5GPP Attachment - 0
5GPP Attachment - 0
5GPP Attachment - 0
Ares(2018)6153612 - 30/11/2018
Deliverable D3.2
Air Interface
Revision History
Revision Date Issued by Description
2018/11/30 Final Deliverable
Abstract
This report investigates the 3GPP Release’15 (Rel’15) of 5G New Radio (NR), and extends
the air interface to point-to-multipoint (PTM) communications. Two modes have been
proposed in order to fulfil the different 3GPP requirements needed for broadcast and
multicast. The 5G-Xcast Mixed Mode enables a dynamic and seamless switching between
Point-to-Point (PTP) and PTM transmissions both in the downlink and the uplink. It reuses
the NR Rel’15 air interface specification. The 5G-Xcast Terrestrial Broadcast Mode enables
the reception of the service to users without uplink capabilities, i.e. being a downlink-only
mode. One of its design principles is the transmission over large coverage areas in order to
enable High-Power High-Tower (HPHT) network configurations. This deliverable also
provides an evaluation of the 5G NR Rel’15 unicast specifications. In addition, the two 5G-
Xcast PTM solutions are evaluated in order to demonstrate the more efficient use of the radio
resources and their advantages over PTP for the scenarios considered in 5G-Xcast
Keywords
5G New Radio, Point-to-Multipoint, Point-to-Point, Multicast, Broadcast Radio Access
Network, Key Performance Indicators, Link Level Simulations, System Level Simulations,
Coverage Simulations, Media & Entertainment vertical, Public Warning vertical, Automotive
vertical, IoT vertical, IMT-2020 evaluation.
5G-Xcast_D3.2
Executive Summary
This report investigates the 3GPP Release’15 (Rel’15) of 5G New Radio (NR), and
extends the air interface to point-to-multipoint (PTM) communications. Two specific 5G
PTM technologies are proposed in order to fulfil the different 3GPP requirements needed
for broadcast and multicast, as well as those derived from the 5G-Xcast use cases
defined in Deliverable D2.1. These technologies are the Mixed Mode and the Terrestrial
Broadcast Mode. 5G-Xcast partners have been contributing to the 3GPP Rel’16
discussions in order to detect missing functionalities that are needed to fulfill the agreed
PTM requirements.
The 5G-Xcast Mixed Mode enables a dynamic and seamless switching between Point-
to-Point (PTP) and PTM transmissions both in the downlink and the uplink. This solution
is envisaged for the different verticals of the 5G-Xcast project, i.e. media and
entertainment, automotive, internet of things (IoT) and public warning. It reuses the NR
Rel’15 air interface specification as much as possible to ensure the maximum
compatibility with PTP. However, some modifications are included such as the discovery
of the scheduling information to a group of users is enabled by the introduction of a Group
Radio Network Identifier (G-RNTI), and a multiple cell coordination that is enabled by
forcing the same cell scrambling sequence to the neighbouring Next Generation NodeBs
(gNB). Negative numerologies as well as the concept of mini-slots are also included to
support SFN areas and larger deployments.
The 5G-Xcast Terrestrial Broadcast Mode enables the reception of the service to users
without uplink capabilities, i.e. being a downlink-only mode. One of its design principles
is the transmission over large coverage areas in order to enable the media delivery
through deployments including High-Power High-Tower (HPHT) network configurations.
To make this possible, a new physical channel has been defined. It includes new cyclic-
prefix values and reference signals, as well as a Cell Acquisition Subframe (CAS) and a
very narrow carrier spacing (even more than those provided in the Mixed Mode) in the
frame structure to allow very large Inter-Site Distances (ISDs) in Single Frequency
Networks (SFNs), as required by the 3GPP requirements.
This deliverable also provides an evaluation of the 5G NR Rel’15 unicast specifications
against specific Key Performance Indicators (KPIs), as defined in the IMT-2020
guidelines. It includes analysis, inspection and link-level simulations. In addition, the two
5G-Xcast PTM solutions are evaluated in order to demonstrate the more efficient use of
the radio resources and their advantages over PTP for the scenarios considered in 5G-
Xcast.
1
5G-Xcast_D3.2
Table of Contents
Executive Summary ........................................................................................................ 1
Table of Contents ........................................................................................................... 2
List of Figures ................................................................................................................. 5
List of Tables .................................................................................................................. 7
List of Acronyms and Abbreviations ............................................................................... 8
1 Introduction ............................................................................................................ 10
5G Development Status ................................................................................. 10
Objectives of the document ............................................................................ 10
Structure of the document .............................................................................. 11
2 5G Point-to-Point Air Interface Overview............................................................... 12
Frame Structure ............................................................................................. 12
Physical Channels and Signals ...................................................................... 13
2.2.1 Physical Downlink Channels and Signals ............................................... 13
2.2.2 Physical Uplink Channels and Signals .................................................... 16
Acquisition Procedure .................................................................................... 16
Feedback........................................................................................................ 17
3 5G Point-to-Multipoint Air Interface Design ........................................................... 19
Introduction..................................................................................................... 19
3.1.1 Summary of RAN Technical Requirements ............................................ 19
3.1.2 Summary of Potential Limitations of LTE Broadcast RAN ...................... 19
Single Cell Mixed Mode (SC-MM) .................................................................. 20
3.2.1 Modifications Introduced in the Downlink (PDCCH) ................................ 20
3.2.2 Modifications Introduced in the Uplink (PUCCH) .................................... 21
Multiple Cell Mixed Mode (MC-MM) ............................................................... 22
3.3.1 Common Cell Scrambling Sequence ...................................................... 22
3.3.2 Negative Numerologies and Extended CP.............................................. 23
3.3.3 Channel Estimation ................................................................................. 25
3.3.4 Feedback ................................................................................................ 26
Terrestrial Broadcast Mode ............................................................................ 26
3.4.1 PTBCH with Compatible SCS and Extended CP .................................... 26
3.4.2 Channel Estimation ................................................................................. 27
MIMO Techniques .......................................................................................... 28
4 Performance Evaluation of 5G PTP against IMT-2020 Requirements .................. 31
Enhanced Mobile Broadband for Media & Entertainment .............................. 31
4.1.1 Bandwidth ............................................................................................... 31
4.1.2 Peak data rate ......................................................................................... 32
4.1.3 Peak spectral efficiency .......................................................................... 35
2
5G-Xcast_D3.2
3
5G-Xcast_D3.2
4
5G-Xcast_D3.2
List of Figures
Figure 1. NR Framing structure (µ = 0)......................................................................... 12
Figure 2. Framing structure for SFI=0 (10 subframes, 100% downlink). ...................... 14
Figure 3. SS/PBCH block allocation within a slot ......................................................... 15
Figure 4. CORESET allocation with AL = 1 (left) and PDSCH resources in a slot (right).
...................................................................................................................................... 16
Figure 5. NR Rel’15 acquisition procedure. .................................................................. 17
Figure 6. C-RNTI for PTP versus G-RNTI for PTM communications. .......................... 21
Figure 7. Mini-slots considered for numerologies -1 and -2.......................................... 24
Figure 8. Mini-slot for numerology -3 and incompatibility with 5G Rel’15. .................... 24
Figure 9. MC-MM Control region and DMRS patterns for negative numerologies. ...... 25
Figure 10. PTBCH DMRS Patterns .............................................................................. 27
Figure 11. 5G NR Control Acquisition Subframe .......................................................... 28
Figure 12. Block error rate vs CNR (dB) for SISO AWGN channel with NR. ................ 37
Figure 13. BICM spectral efficiency vs CNR (dB) for SISO AWGN channel. NR vs. LTE
Rel’14 eMBMS and ATSC 3.0. ..................................................................................... 37
Figure 14. BICM spectral efficiency vs CNR (dB) with NR, AWGN MIMO channel. ..... 38
Figure 15. BICM spectral efficiency vs CNR (dB) for all considered scenarios. ........... 39
Figure 16. BICM spectral efficiency vs CNR (dB) for TDL scenarios with 2x2 MIMO. . 39
Figure 17. Block error rate vs CNR for AWGN channel. Different ALs with NR and LTE.
...................................................................................................................................... 40
Figure 18. CNR against user speed for 5G New Radio in TU-6 mobile channel.
Numerologies 0,1 and 2, MCS 3................................................................................... 43
Figure 19. CNR against user speed for 5G New Radio in TU-6 mobile channel.
Numerology 0, MCS 15. ............................................................................................... 44
Figure 20. CNR against user speed for 5G New Radio in TDL-A mobile channel. PDCCH,
real channel estimation. ................................................................................................ 45
Figure 21. User plane latency calculation. .................................................................... 46
Figure 22. User plane latency for numerologies 0 and 1, with slot based scheduling of 14
symbols. Minimum frame alignment. ............................................................................ 47
Figure 21. Peak data rate (top) and gain obtained (bottom) with SC-MM and MC-MM
compared to PTP, for different numbers of users and different number of CCs. .......... 50
Figure 22. BICM spectral efficiency vs. required CNR, MC-MM with 0 dB echo channel.
...................................................................................................................................... 52
Figure 23. BICM spectral efficiency vs. required CNR, TB with 0 dB echo channel. .... 52
Figure 24. Required CNR with different echo delays using MC-MM. ........................... 53
Figure 25. Required CNR with different echo delays using the 5G TB mode. .............. 54
Figure 26. Required CNR vs corresponding echo delay (PDCCH, AL = 2) .................. 55
Figure 27. Normal & extended CP with 0.1 ∗ 𝑇𝑇𝑇𝑇𝑇𝑇 as delay for the second path. .......... 56
Figure 28. Normal & extended CP with 0.6 ∗ 𝑇𝑇𝑇𝑇𝑇𝑇 as delay for the second path. .......... 57
Figure 29 SINR vs Coverage Probability at the location of minimum capacity for the
network configurations shown. Fixed rooftop reception. HPHT1 (left) and HPHT2 (right)
...................................................................................................................................... 58
Figure 30: SINR vs Coverage Probability at the Location of Minimum Capacity for car
mounted reception. MPMT (left), LPLT (right). ............................................................. 59
Figure 31. CNR against user speed for 5G-Xcast PTM modes in TU-6 mobile channel.
