, N. S. Loghin, D. Gomez-Barquero, and W. Ko,
"Physical Layer Time Interleaving for the ATSC 3.0 System," in IEEE
Trans. Broadcast., vol. 62, no. 1, pp. 253-262, March 2016.
[9] L. Zhang et al., "Layered-Division-Multiplexing: Theory and Practice,"
in IEEE Trans. on Broadcast., vol. 62, no. 1, pp. 216-232, March 2016.
[10] 3GPP TR 36.859 V13.0.0, “Study on Downlink Multiuser Superposition
Transmission (MUST) for LTE”, January 2016.
[11] J. J. Gimenez, P. Renka, S. Elliott, D. Vargas and D. Gomez-Barquero,
"Enhanced TV Delivery with eMBMS: Coverage Evaluation for Roof-
Top Reception," 2018 IEEE International Symposium on Broadband
Multimedia Systems and Broadcasting (BMSB), Valencia, 2018, pp. 1-5.
[12] L. Richter and S. Ilsen, "Coverage Evaluation of LTE FeMBMS: a Case
Study Based on a DVB-T2 Network," 2018 IEEE International
Symposium on Broadband Multimedia Systems and Broadcasting
(BMSB), Valencia, 2018, pp. 1-5.
[13] 3GPP R1-1712879: “Forward Compatibility Consideration for ECP
Design and NR-MBMS”, August 2017.
[14] EBU Tech. Rep. 3348, “Frequency and Network Planning Aspects of
DVB-T2,” Oct. 2014
[15] E. Arikan, "Channel Polarization: A Method for Constructing Capacity-
Achieving Codes for Symmetric Binary-Input Memoryless Channels,"
in IEEE Transactions on Information Theory, vol. 55, no. 7, pp. 3051-
3073, July 2009.
[16] G. K. Walker, X. Zhang and X. F. Wang, "Extended duration cyclic
prefix with low overhead for LTE Broadcast” U.S. Patent
2015/0078292 A1, issued March 19, 2015.
[17] J. J. Gimenez, D. Gomez-Barquero, J. Morgade and E. Stare, "Wideband
Broadcasting: A Power-Efficient Approach to 5G Broadcasting," in
IEEE Communications Magazine, vol. 56, no. 3, pp. 119-125, March
2018.
[18] 3GPP TR 38.213 V15.4.0, “NR; Physical Layer Procedures for control”,
January 2019.
[19] ETSI TS 102 831 v1.2.1, “Digital Video Broadcasting (DVB);
Implementation guidelines for a second generation digital terrestrial
television broadcasting system (DVB-T2)”, August 2012
[20] L. Zhang, et al., “Enhanced DFT based channel estimation for LDM
systems over SFN channels,” in Proceedings IEEE International
Symposium on Broadband Multimedia Systems and Broadcasting
(BMSB), Ghent, Belgium, June 2015.
[21] J. L. Carcel, J. J. Gimenez and D. Gomez-Barquero, "Zero-guard OFDM
performance in SFN with ATSC 3.0 ultra-robust transmission modes,"
2017 IEEE International Symposium on Broadband Multimedia Systems
and Broadcasting (BMSB), Cagliari, 2017, pp. 1-5.
[22] E. Garro, M. Fuentes, J. J. Gimenez, J. L. Carcel, Eds., “Air Interface”,
5G-Xcast Deliverable D3.2, November 2018.
[23] M. Fuentes, et. al., “Physical Layer Performance Evaluation of LTE-
Advanced Pro Broadcast and ATSC 3.0 Systems,” to appear at IEEE
Trans. Broadcasting.
[24] S. I. Park et al., “Low Complexity Layered Division Multiplexing for
ATSC 3.0,” IEEE Trans. on Broadcast., vol. 62, no. 1, pp. 233-243,
March 2016.
[25] Zhang, et. al., “Using Layered-Division-Multiplexing to Deliver Multi-
Layer Mobile Services in ATSC 3.0,” IEEE Trans. Broadcasting, vol.
65, no. 1, March 2019.
[26] EBU, “Simulation Parameters for Theoretical LTE eMBMS Network
Studies”, Tech. Rep 034.
[27] P. G. Brown, K. T. Tsioumparakis, M. Jordan, A. Chong, “UK Planning
Model for Digital Terrestrial Television Coverage,” BBC Research &
Development, White Paper WHP 048, Sept. 2002.
[28] D. Gomez-Barquero, D. Navratil, S. Appleby and M. Stagg, “Point-to-
Multipoint Communication Enablers for the Fifth-Generation of
Wireless Systems”, IEEE Communications Standards Magazine, vol. 2,
no. 1, March 2018.
[29] A. Awada, M. Säily and L. Kuru, "Design and performance impact of
long cyclic prefixes for eMBMS in LTE networks," 2016 IEEE Wireless
Communications and Networking Conference, Doha, 2016, pp. 1-7.