...................................................................................................................................... 60
Figure 21. User plane latency for numerology -2, with non-slot based scheduling of 2
symbols. Minimum frame alignment. ............................................................................ 61
Figure 32. PDCCH block diagram. ............................................................................... 68
Figure 33. RE mapping procedure in PDCCH for aggregation level 1 ......................... 70
Figure 34. NR Link-level transmitter block diagram. ..................................................... 70
5
5G-Xcast_D3.2
6
5G-Xcast_D3.2
List of Tables
Table 1. 5G air interface parameters: numerology (μ), subcarrier spacing, useful symbol
duration (TU), CP duration (TCP) and slots per subframe. ............................................. 13
Table 2. Number of Resource Elements (REs) allocated for PDCCH CORESET with
different number of UEs for PTP and SC-MM. ............................................................. 21
Table 3. Maximum ISD with 5G NR Rel’15 unicast numerologies. ............................... 23
Table 4. Extended cyclic prefixes for 5G negative numerologies. ................................ 24
Table 5. MC-MM Reference Signal Patterns ................................................................ 26
Table 6. Terrestrial Broadcast new numerologies and extended CP combinations. .... 27
Table 7. Terrestrial Broadcast DMRS patterns ............................................................. 28
Table 8. Assessment method used per KPI and associated 5G-Xcast verticals. ......... 31
Table 9. NR maximum supported bandwidth................................................................ 32
Table 10. FDD DL peak data rate values ..................................................................... 33
Table 11. TDD DL peak data rate values ..................................................................... 34
Table 12. FDD UL peak data rate values ..................................................................... 34
Table 13. TDD UL peak data rate values ..................................................................... 34
Table 14. Peak spectral efficiency for FDD DL. ............................................................ 35
Table 15. Peak spectral efficiency for TDD DL. ............................................................ 35
Table 16. Peak spectral efficiency for FDD UL. Numerologies from 0 to 2, with and
without MIMO................................................................................................................ 36
Table 17. Peak spectral efficiency for TDD UL. Numerologies from 0 to 3, with and
without MIMO................................................................................................................ 36
Table 18. Minimum CNR threshold of PDCCH in AWGN, TDL-A, and TLD-C channel.
...................................................................................................................................... 41
Table 19. Doppler limit (Hz) and related speed (km/h) for 700 MHz and 4 GHz bands.
...................................................................................................................................... 42
Table 20. User plane latency (ms) for all considered configurations. ........................... 48
Table 21. Assessment method used per KPI and associated technology. ................... 49
Table 22. Spectral efficiency vs number of users. ........................................................ 51
Table 23 Network Parameters ...................................................................................... 57
Table 24 Channel characteristics for considered environments ................................... 58
Table 25 Achievable SINR (dB) at 95% locations. Fixed rooftop reception. ................. 58
Table 26 Carrier spacings and CP overhead for numerologies used in simulations .... 59
Table 27 Achievable SINR (dB) at 95% locations. Car mounted reception .................. 59
Table 28. User plane latency (ms) with the Mixed Mode. ............................................. 62
Table 29. User plane latency (ms) with 5G Terrestrial Broadcast. ............................... 62
Table 30. KPI analysis and the associated IMT-2020 requirements. ........................... 65
Table 31. DCI Format 1_0 content. .............................................................................. 68
Table 32. FDD FR1 cases. ........................................................................................... 78
7
5G-Xcast_D3.2
8
5G-Xcast_D3.2
9
5G-Xcast_D3.2
1 Introduction
5G Development Status
The first 3GPP release of 5G, i.e. Release’15 (Rel’15) has been structured in three
phases. An early drop non-standalone (NSA) version was initially approved in December
2017. This version relies on both Long Term Evolution (LTE) and New Radio (NR) air
interfaces and reuses the LTE core network. It also focuses on the user plane and makes
use of LTE for the control plane. In June 2018, a 5G stand-alone (SA) version was
specified. It additionally includes a fully 5G core network and has full user and control
plane capabilities. The last drop of Rel’15 is expected to be released at the end of 2018
and it will enable the 5G core to inter-work equally with both LTE RAN and NR RAN.
The NR air interface brings a large number of improvements compared to LTE to address
the new IMT-2020 requirements of enhanced mobile broadband (eMBB) and ultra-
reliable low-latency communication (URLLC) services 1, as well as to cover an extensive
number of use cases for the digitalization of new industries (also known as verticals).
The main improvements in the air interface are more efficient Forward Error Correction
(FEC) codes, large bandwidths, new OFDM waveform numerologies that adapt to the
new spectrum bands and bandwidth allocations, dynamic frame structures, or massive
MIMO (Multiple-Input Multiple-Output) schemes, among others [1]. However, NR Rel’15
only supports point-to-point (PTP), which may be inefficient when transmitting the same
content to a large number of users simultaneously [2]. In fact, the deliverable D2.1 of 5G-
Xcast [3] describes a series of use cases from 4 verticals where point-to-multipoint (PTM)
would improve the overall performance in 5G: media and entertainment (M&E), vehicle-
to-everything (V2X) communications, Internet-of-Things (IoT) and public warning. PTM
transmissions are also useful in other verticals such as airborne communications (e.g.
use of drones).
3GPP identified a flexible broadcast/multicast service as a basic feature to be used in
5G systems [4] and reference [5] sets out multicast/broadcast requirements in 3GPP for
next-generation access technologies. The support of broadcast and multicast
capabilities in 5G was discussed in Rel’16 and a new Study Item (SI) called “LTE-based
5G Terrestrial Broadcast” was approved. Although this SI is highly related to the project,
it is based on LTE enTV solutions. Therefore, it has been included as part of deliverable
D3.1. There are no current plans in 3GPP for the use of PTM in 5G NR where this study
has been postponed, at least, to Rel’17.
1 IMT-2020 requirements for massive Machine-Type Communications (mMTC) will be addressed with LTE
IoT (Internet-of-Things) solutions LTE-M (LTE for Machines) and NB-IoT (Narrow-Band IoT).
10
5G-Xcast_D3.2
11
5G-Xcast_D3.2
Frame Structure
Both downlink and uplink physical channels in NR are organized into frames with 10 ms
duration each. Frames are divided in turn in 10 subframes, with each subframe having a
fixed duration of 1 ms. The number of OFDM symbols in a subframe will depend on the
carrier spacing in the frequency domain, which is directly linked to the term numerology,
and the type of cyclic prefix (CP). Multiple numerology options are defined by µ, a
positive integer factor. OFDM symbols are also grouped in slots. With normal CP each
slot conveys 14 OFDM symbols. With extended CP (only available with numerology
µ=2), there are 12 OFDM symbols per slot. An illustrative example of framing structure
for numerology µ=0 is shown in Figure 1.
In the frequency domain, each OFDM symbol contains a fixed number of subcarriers.
This number also depends on the numerology selected, but also on the total bandwidth.
Note that each OFDM symbol can be assigned to downlink or uplink transmissions
depending on the Slot Format Indicator (SFI), which allows flexible assignment for Time
Division Duplex (TDD) or Frequency Division Duplex (FDD) operation modes. In any
case, a Resource Element (RE) is defined as 1 subcarrier in a OFDM symbol, and a
Resource Block (RB) is defined as a group of 12 REs in frequency. The REs are
separated with a specific carrier spacing (SCS), calculated as follows:
12
5G-Xcast_D3.2
Different bandwidths are available depending on the different frequency ranges [10].
While FR1 (450 MHz - 6 GHz) allows bandwidths from 5 MHz up to 100 MHz, FR2 (24.25
GHz - 52.6 GHz) offers values from 50 MHz up to 400 MHz. A more detailed explanation
is provided in Section 3.
2 For Normal CP, two CP lengths are (first OFDM symbol / rest of OFDM symbols).
13
5G-Xcast_D3.2
Physical channels and signals are allocated in a frame as shown in Figure 2. In this
particular case, the SFI value is 0, which means that the 14 OFDM symbols are allocated
to the downlink. From this figure, the physical channels and physical signals are allocated
as follows:
PSS / SSS / PBCH / PBCH-DMRS:
Grouped in SS/PBCH blocks, where each SS/PBCH block consists of 240 subcarriers
and 4 OFDM symbols [11]. PSS signals are allocated in the first OFDM symbol within a
SS/PBCH block, while SSS signals are distributed across the third OFDM symbol. The
PBCH and PBCH-DMRS signals are transmitted in the second, third and fourth OFDM
symbol. Cells set to 0s are used as padding to complete the block structure. There are
four available numerology options within the block, µ = {0,1,3,4}, which are selected
depending on the frequency range. The allocation of SS/PBCH blocks in the frequency
domain is specified by the higher-layer parameter ssb-subcarrierOffset. In the time
domain, SS/PBCH blocks are sent in periodical burst sets, where the number of
SS/PBCH blocks transmitted in each burst (number of antenna beams) depends on the
numerology and the frequency band of operation [12]. An example of SS/PBCH block
allocation within a slot is given in Figure 3.
14
5G-Xcast_D3.2
PDCCH:
The control information specifies the data scheduling and allocation for each UE by
means of the Downlink Control Information (DCI). This information is mapped within the
PDCCH in one or more control-channel elements (CCE). The number of CCEs allocated
in the PDCCH depends on the Aggregation Level (AL), which has five possible values
{1,2,4,8,16}. Each CCE consists of 6 REGs, where a REG is defined as one RB allocated
in one OFDM symbol. REGs are mapped in control-resource sets (CORESETs) for a
given numerology. The total number of REGs associated to each UE is mapped within
PDCCH in CORESETs packets allocated in a specific control region. Hence, the
minimum CORESET length is equal to 6 RBs x 12 REs/RB = 72 REs (1,2 MHz for µ =
0). The allocation of CORESETs in the frequency domain is specified by higher-layer
parameters. Regarding the time domain allocation, CORESETs can be transmitted at
OFDM symbols 0,1 or 2 of subframes which do not contain SS/PBCH blocks. CORESET
content can be distributed at most over three consecutive OFDM symbols, depending on
high layer parameters. CORESET also includes DMRS signals to allow the correct
demodulation of the PDCCH. Figure 4 (left) illustrates a possible CORESET allocation
with AL = 1.
PDSCH:
It contains SIBs and data content from the DL-SCH transport channel. In particular,
PDSCH is distributed in the remaining REs where the SS/PBCH and PDCCH are not
allocated. The number of RBs associated to PDSCH transmissions depends on the
available bandwidth and numerology. As for PBCH and PDCCH, PDSCH also includes
DMRS in order to ease the demodulation process. DMRS allocation depends on the
selected DMRS pattern. In addition, PDSCH also includes PT-RS and CSI-RS. An
example of PDCCH and PDSCH allocation as well as RS is shown in Figure 4 (right).
15
5G-Xcast_D3.2
Figure 4. CORESET allocation with AL = 1 (left) and PDSCH resources in a slot (right).
Acquisition Procedure
Acquisition is a basic procedure that enables the connection of UEs to the network and
provides basic information required to receive the data information carried in the PDSCH.
The acquisition procedure involves the use of PSS, SSS, PBCH, PDCCH, PDSCH and
PRACH as shown in Figure 5.
The acquisition procedure starts when the UE receives the SS/PBCH block. It includes
PSS and SSS, which provide frame synchronization and information of the physical cell
identity. Both synchronization signals are transmitted together with the PBCH. PBCH
payload contains a Master Information Block (MIB), which provides minimum system
16
5G-Xcast_D3.2
information to all UEs. It also specifies the parameter configuration needed to access the
Remaining System Information (RMSI) CORESET, which is sent over the PDCCH. An
RMSI CORESET carries a special DCI to provide the System Information Block 1 (SIB1)
scheduling. SIB1 contains information related to the availability and scheduling of other
SIBs within the cell (whether they are provided via periodic broadcast basis or only on-
demand basis) [13]. SIB1 is sent over PDSCH.
When UEs request a particular SIB, the PRACH uplink channel starts the initial access
with message 1 (Msg 1). Following the initial access request, gNB sends a random
access response (Msg 2) through PDCCH and PDSCH. Then, the UE requests the RRC
connection with Msg 3, sent via PUSCH. RRC Connection is carried through a message
exchange process. Once RRC Connection has been completed, the UE acquires the
Cell-Radio Network Temporary Identifier (C-RNTI), which uniquely identifies the link
between the gNB and the UE. Afterwards, gNB sends in the PDCCH the DCI, which is
CRC encoded and specifies where specific data is scheduled. The CRC sequence is
scrambled by the C-RNTI, which disables the reception of the serving UE content for the
rest of UEs. Once the DCI is decoded, the UE obtains the data allocation inside the
PDSCH. Finally, the UE accedes to its corresponding data region, which is also
scrambled with the C-RNTI.
Feedback
Feedback procedures are possible thanks to the uplink. They are also known as link
adaptation schemes. The three main link adaptation schemes are: Hybrid Automatic
Repeat Request (HARQ), Adaptive Modulation and Coding (AMC), and Close-Loop
MIMO.
• HARQ: used to perform physical layer retransmissions enabling transmitters to
provide higher code rates for a fixed MCS selection while decreasing the number of
received errors. HARQ ACK are transmitted in PUCCH, while data retransmissions
are sent via PDSCH.
17
5G-Xcast_D3.2
18
5G-Xcast_D3.2
19
5G-Xcast_D3.2
of the LTE eMBMS modes, e.g. LTE enTV, may disable the essential features of the
network, e.g., supporting dynamic optimisation of resource allocation or providing useful
audience metrics. Compared to ATSC 3.0 and DVB technologies such as DVB-NGH,
where spatial multiplexing techniques, longer CPs and high constellation sizes are
available, eMBMS Rel’14 is very limited and unable to support certain network
deployments, e.g., large area deployment with high spectral efficiencies.
The setup of an MBSFN area and related radio parameter configurations within the RAN
is currently done statically, thereby limiting the dynamic provisioning of such services
based on real-time traffic demands. Additionally, eMBMS requires a separate user-plane
infrastructure for connectivity of the RAN with the core network as well as for special
MBSFN subframes over the air interface. While SC-PTM enables scheduling of data
using PDSCH, the infrastructure requirements are similar to MBSFN. These aspects lead
to a significantly large footprint in added infrastructure investments for the network
operator as well as additional implementation complexity in the UE. Furthermore, this
prompts the requirement of special middleware for reliable reception of such data.
One of the key design principles adopted for the 5G RAN design would be to limit the
added footprint for delivering PTM services over the existing PTP infrastructure and
physical layer design. This would limit investment costs and implementation complexity.
Currently, the LTE eMBMS radio session setup procedure is complex and time-
consuming. This requires simplification in 5G for enabling fast and efficient 5G-Xcast
RAN sessions.
20
5G-Xcast_D3.2
The introduction of the G-RNTI enables a dynamic scheduling between PTP and PTM
within the PDSCH channel, thus allowing to create a flexible and scalable air interface
solution. The basic mechanism behind this process is illustrated in Figure 6.
In PTM transmissions, a DCI could be transmitted within a single CORESET for a group
of UEs interested in the same content. This solution avoids replication and a transmission
where several CORESETs announce the same data to all users. This modification
reduces considerably the PDCCH overhead within a NR frame. As an example, Table 2
shows the number of REs in CORESETs needed for PTP with different numbers of UEs,
and compares the results to the SC-MM solution proposed. Naturally, the larger the
number of users, the higher the overhead reduction introduced. For instance, when 10
UEs request the same content, the PDCCH overhead is reduced a 90%.
Table 2. Number of Resource Elements (REs) allocated for PDCCH CORESET with
different number of UEs for PTP and SC-MM.
Overhead
Number of Number of Number of
Reduction
UEs REs PTP REs SC-MM
(%)
2 144 50%
3 216 66%
72
5 360 80%
10 720 90%
21
5G-Xcast_D3.2
is sufficiently large, at least one of them experiences very poor link quality, and thereby
only a more robust MCS needs to be considered. In the simplest case with a very large
number of UEs requesting the same content, a fixed MCS can be chosen based on a
statistical BLER requirement. In such a case, no feedback is therefore necessarily
required. The use of link adaptation schemes can be considered as one the most efficient
feedback solutions in multicast contexts [14]. Section 5.2.2.1 of D3.4 [15] conducts link
adaptation analysis for PTM transmission with heuristic fixed offsets as well as adaptive
MCS via CQI report from a UE that has the worst radio link, SU-MIMO Precoding Matrix
Indicator (PMI) and Rank Indicator (RI) settings. It was found that the cyclic PMI
outperforms 95% coverage as compared to fixed PMI for rank 1 and the diversity benefits
of rank 2 are exhibited at lower packet loss rates for the same PMI setting, but saturate
at higher packet loss rates. In addition, it was also shown that for adaptive MCS, the
worse-UE PMI settings outperform cyclic PMI settings.
Group HARQs can be additionally considered for the SC-MM solution. When several
UEs request a HARQ/ACK retransmission via PUCCH, gNB sends a joint retransmission
via PDSCH to all of them. Different studies have proved that HARQ feedback provides
noticeable performance improvements in SC-PTM systems. However, the use of HARQ
may also introduce some limitations depending on the number of UEs requests.
Considerable overheads and latency may be introduced when large numbers of UEs ask
for retransmissions [17]. In addition, the use of HARQ also requires of an increase of the
buffer capacity at the transmitter and the receiver to store the different packet
retransmissions.
22
5G-Xcast_D3.2
SCS
μ Type CP CP (μs) ISD (km)
(kHz)
0 15 Normal 4.69 1.41
1 30 Normal 2.34 0.70
Normal 1.17 0.35
2 60
Extended 4.16 1.25
3 120 Normal 0.59 0.18
4 240 Normal 0.29 0.09
Although these ISD values are suitable for some specific scenarios such as stadiums,
campus or malls, other larger scenarios such as urban and rural environments require
higher ISD values not fulfilled with current numerologies. Naturally, a set of different
enhancements may be introduced to introduce longer ranges in SFN operations.
The use of negative numerologies combined with extended CPs represent one of the
main improvements to achieve this goal. Negative numerologies are selected through
negative integer values for µ. A negative µ implies a narrower subcarrier spacing, longer
OFDM symbol length and therefore a longer cyclic prefix, directly equivalent to larger
ISD distances in SFN environments. The combination of this concept with the use of an
extended CP of 1/4 (only available in PTP for µ = 2), may increase even more the SFN
coverage.
The main problem encountered with negative numerologies is the framing limitation.
Since 5G framing defines a fixed duration per subframe of 1 ms with a fixed number of
OFDM symbols per slot, the number and duration of slots are affected by the numerology
(See Section 2.1). With negative numerologies, the slot duration increases and slots
span over more than one subframe in the time domain. Due to the slot expansion,
small changes would need to be introduced in the framing structure to create a solution
compatible with negative integer numerologies. According to [18], a mini-slot structure
would enable the use of negative numerologies, with slots spanning over different
subframes or even frames. Mini-slots are a small framing unit formed by a group of 2, 4
or 7 OFDM symbols (normal frames are formed by 14 symbols). They offer a better
granularity and flexibility, as well as lower latencies that may benefit some 5G scenarios
[19].
An example of compatible solution to integrate mini-slots in 5G negative numerology
framing is shown in Figure 7. When numerology µ = - 1 and extended CP are selected,
mini-slots cover two OFDM symbols. In order to occupy the whole slot length, six mini-
slots are required. With µ = - 2, mini-slots can group up to four OFDM symbols with three
mini-slots filling each slot.
23
5G-Xcast_D3.2
As Figure 8 shows, if larger numerology values are considered, the number of mini-slots
required to fill the slot structure is not an integer. This limits the use of negative
numerologies to the value -2. In 5G-Xcast, only numerologies µ = 0, µ = - 1 and µ = - 2
are considered as compatible solutions with a mini-slot frame structure.
Table 4 shows the symbol duration, extended CP duration and ISD obtained when using
negative numerologies with values down to µ = - 2.
Table 4. Extended cyclic prefixes for 5G negative numerologies.
The use of an extended CP allows to deploy SFNs in small, medium and large cells for
LPLT environments with a constant CP overhead (𝑂𝑂𝑂𝑂𝐶𝐶𝐶𝐶 ):
𝑇𝑇𝑈𝑈
𝑂𝑂𝑂𝑂𝐶𝐶𝐶𝐶 = = 0.2 (2)
𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶
24
5G-Xcast_D3.2
negative numerologies, PDCCH will be allocated at the first one or two OFDM symbols,
depending on the new numerology configuration.
3.3.3 Channel Estimation
New DMRS signals need to be designed as well to enable the channel estimation
process in large SFN areas. Patterns can be defined by 𝐷𝐷𝑥𝑥 , which is the separation of
reference signals along different OFDM symbols in frequency domain, and 𝐷𝐷𝑦𝑦 , which
specifies the separation of reference signals for the same subcarrier in time domain. 𝐷𝐷𝑥𝑥
and 𝐷𝐷𝑦𝑦 directly influence the Nyquist limit (𝑇𝑇𝑃𝑃 ) and the maximum mobility tolerance
(𝑓𝑓𝐷𝐷𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 ), which at the same time affects the SFN operation performance. This impact is
shown as follows:
𝑇𝑇𝑈𝑈
𝑇𝑇𝑃𝑃 = (3)
𝐷𝐷𝑥𝑥
1
𝑓𝑓𝐷𝐷𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = (4)
2𝐷𝐷𝑦𝑦 (𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶 )
Figure 9. MC-MM Control region and DMRS patterns for negative numerologies.
The use of the same DMRS pattern for all numerology configurations logically leads to
1
the same DMRS overhead (𝑂𝑂𝑂𝑂𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = = 0.1666). More details about the DMRS
𝐷𝐷𝑥𝑥 𝐷𝐷𝑦𝑦
patterns considered for the MC-MM solution are given in Table 5.
25
5G-Xcast_D3.2
SCS
μ 𝑫𝑫𝒙𝒙 𝑫𝑫𝒚𝒚 Overhead (%) 𝐓𝐓𝐮𝐮 (μs) CP (μs) 𝐓𝐓𝐩𝐩 (μs) 𝒇𝒇𝑫𝑫𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍 (𝑯𝑯𝑯𝑯)
(kHz)
0 15 66,66 16,66 33,33 2000
-1 7,5 2 3 16,66 133,33 33,33 66,66 1000
-2 3,75 266,66 66,66 133,33 500
The obtained Nyquist limits do not imply a limitation in the SFN operation since they are
larger than the corresponding CP. Regarding the maximum mobility conditions, the
obtained speed limit is set up to 771 km/h (µ = -2) and 3085 km/h (µ = 0) in a frequency
range of 700 MHz and up to 135 km/h (µ = -2) and 540 km/h (µ = -2) in the range of 4
GHz.
3.3.4 Feedback
Link adaptation schemes used in SC-MM are here reused. Moreover, this mode
considers feedback coordination between cells in multi-cell environments. While in SC-
MM HARQ retransmissions are done via PDSCH from a single gNB, MC-MM may
require coordinated retransmissions from more than one gNB depending on the UE
location. For example, HARQ efficiency for one UE located between multiple serving
cells may be increased if coordinated retransmissions are received from multiple gNBs.
Nevertheless, in a multi-cell scenario, it is very likely to have reliable delivery calls for a
very robust fixed MCS when a large number of UEs receive the same content. In that
case, the role of feedback might not be crucial. Similar to SC-MM, a fixed MCS chosen
in a statistical manner could be used, rendering any uplink feedback unnecessary.
26
5G-Xcast_D3.2
more flexible way, at the expense of breaking the backwards compatibility with 5G NR
PTP, SC-MM and MC-MM.
Three extended CP combinations are proposed to cover ISDs from 30 km up to 120 km.
Wider CP values naturally are associated to narrow subcarrier spacing values. The set
of CP values is shown in Table 6.
As it can be seen, these CP combinations allow covering different ISDs while keeping
the CP overhead of 20% from previous designed modes. The specific case of 0,625 kHz
SCS needs an additional change, with each subframe having 2 ms length (5
subframes per frame).
3.4.2 Channel Estimation
The introduction of new SCS configurations leads to the design of specific DMRS
patterns to ease the channel estimation process in large scale SFN scenarios. Thanks
to the creation of a PTBCH, it is possible to create specific patterns able to optimize the
equalization interval length and the maximum mobility tolerance easily. Dense patterns
increase the SFN equalization interval and mobility tolerance at the expense of
introducing larger overheads. The DMRS allocation in PTBCH is shown in Figure 10.
27
5G-Xcast_D3.2
Table 7 shows the proposed DMRS patterns. These patterns permit longer equalization
intervals while keeping the same overhead, i.e. 16.66 %, with all possible configurations.
The TB mode is also able to cope with high mobility conditions due to the dense pattern
design in time domain (𝐷𝐷𝑦𝑦 = 2). In particular, the introduction of a CP of 100 µs supposes
a good alternative when high mobility requirements are needed. According to the
obtained Doppler limits, speeds up to 192 km/h (SCS = 0.625 kHz) and 771 km/h (SCS
= 2.5 kHz) for 700 MHz and up to 33 km/h (SCS = 0.625 kHz) and 135 km/h (SCS = 2.5
kHz) for 4 GHz are supported.
Table 7. Terrestrial Broadcast DMRS patterns
SCS
𝑫𝑫𝒙𝒙 𝑫𝑫𝒚𝒚 Overhead (%) 𝐓𝐓𝐮𝐮 (μs) CP (μs) 𝐓𝐓𝐩𝐩 (μs) 𝒇𝒇𝑫𝑫𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍 (𝑯𝑯𝑯𝑯)
(kHz)
2,5 400 100 166.66 500
1,25 3 2 16,66 800 200 266.66 250
0,625 1600 400 533.33 125
Terrestrial Broadcast also includes the design of a specific control region to enable the
100% allocation and correct reception of SFN content transmitted in the PTBCH within
the 5G NR frame. This control region, envisaged as 5G Cell Acquisition Subframe (5G
CAS), is sent in 1 out of 40 subframes and conveys PSS, SSS, PBCH, PDCCH and
PSCH content. 5G CAS is designed over the MC-MM physical layer design and therefore
it reuses the negative numerology and extended CP approach. Nevertheless, 5G CAS
use may limit the SFN coverage area. While the reuse of MC-MM physical layer only
allows to cover ISDs up to 20 km, Terrestrial Broadcast enables coverage areas up to
120 km. As a consequence, while 5G CAS performance may be affected by SFN self-
interference when echoes arrive outside CAS CP, PTBCH may not be affected due to its
larger CP configuration. However, the 5G CAS carries lower bit rates than the PTBCH
and it should be designed with robust MCS and coding schemes that allow coping with
the higher SFN self-interference.
MIMO Techniques
In this section, a general overview of the MIMO precoding for multicast and broadcast is
given, in the context of physical layer techniques for PTM transmissions, aiming to
provide some insightful guidelines for the implementation of MIMO techniques in
broadcasting and multicasting. The first subsection considers the situation when the
transmitter has channel state information (CSI) of the channel coefficients experienced
by the UEs, while the second subsection considers the situation where the transmitter
has no knowledge of the CSI.
28
5G-Xcast_D3.2
29
5G-Xcast_D3.2
which translates to a single multicast group, interference effect is ignored since all users
receive a single message. This was approximated again by Sidiropoulos et al. [31] as an
NP-hard broadcast problem by SDR and Gaussian randomisation.
3.5.2 MIMO Precoding without CSI at the Transmitter
In the situation where the transmitter does not have CSI, the MIMO precoding cannot be
designed according to the specific channel conditions experienced by the UEs but it has
to be designed to enhance the spatial diversity of the system and/or to accommodate
different deployment scenarios. DVB-NGH and ATSC 3.0 include precoding with spatial
multiplexing for MIMO 2x2 (i.e. two transmit and receive antennas) based mainly on
rotation matrices [38-39]. The rotation matrix linearly combines the information streams
of each transmit antenna, according to a rotation angle, increasing the spatial diversity
of the transmitted signal. Optimal rotation angles for MIMO 2x2 with cross-polar antennas
was investigated in [40] where it was shown that the optimal rotation angle mainly
depends on the physical layer code-rate. The concept of rotation matrices can be
extended to accommodate more general antenna configuration arrangements with
different number of transmit and receive antennas. This is especially important for the
design of the NR air interface where the base-stations can have large antenna arrays.
Hence, an important research area is to investigate MIMO precoding techniques without
CSI at the base-station for large antenna arrays. Other MIMO precoding stages can
include stream-power-allocation and power-imbalance matrices. The stream-power-
allocation matrix allows to design the power allocation to each of the information streams
that can be modulated with constellations with different cardinality. On the other hand,
the power-allocation matrix allows to design the final output power per transmit antenna
that can be useful to accommodate different deployment scenarios that may have
specific restrictions on the total power radiated per polarisation. As for the rotation
matrices, the extension of these precoding stages to larger antenna arrays is also an
important area of research.
30
5G-Xcast_D3.2
IMT-2020
KPI Vertical Methodology
Scenario
Bandwidth eMBB, URLLC M&E, PW, V2X Inspection
Peak data rate eMBB M&E Analytical
Peak spectral efficiency eMBB M&E Analytical
BICM spectral efficiency eMBB M&E Link Level
Mobility URLLC, eMBB PW, V2X, M&E Link Level
Latency URLLC PW, V2X Analytical
It is worth saying that the importance of a KPI highly depends on the scenario under
evaluation. As Table 8 shows, eMBB is the most demanding scenario in terms of data
rate, spectral efficiency, bandwidth, latency or mobility. These KPIs are needed for the
successful transmission of high throughputs in M&E use cases. On the other hand,
URLLC scenarios demand for reliable communications with very short latency but most
importantly very high user speeds.
The next subsection introduces the first scenario, eMBB, and evaluates the most
important KPIs related to it.
31
5G-Xcast_D3.2
CA BWmax
Frequency range µ BWmax (MHz) NRB CA
(GHz)
0 50 270 0.8
FR1
1 100 273 1.6
(450 MHz - 6 GHz)
2 100 135 16 1.6
FR2 2 200 264 3.2
(24.25 GHz - 52.6 GHz) 3 400 264 6.4
where:
- 𝑱𝑱 is the number of aggregated component carriers in a frequency band. It has
integer values from 1 to 16.
- 𝜶𝜶(𝒋𝒋) is the normalized scaling factor related to the ratio of resources used in the
DL/UL for the 𝑗𝑗 component carrier. In FDD, 𝛼𝛼 (𝑗𝑗) = 1 for both DL and UL. In TDD
and other duplexing techniques, both for DL and UL, 𝛼𝛼 (𝑗𝑗) is calculated based on
32
5G-Xcast_D3.2
TDD
Following the same procedure, TDD DL peak data rate values are:
33
5G-Xcast_D3.2
For FDD, one component carrier is able to provide peak data rate values up to 600 Mbps
with SISO antenna configurations and 4.87 Gbps with 8 layers in MIMO antenna
configurations in FR1. Considering TDD techniques, peak data rates up to 1.80 Gbps for
SISO and 10.85 Gbps for MIMO can be achieved in FR2.
By aggregating multiple component carriers (CC), higher peak data rate values can
be achieved. Component carriers can be either contiguous or non-contiguous in the
frequency domain. The number of component carriers has been set to the maximum, i.e.
16 component carriers. With this configuration, peak data rates up to 9.75 Gbps and
78.05 Gbps can be reached for FDD SISO and MIMO modes. In TDD, values up to
28.90 Gbps and 173.57 Gbps can be reached with SISO and MIMO configurations,
respectively. One can affirm that the use of MIMO and carrier aggregation is one of the
main factors that enables to meet the ITU-R peak data rate requirement. By enabling
just carrier aggregation in a SISO configuration, only TDD FR2 case meets this
requirement.
B. Uplink
The peak data rate in the uplink is also provided for FDD and TDD techniques. Same
SISO and MIMO configurations with single and aggregated component carriers are
considered. The assumed parameter configuration is described in Annex B.
FDD
Considering an FDD mode where all resources are assigned to UL transmissions, peak
data rate is calculated as follows:
Table 12. FDD UL peak data rate values
𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 Req.
BWmax
µ SISO Layers MIMO CA SISO+CA MIMO+CA (Gbit
(MHz)
(Gbit/s) (Gbit/s) (Gbit/s) (Gbit/s) /s)
1 50 0.30 4 1.22 16 4.90 19.60
2 100 0.62 4 2.49 16 9.99 39.99 10
3 100 0.62 4 2.49 16 9.99 39.54
TDD
Following the same procedure, TDD UL peak data rate values are calculated:
Table 13. TDD UL peak data rate values
𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 𝜸𝜸𝒑𝒑 Req.
BWmax
µ SISO Layers MIMO CA SISO+CA MIMO+CA (Gbit
(MHz)
(Gbit/s) (Gbit/s) (Gbit/s) (Gbit/s) /s)
0 50 0.18 0.75 3.00 12.03
FR1 1 0.38 1.52 6.11 24.46
100
2 0.37 4 1.50 16 6.02 24.08 10
2 200 0.73 2.94 11.79 47.16
FR2
3 400 1.47 5.91 23.64 94.57
34
5G-Xcast_D3.2
For FDD, one component carrier is able to provide peak data rate values up to 620
Mbps with SISO antenna configurations and 2.49Gbps with MIMO 4 layers configuration
in frequency ranges between 450 MHz and 6 GHz. Considering TDD techniques for
frequency ranges of 450 MHz - 6 GHz and 24.25 GHz - 52.6 GHz, peak data rates up to
1.47 Gbps for SISO and 5.91 Gbps for MIMO 4 layers can be obtained.
By aggregating multiple component carriers, higher peak data rate values can also
be reached for uplink transmissions. The number of component carriers has also been
set to 16 component carriers. With this configuration, peak data rates up to 9.99 Gbps
and 39.54 Gbps can be reached for FDD SISO and MIMO modes. In TDD, values up to
23.64 Gbps and 94.57 Gbps can be reached with SISO and MIMO configurations,
respectively. As it can be seen, the use of MIMO and carrier aggregation techniques
is also the key for uplink since it allows to meet the 10 Gbps ITU-R requirement.
A. Downlink
DL peak spectral efficiency is calculated for both FDD and TDD techniques. For FDD,
peak spectral efficiency is only calculated for FR1 while for TDD, both FR1 and FR2 are
considered. Peak spectral efficiency has been calculated per component carrier with
SISO and MIMO configurations. To enable the calculation, previous peak data rate
values have been considered.
FDD
Table 14. Peak spectral efficiency for FDD DL.
𝜼𝜼𝒑𝒑 (bit/s/Hz) 𝜼𝜼𝒑𝒑 (bit/s/Hz)
Numerology BWmax (MHz)
SISO MIMO
FR1 0 50 6.02 48.17
1 100 6.09 48.78
2 100 5.98 47.89
TDD
Table 15. Peak spectral efficiency for TDD DL.
Bandwidth 𝜼𝜼𝒑𝒑 (bit/s/Hz) 𝜼𝜼𝒑𝒑 (bit/s/Hz)
Numerology
(MHz) SISO MIMO
FR1 0 50 5.91 47.34
1 100 5.99 47.93
2 100 5.88 47.03
2 200 5.88 35.29
FR2
3 400 5.91 35.48
35
5G-Xcast_D3.2
As shown in Table 14 and Table 15, one component carrier is able to provide peak
spectral efficiency values up to 48.78 bps/Hz for FDD and up to 47.93 bps/Hz for TDD
techniques thanks to the use of MIMO 8 layers configuration. Both FDD and TDD
configurations are able to meet the ITU-R requirement (30 bps/Hz) for all the evaluated
bandwidths and numerologies.
B. Uplink
UL peak spectral efficiency is also calculated for both FDD and TDD. Same assumptions
about frequency ranges and antenna configurations have been made. The previous UL
peak data rate values have also been used for this calculation.
FDD
Frequency Numerology Bandwidth 𝜼𝜼𝒑𝒑 (bit/s/Hz) 𝜼𝜼𝒑𝒑 (bit/s/Hz)
Band (MHz) SISO MIMO
FR1 0 50 6.12 24.51
1 100 6.24 24.99
2 100 6.17 24.71
Table 16. Peak spectral efficiency for FDD UL. Numerologies from 0 to 2, with and
without MIMO.
TDD
Frequency Numerology Bandwidth 𝜼𝜼𝒑𝒑 (bit/s/Hz) 𝜼𝜼𝒑𝒑 (bit/s/Hz)
Band (MHz) SISO MIMO
FR1 0 50 5.91 23.66
1 100 6.01 24.05
2 100 5.92 23.68
2 200 5.79 23.18
FR2
3 400 5.81 23.24
Table 17. Peak spectral efficiency for TDD UL. Numerologies from 0 to 3, with and
without MIMO.
As shown in Table 16 and Table 17, one component carrier is able to provide peak
spectral efficiency values up to 24.99 bps/Hz and 24.05 bps/Hz for both FDD and TDD
techniques thanks to the use of with MIMO 4 layers configurations. All the numerology
and bandwidth combinations are able to provide values above the ITU-R requirement,
which is fixed to 15 bps/Hz.
4.1.4 BICM spectral efficiency
This KPI indicates the number of data bits per channel used to provide a minimum
carrier-to-noise ratio (CNR). The CNR is evaluated through link-level simulations. The
spectral efficiency can by simply calculated as:
𝜂𝜂𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑄𝑄𝑚𝑚 ∙ 𝑅𝑅 (7)
The selected Quality of Service (QoS) is a block error rate BLER < 0.1%. Ideal channel
estimation is used in all cases. In this section, both PDSCH and PDCCH channels have
been included.
4.1.4.1 Physical Downlink Shared Channel (PDSCH)
36
5G-Xcast_D3.2
included in the TS 38.213 [11] have been considered. The BW selected is 5 MHz with
numerology 0, i.e. a carrier spacing of 15 kHz.
0
10
-1
10
Block Error Rate (BLER)
-2
10
-3
10
-5 0 5 10 15 20 25
CNR (dB)
Figure 12. Block error rate vs CNR (dB) for SISO AWGN channel with NR.
As shown in the figure, NR Rel’15 provides a good granularity. In fact, only high CRs
with 64QAM and low CRs with 256QAM provide similar performance. That is why they
do not appear in both tables as specified in 3GPP.
Figure 13 shows the BICM spectral efficiency (bpc) as a function of the required CNR in
an AWGN channel for NR, compared with other technologies such as ATSC 3.0, 4G LTE
PTP, SC-PTM and MBSFN. The channel capacity is also shown for comparison. The
performance of NR when using a single antenna port and layer is better than 4G LTE.
The use of Low-Density Parity Check (LDPC) codes provides significant performance
gains. However, there is room for improvement since the gap to ATSC 3.0 is still high,
especially for high CNR values.
8
Shannon limit
7 ATSC 3.0
MBSFN, f = 1.25 kHz
6 SC-PTM
5G New Radio
BICM Spectral Efficiency (bit/s/Hz)
0
-7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23 25 27
CNR (dB)
Figure 13. BICM spectral efficiency vs CNR (dB) for SISO AWGN channel. NR vs. LTE
Rel’14 eMBMS and ATSC 3.0.
37
5G-Xcast_D3.2
On the one hand, ATSC 3.0 employs Non-Uniform Constellations (NUC), which provide
an important performance improvement due to the geometrical signal shaping. Note that
no gains are obtained with QPSK, since this constellation does not have room for
possible optimization. Additional gains are also obtained due to the use of longer
codewords that in turn employ long LDPC lengths. The codeword in ATSC 3.0 has 64800
bits, while the maximum length in NR is 25344 bits and it is only used with high MCS
indexes.
b) MIMO Performance
The following section shows the performance of 5G when using a MIMO configuration in
an AWGN channel. This deliverable has considered two configurations: 2x2 and 4x4
antennas, with 2 and 4 layers respectively (note that NR implements up to 8 layers), as
shown in Figure 14. The channel capacity is also shown for comparison.
16
5G New Radio - SISO
14 5G New Radio - MIMO 2 2
5G New Radio - MIMO 4 4
12 Channel capacity - SISO
Channel capacity - MIMO 2 2
BICM Spectral Efficiency (bit/s/Hz)
0
-7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23 25
CNR (dB)
Figure 14. BICM spectral efficiency vs CNR (dB) with NR, AWGN MIMO channel.
The use of multiple antennas drastically increases the BICM spectral efficiency of the
systems. For example, NR provides a capacity of approximately 5.1 bpc for a minimum
CNR of 18 dB with a single antenna. If MIMO 2x2 is used, the capacity is increased to
8.4 bpc. With MIMO 4x4, this capacity reaches a value of 13.3 bpc.
Observing the results for MIMO 2x2, it is possible to affirm that NR exceeds the SISO
capacity if the CNR is high enough, in this case from 2 dB. The same behaviour can be
seen with MIMO 4x4, which exceeds the 2x2 MIMO channel capacity when the CNR is
7 or higher.
c) IMT-2020 Scenarios
Figure 15 shows the BICM spectral efficiency vs. CNR with NR for the IMT-2020
considered scenarios, i.e. indoor hotspot, dense urban and rural. All scenarios are
evaluated considering Line-of-Sight (LoS) and compared with AWGN and the
independent and identically distributed (i.i.d.) Rayleigh channels. The frequency selected
is 700 MHz.
38
5G-Xcast_D3.2
8
Shannon limit
7 AWGN
i.i.d. Rayleigh channel
0
-5 0 5 10 15 20 25 30 35
CNR (dB)
Figure 15. BICM spectral efficiency vs CNR (dB) for all considered scenarios.
The indoor hotspot scenario is modelled with the TDL-A channel model [43, 45]. The
selected bandwidth is 10 MHz, with numerology 0. In this scenario, a user speed of 3
km/h is used. The delay spread is 20 ns. Particular indexes with low constellation orders
and high CRs have been discarded. In this case, it is better to use a higher modulation
order with a more robust CR that improves the performance while keeping the same
BICM spectral efficiency. The dense urban scenario is modelled with the TDL-C channel
model. In this scenario, a user speed of 30 km/h and a delay spread of 100 ns are used.
Finally, the rural scenario is also modelled with the TDL-C channel, but in this case, the
user speed is set to 120 km/h. The delay spread also changes for this scenario, having
30 ns in all cases.
12
Shannon limit
AWGN
10 Indoor Hotspot (LoS)
Dense Urban (LoS)
Rural (LoS)
8
BICM Spectral Efficiency (bit/s/Hz)
0
-5 0 5 10 15 20 25 30
CNR (dB)
Figure 16. BICM spectral efficiency vs CNR (dB) for TDL scenarios with 2x2 MIMO.
Figure 16 depicts the BICM spectral efficiency vs. CNR for the IMT-2020 considered
scenarios, i.e. indoor hotspot, dense urban and rural. All scenarios are evaluated with
and LoS and compared with the AWGN MIMO channel. Two transmitter and receiver
39
5G-Xcast_D3.2
antennas are used. A single codeword is transmitted by using two independent layers
as specified in [12].
4.1.4.2 Physical Downlink Control Channel (PDCCH)
a) Technology Comparison: NR vs LTE
Figure 17 shows the BER and BLER as a function of the required CNR (dB) for the
different aggregation levels (AL) of Control Channel Elements (CCE), including both New
Radio configuration and LTE in AWGN channel.
0
10
New Radio, AL = 1
New Radio, AL = 2
-1 New Radio, AL = 4
10
New Radio, AL = 8
LTE Advanced-Pro, AL = 1
LTE Advanced-Pro, AL = 2
Block Error Rate (BLER)
-2
10
-3
10
-4
10
CNR (dB)
Figure 17. Block error rate vs CNR for AWGN channel. Different ALs with NR and LTE.
The reason of choosing the number of DCI information bits 12 is because for higher
numbers zeros shall be appended to the DCI format until the payload size equals 12 [12].
This means 12 bits is the smallest possible number of DCI information bits, which should
be equivalent to the smallest possible CNR requirement for all conditions (different
aggregation levels).
From the results in Figure 17, we can see that a higher aggregation level generally gives
more protection to codewords, which is reflected in the required CNR for both LTE and
NR situation, by trading more occupied bandwidth. Due to the different CRC and
aggregation levels for LTE and NR, it is not easy to achieve a fair comparison. Generally,
Polar codes should outperform TBCC. If compared under the same AL with the
parameters shown, NR requires about 2.3dB less than LTE to achieve BLER < 10-3.
40
5G-Xcast_D3.2
b) IMT-2020 Scenarios
In this section, the SISO performance for the IMT-2020 scenarios is presented and
compared with AWGN, according to different scenarios described in subsection 4.1.5.1.
Perfect channel estimation has been used in a frequency range of 700 MHz.
The following table shows the minimum CNR needed for:
• TDL-A channel model for indoor hotspot scenario with 30ns delay spread and 3km/h.
• TDL-C channel model for rural scenario with 300ns delay spread and 30km/h.
Table 18. Minimum CNR threshold of PDCCH in AWGN, TDL-A, and TLD-C channel.
Scenario AL 1 AL 2 AL 4 AL 8
AWGN 0.3 dB -3.8 dB -7 dB -10 dB
Indoor hotspot 1.5 dB -3.2 dB -6.5 dB -9.4 dB
Rural - -3.7 dB -6.8 dB -9.9 dB
From the results, we can see that under perfect channel estimation, the higher movement
speed equivalent to better Doppler diversity which makes the CNR requirement for each
aggregation level of the TDL-C channel (with 30km/h movement speed) outperform the
corresponding point with TDL-A channel (with 3km/h movement speed).
Compared to the AWGN results, at lower aggregation level which is equivalent to high
data rates, the BLER performance for both TDL-A and TDL-C channels are worse than
AWGN result. But because of the fixed code word length, when aggregation level goes
higher, the code rate dramatically decreases and the TDL channel performances are
almost aligned with the AWGN channel. For channel models with LOS path (results not
shown here), it will give exactly the same results as the AWGN channel due to the large
Rice factor (K) of the LOS path. (i.e. TDL-D and TDL-E).
41
5G-Xcast_D3.2
1
𝑓𝑓𝐷𝐷𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = 𝐻𝐻𝐻𝐻 (9)
2𝐷𝐷𝑦𝑦 (𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶 )
where 𝐷𝐷𝑦𝑦 is the length of the reference signal sequence in OFDM symbols, 𝑇𝑇𝑈𝑈 is the
useful symbol duration, and 𝑇𝑇𝐶𝐶𝐶𝐶 is the cyclic prefix length in time. Hence, the performance
depends on the carrier spacing, system bandwidth, the operational frequency band and
the accuracy of channel estimation. For this study, a wide range of Doppler shifts is
evaluated. The obtained results can be easily mapped to the frequency bands under
evaluation by using the previous formulas. Two frequency bands are here evaluated:
700 MHz and 4 GHz.
The theoretical Doppler limits for each frequency band depend on the DMRS signal used.
Assuming Mapping Type A, DMRS configuration type 1 and DL-DMRS-add-pos = 1, i.e.
2 symbols (𝐷𝐷𝑦𝑦 = 7), the Doppler limit can be calculated as:
1 (10)
𝑓𝑓𝐷𝐷 = = 𝟗𝟗𝟗𝟗𝟗𝟗. 𝟖𝟖 𝑯𝑯𝑯𝑯
2·7· 10−6 (66.6 + 5.2)
which corresponds to a user speed of 1534 km/h at 700 MHz and 268 km/h at 4 GHz.
Note that a numerology 0 has been used in this calculation and higher numerologies will
increase the speed, since lower 𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶 will be obtained. Likewise, using a different
pattern will change the user speed limit as follows:
The IMT-2020 requirement is a user speed of 500 km/h [46]. As can be observed, every
configuration at the 700 MHz band fulfils the selected criterion. This is not the case at 4
GHz, where using numerology 0 hampers the reception if only 2 DMRS symbols are
used (marked in bold).
Table 19. Doppler limit (Hz) and related speed (km/h) for 700 MHz and 4 GHz bands.
User limit User limit
DMRS
µ 𝑫𝑫𝒚𝒚 𝒇𝒇𝑫𝑫 (km/h) (km/h)
Symbols
@ 700 MHz @ 4 GHz
2 7 994.8 1534 268
0
4 3 2319.3 3576 625
2 7 1992.6 3074 537.99
1
4 3 4638.6 7155 1251
2 7 3982.7 6144 1075
2
4 3 5566.4 14313.1 2496
These values are the maximum theoretical limits, which can be only reached if the
transmitted configuration is robust enough. The use of modulation and coding will lower
these values significantly, as the next section shows.
b) Link-level Evaluation
In order to evaluate the mobility performance of NR in the IMT-2020 evaluation context,
several channel models with a range of different user speeds have been evaluated. The
considered scenario is a Typical Urban (TU-6) with variable speed.
The three considered numerologies for frequency range 1 (450 MHz to 6 GHz) are
evaluated, i.e. 15, 30 and 60 kHz. This parameter is important when evaluating mobility,
since higher numerologies are more robust against Doppler shifts. Mapping Type A with
DMRS configuration type 1 is assumed. Different pilot patterns, i.e. patterns with 2 and
4 symbols are also explored. Real channel estimation (linear in time, linear in frequency)
with MCS 3 is used in all cases.
42
5G-Xcast_D3.2
16
14
12
CNR (dB)
10
8 = 0, 2 DMRS symbols
= 0, 4 DMRS symbols
= 1, 4 DMRS symbols
6
= 1, 2 DMRS symbols
= 2, 4 DMRS symbols
= 2, 2 DMRS symbols
4
100 300 500 700 900 1100 1300 1500 1700 2000
Figure 18. CNR against user speed for 5G New Radio in TU-6 mobile channel.
Numerologies 0,1 and 2, MCS 3.
Results in Figure 18 show that for Doppler shifts up to 500 Hz, the performance with all
configurations and real channel estimation is good enough to keep an acceptable CNR.
In this case, the higher the speed the lower the CNR required. This behaviour is obtained
because of the Doppler range used. At high Doppler shifts, it may increase as occurs
with the rest of configurations. Then, degradation appears. For higher user speeds and
some configurations, the Doppler shift starts to cause significant Inter-Carrier
Interference (ICI) and channel estimation errors. This leads to performance degradation.
The way to increase the Doppler shift limits is by using more DMRS symbols or by
increasing the numerology.
Configurations with Doppler limits are:
• Numerology 0
o 2 DMRS symbols: 600 Hz (925 km/h @ 700 MHz, 162 km/h @ 4 GHz).
o 4 DMRS symbols: 1300 Hz (2005 km/h @ 700 MHz, 351 km/h @ 4 GHz).
• Numerology 1
o 2 DMRS symbols: 1500 Hz (2314 km/h @ 700 MHz, 405 km/h @ 4 GHz).
The rest of configurations are good enough to support user speeds higher than 500 km/h
at 4GHz. Therefore, numerology 1 with more than 2 DMRS symbols is at least
needed for at 4 GHz with this modulation and coding scheme. On the other hand, all
configurations fulfil the requirement at 700 MHz. These results are slightly worse than
the theoretical limits obtained in Table 19.
The use of a higher MCS 15 further reduces these limits, as shown in Figure 19 for
numerology 0.
43
5G-Xcast_D3.2
30
= 0, 2 DMRS symbols
= 0, 4 DMRS symbols
28
26
CNR (dB)
24
22
20
50 100 150 200 250 300 350 400 450 500 550 600
Figure 19. CNR against user speed for 5G New Radio in TU-6 mobile channel.
Numerology 0, MCS 15.
Results in Figure 19 show that the degradation for the case of 2 DMRS symbols starts
in this case at 300 Hz. The value is reduced to half compared to MCS 3. A similar
behaviour can be observed when using 4 DMRS symbols. The Doppler limit is reduced
to 600 Hz in this case, having 1200 Hz with MCS 3.
4.2.1.2 Physical Downlink Control Channel (PDCCH)
This section studies the speed tolerance of the NR PDCCH with practical channel
estimation algorithms.
a) Theoretical Doppler limit
Considering the frame structure shown in Figure 2 and the CORESET allocation shown
in Figure 4 (aligned with the frame structure in [12]), it can be observed that there is no
gap between PDCCH DMRS symbols in the time domain. More specifically, for NR:
• In the frequency domain, PDCCH DMRS symbols are allocated in every 4
subcarriers;
• In the time domain, the number of PDCCH DMRS symbols is depended on the
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
value of 𝑁𝑁𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , which is determined by PCFICH and can be 1, 2 or 3.
In order to estimate the channel, a two-dimensional (i.e., frequency and time) sampling
should satisfy:
• In the frequency domain, the sampling rate must be faster than or equal to the
maximum delay spread of the channel;
• In the time domain, the sampling rate must be greater than or equal to the
maximum Doppler spread of the channel.
The maximum distance between two PDCCH DMRS symbols in the time domain, 𝑛𝑛𝑚𝑚𝑚𝑚𝑚𝑚 ,
is therefore given by:
1
𝑛𝑛𝑚𝑚𝑚𝑚𝑚𝑚 ≤ (11)
2(𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶 )𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚
where 𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚 represents the maximum Doppler spread of the channel. Due to the fact that
the PDCCH DMRS symbols cover all the time domain REs on selected subcarriers, i.e.,
44
5G-Xcast_D3.2
𝐷𝐷𝑦𝑦 = 1, thus the number of PDCCH DMRS symbols is sufficient enough to capture the
time variation of the channel with potentially a wide range of user speeds, as given by:
1
𝑓𝑓𝐷𝐷 = = 𝟔𝟔𝟔𝟔𝟔𝟔𝟔𝟔. 𝟔𝟔 𝑯𝑯𝑯𝑯 (12)
2 · 1 · 10−6 (66.6 + 5.2)
which corresponds to a user speed of 10743 km/h at 700 MHz and 1880 km/h at 4 GHz.
As previously mentioned, the numerology 0 has been used in this calculation and higher
numerologies will increase the speed, since lower 𝑇𝑇𝑈𝑈 + 𝑇𝑇𝐶𝐶𝐶𝐶 will be obtained.
b) Link-level Evaluation
To evaluate the performance of PDCCH in the mobile environments, link level
simulations are performed to, first, verify if any error floor occurs in the BICM
performance when the practical channel estimation applies and different user speeds
are considered. If not, then the required CNR to achieve BLER < 0.1% against the
Doppler shift or user speeds can be evaluated. The two-dimensional pilot-based
estimation with the linear interpolation is used. Different aggregation levels have been
considered.
8
6
AL 1
4
CNR (dB)
2 AL 2
0
AL 4
-2 AL 8
-4
100 200 400 600 800 1000 1200 1400 1600 1800 2000
Figure 20. CNR against user speed for 5G New Radio in TDL-A mobile channel.
PDCCH, real channel estimation.
As shown in the above figure, the PDCCH can handle all required user speeds for all
aggregation levels, as in [3] for all the considered frequency bands. Also, the higher
the aggregation level, the lower the required CNR, due to the better coding rate used
(half with the next level).
4.2.2 Latency
The next section describes the user plane latency that 5G NR Rel’15 introduces for
different configurations and analyses if the IMT-2020 requirement of 1 ms for URLLC is
met. The transmission and HARQ retransmission between UE and BS can be modelled
as follows:
𝑇𝑇𝑈𝑈𝑈𝑈 = 𝑇𝑇1 + 𝑝𝑝(𝑇𝑇2 + 𝑇𝑇3 ) (13)
45
5G-Xcast_D3.2
where T1 represents the time needed to transmit from the gNB to the UE, T2 is the time
required for a HARQ request, T3 is the time needed to retransmit the content and p is the
probability of a retransmission.
Example:
a) First transmission
We assume UE capability 2 [47], as it is representative of URLLC. As an example, this
analysis presents first the case of numerology 0, with slot based scheduling of 14
symbols and probability of retransmission 𝑝𝑝 = 0.1. First of all, the gBN needs some time
to process the data. This step is calculated as follows:
𝑇𝑇𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝,2 max(𝑁𝑁2 (2048 + 144) · 𝜅𝜅 · 2−µ · 𝑇𝑇𝑐𝑐 /(𝑆𝑆𝑆𝑆𝑆𝑆 · 2048),0)
𝑡𝑡𝐵𝐵𝐵𝐵,𝑡𝑡𝑡𝑡 = = (17)
2 2
where 𝑁𝑁2 is a parameter that depends on the SCS. For numerology 0 and UE capability
2, 𝑁𝑁2 is 5. Therefore, the processing time in the gNB, 𝑡𝑡𝐵𝐵𝐵𝐵,𝑡𝑡𝑡𝑡 , is 178,4 µs. To start
transmitting the content, the gNB needs to be aligned with the first possible symbol to
transmit, in this case the symbol number 0. This value depends on the moment when
the gNB starts the process. The gNB waits a minimum time 𝑡𝑡𝐹𝐹𝐹𝐹1 of 35,9 µs and a
maximum time of 964,5 µs. On average, the time needed is 500 µs. Then, the TTI is
46
5G-Xcast_D3.2
transmitted. 14 OFDM symbols with numerology 0 need 1 ms, which is the time of a
subframe. Finally, the UE processing time is calculated as follows:
𝑇𝑇𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝,1 max(𝑁𝑁1 (2048 + 144) · 𝜅𝜅 · 2−µ · 𝑇𝑇𝑐𝑐 /(𝑆𝑆𝑆𝑆𝑆𝑆 · 2048),0)
𝑡𝑡𝑈𝑈𝑈𝑈,𝑟𝑟𝑟𝑟 = = (18)
2 2
The value of 𝑁𝑁1 also depends on the SCS. In this case, 𝑁𝑁1 is 8, and therefore the
processing time, 𝑡𝑡𝑈𝑈𝑈𝑈,𝑟𝑟𝑟𝑟 , is 107 µs. The total time needed for the data transmission without
HARQ, 𝑇𝑇1 , is therefore 1,8 ms.
b) HARQ petition
If the data is not correctly received, then the UE sends the HARQ petition. Note that
processing times in the UE and BS are the same in this case. The UE then needs 107
µs to process the petition, and does not need to wait for this particular case, as Figure X
shows. Therefore, 𝑡𝑡𝐹𝐹𝐹𝐹2 is 0 µs. The time for the HARQ petition is 1 OFDM symbol, and
thus 𝑡𝑡𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 is 71,4 µs. The gNB the processes the request in 178,4 µs. In total, the HARQ
petition needs a time 𝑇𝑇2 of 356,4 µs.
c) HARQ retransmission
The gNB then needs time to process the retransmission, with the same value of 178,4
µs. The time for frame alignment in this case is 357 µs in any case, to reach the symbol
number 0 and retransmit. Then, the TTI is retransmitted in 1 ms, and the UE processes
the data again in 107 µs. The total time of retransmission 𝑇𝑇3 is 1,59 ms. The total user
plane latency with a probability of retransmission 𝑝𝑝 = 0.1 is therefore 1,98 ms.
Figure 22. User plane latency for numerologies 0 and 1, with slot based scheduling of
14 symbols. Minimum frame alignment.
Extrapolation to all configurations:
The process can be easily extrapolated to all numerologies, as well as different slot
configurations. In this deliverable, we assume FDD for the latency calculation. The
results are summarized in Table 20.
47
5G-Xcast_D3.2
Table 20. User plane latency (ms) for all considered configurations.
Slot HARQ
µ=0 µ=1 µ=2
configuration probability
𝑝𝑝 = 0 0,50 0,27 0,23
2 symbols 𝑝𝑝 = 0,1 0,58 0,32 0,27
𝑝𝑝 = 1 1,35 0,77 0,66
𝑝𝑝 = 0 0,71 0,38 0,28
4 symbols 𝑝𝑝 = 0,1 0,82 0,44 0,34
𝑝𝑝 = 1 1,85 0,95 0,78
𝑝𝑝 = 0 1,03 0,54 0,36
7 symbols 𝑝𝑝 = 0,1 1,18 0,62 0,41
𝑝𝑝 = 1 2,53 1,29 0,86
𝑝𝑝 = 0 1,80 0,92 0,55
14 symbols 𝑝𝑝 = 0,1 1,98 1,02 0,63
𝑝𝑝 = 1 3,78 1,91 1,30
48
5G-Xcast_D3.2
As shown in Table 21, depending on the PTM air interface solution, a set of different
KPIs has been considered, due to the different nature and type of service that these
modes were created for.
SC-MM introduces the design of a new multicast CORESET, which is used to convey all
the control information for all users in the cell. When the number of users in the cell is
high, it allows to reduce the PDCCH overhead from 50% up to 90%. As a consequence,
the use of a new multicast CORESET has an impact on the peak data rate and peak
spectral efficiency calculation.
MC-MM includes the design of a SFN operation mode that allows to cover ISDs up to 20
km thanks to the design of negative numerologies combined with extended CP.
According to that, BICM Spectral Efficiency is calculated in SFN scenarios for the
different numerology options. In addition, SFN coverage is obtained through link-level
simulations for different echo delays. Finally, mobility tolerance is analysed for the new
SCS, related to the new negative numerology design.
TB extends the SFN coverage up to 100 km thanks to the use of the new physical
channel PTBCH with narrow numerologies and extended CP. PTBCH introduces new
CP lengths, DMRS patterns and control regions that may have an impact on the peak
data rate and peak spectral efficiency. BICM spectral efficiency and coverage are also
analysed for the new SFN operation mode Finally, Terrestrial Broadcast mobility
tolerance is also evaluated.
49
5G-Xcast_D3.2
CC= 16
35
5G NR - PTP
SC-MM & MC-MM
30
CC= 12
25
Maximum Data Rate (Mbps)
20 CC= 8
15
CC= 4
10
CC= 2
5
0
1 20 40 60 80 100 120 135
Number of users
10
CC= 16
= 38.54 Gbps
8
CC= 12
= 28.90 Gbps
Peak data rate gain (Mbps)
6
CC= 8
= 19.24 Gbps
CC= 4
= 9.62 Gbps
2
CC= 2
= 4.80 Gbps
0
0 20 40 60 80 100 120 135
Number of users
Figure 23. Peak data rate (top) and gain obtained (bottom) with SC-MM and MC-MM
compared to PTP, for different numbers of users and different number of CCs.
There is a clear data rate gain compared to PTP services, especially for high numbers
of users. This is due to the use of a single CORESET, saving 72 REs per user. Results
50
5G-Xcast_D3.2
show that the use of unicast is not efficient in this scenario, and the data rate gain
becomes even higher when multiple carriers (by means of carrier aggregation) are used.
For example, with 16 CC and 135 users, the maximum rate gain is 9.15 Gbps, since it
is fixed to 38.62 Gbps with the MM regardless of the number of users.
In the case of the TB mode, CORESETs are transmitted within the CAS, which is
transmitted periodically in 1 out of 40 subframes. CAS transmissions and the new
PTBCH DMRS design imply an increase in the overhead, e.g. OH = 18.75% for SCS
1.25 kHz, in comparison to MC-MM, and consequently a throughput loss. The trade-off
here comes in data rate and coverage. The data rate with the TB is 30.78 Gbps.
However, although the data rate is lower, this permits to enable large coverage areas up
to 120 km.
5.1.3 Spectral Efficiency
The spectral efficiency can be easily calculated by following the procedure described in
section 4.1.3 for PTP. There is a direct relationship between spectral efficiency and data
rate, and therefore the PTM gain has similar behaviour. The specific spectral efficiency
values and gains for both modes are shown in Table 22. A numerology µ = 0 and 50
MHz of bandwidth are used.
Table 22. Spectral efficiency vs number of users.
Number of users PTP (bit/s/Hz) MM (bit/s/Hz) Gain (%)
1 48.17 0
5 47.93 0.48
10 47.51 1.37
25 46.23 48.17 4.03
50 44.09 8.45
100 39.83 17.31
135 36.84 23.51
Figure 24 shows the BICM spectral efficiency (bit/s/Hz) as a function of the required CNR
for the different MCS available in NR. Different negative numerologies have been
simulated for an echo delay equal to 16.67 µs (ISD = 5 km) and compared with LTE SC-
PTM. The channel capacity is also shown for comparison.
51
5G-Xcast_D3.2
Channel Capacity
5G MC-MM, =0
4 5G MC-MM, = -1
5G MC-MM, = -2
LTE MBSFN, SCS = 15 kHz
BICM Spectral Efficiency (bit/s/Hz)
0
-3 -1 1 3 5 7 9 11 13 15 17 19 21 23
CNR (dB)
Figure 24. BICM spectral efficiency vs. required CNR, MC-MM with 0 dB echo channel.
Channel Capacity
5G Terrestrial Broadcast (2.5 kHz)
5
5G Terrestrial Brodacast (1.25 kHz)
5G Terrestrial Broadcast (0.625 kHz)
4G LTE enTV (1.25 kHz)
4
BICM Spectral Efficiency (bits/s/Hz)
0
0 5 10 15 20 25
CNR(dB)
Figure 25. BICM spectral efficiency vs. required CNR, TB with 0 dB echo channel.
52
5G-Xcast_D3.2
allow the configuration of multiple echo delays inside and outside the CP region.
Coverage has been analysed by means of link-level simulations for all the physical
channels involved in SFN transmissions, i.e. PDSCH (MC-MM), PTBCH (TB) and
PDCCH. QoS requirements and channel estimation techniques from previous sections
are reused. For each physical channel, different SCS and CP configurations are
evaluated.
5.1.5.1 Mixed Mode: Physical Downlink Shared Channel (PDSCH)
The PDSCH coverage has been evaluated for different CP lengths associated to
negative numerologies in MC-MM, i.e. 16.66 µs (ISD = 5 km), 33.33 µs (ISD = 10 km)
and 66.66 µs (ISD = 20 km). All configurations have been simulated with MCS 2.
CP = 16.6 s CP = 33.3 s CP = 66.6 s
10
Numerology =0
Numerology = -1
Numerology = -2
8
6
CNR (dB)
Figure 26. Required CNR with different echo delays using MC-MM.
As Figure 26 depicts, the performance remains constant for all the evaluated
configurations when echoes arrive inside the CP. Otherwise, degradation due to
multipath effect is introduced. In particular, when echoes arrive between the CP and the
Nyquist limit, the degradation depends on the robustness of the MCS selected. On the
other hand, echoes outside the Nyquist limit suppose the total degradation of the system.
Therefore, the MC-MM coverage in SFN scenarios is limited by the Cyclic Prefix
and the Nyquist limit. Configurations with longer CP and longer Nyquist limits provide
higher coverage. When transmitters are deployed with ISD equal to the CP length, there
is no degradation. If the ISD is set up to the Nyquist limit, longer coverage is introduced
at the expense of introducing more degradation to the system [50].
5.1.5.2 TB Mode: Physical Terrestrial Broadcast Channel (PTBCH)
The performance for the designed CP lengths in TB, i.e.100 µs (ISD = 30 km), 200 µs
(ISD = 60 km) and 400 µs (ISD = 120 km) is shown in Figure 27. LTE enTV coverage,
with CP 200 µs, is also used as a reference. Same MCS index is also evaluated.
53
5G-Xcast_D3.2
CP = 200 s
CP = 100 s CP = 200 s CP = 400 s
10
5G TB: CP 100 s
5G TB: CP 200 s
5G TB: CP 400 s
8
LTE enTV: CP 200 s
6
CNR (dB)
Tp = 266.6 s
Tp = 133.3 s Tp = 266.6 s Tp = 533.3 s
0
0 50 100 150 200 250 300 350 400 450 500 550
Echo Delay ( s)
Figure 27. Required CNR with different echo delays using the 5G TB mode.
Figure 27 shows how 5G TB outperforms LTE enTV in terms of SFN coverage thanks
to the definition of CP = 400 µs (ISD = 120 km). In addition, TB also introduces a new
CP = 100 µs that enables to cover smaller ISD = 30 km. As described previously, CNR
degradation appears when the echo delay arrives beyond the CP region. As a
particularity, the percentage of Nyquist limit region where degradation is acceptable gets
reduced when the highest coverage mode is selected.
5.1.5.3 Physical Downlink Control Channel (PDCCH)
In this section, the performance of the NR-PDCCH channel with different coverage
settings has been evaluated. In theory, with a specific 𝜇𝜇 and therefore a fixed useful
symbol duration 𝑇𝑇𝑈𝑈 , the maximum channel delay spread is highly dependent on the
DMRS pattern, or to be specific, the pilot granularity in the frequency domain 𝐷𝐷𝑥𝑥 . The
maximum distance between two frequency domain PDCCH DMRS symbols, 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 , is
given by:
𝑇𝑇𝑈𝑈
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 ≤ (19)
𝜏𝜏𝑚𝑚𝑚𝑚𝑚𝑚
where 𝜏𝜏𝑚𝑚𝑚𝑚𝑚𝑚 represents the maximum delay spread of the channel. Take the 15kHz
subcarrier spacing as an example, also as shown in the frame structure in Figure 2 and
the CORESET allocation in Figure 4, 𝐷𝐷𝑥𝑥 = 4, one can derive 𝑇𝑇𝑈𝑈 = 1⁄Δ𝑓𝑓 = 66.7𝜇𝜇𝜇𝜇.
Therefore, the maximum channel delay spread that can be tolerated would be 𝜏𝜏𝑚𝑚𝑚𝑚𝑚𝑚 ≤
𝑇𝑇𝑈𝑈 ⁄𝐷𝐷𝑥𝑥 ≈ 16.66𝜇𝜇𝜇𝜇, greater than the one from the channel such as TDL-A and TDL-C.
Thus, similar to the NR-PDSCH, the 0dB echo channel has been modified and extended
to include different delays.
Differently from PDSCH, there is no need to perform interpolations in time domain due
to the tight pilot distribution as discussed in section 4.2.1.2. The DFT-based interpolation
is applied in frequency domain, which is a widely-used channel estimator and shows
promise in the literature [49], compared with the estimator with the other types of
interpolations such as the linear interpolation. Two set of simulations are considered:
• Re-use of CP/FFT relationship in the PTP scenario, i.e., normal CP length with
7% FFT size (only for 𝜇𝜇 = 0).
54
5G-Xcast_D3.2
• Extended CP with 25% of FFT size. Note that due to the increase of CP length,
the number of OFDM symbol in the control region decreases (𝜇𝜇 = 0, -1 and -2)
The pilot granularity remains the same in both scenarios, and the aggregation level 2 is
considered. The corresponding simulation results are presented in Figure 28.
15
= 0, CP = 16.6 s
= -1, CP = 33.3 s
= -2, CP = 66.6 s
= 0, CP = 4.67 s
10
CNR (dB)
With numerology 0, both normal and extended CPs can work properly with the time delay
equivalent to the normal CP duration. Then the larger the time delay, the higher the
required CNR for normal CP, due to the increased ISI. However, the required CNR for
the extended CP is increased even within the CP duration. Regarding the maximum
channel delay spread to be tolerated, both normal CP and extended CP can support up
to around 16𝜇𝜇𝜇𝜇, aligned with the theoretical limit as calculated previously.
Considering the different values of 𝜇𝜇, due to the increased useful symbol duration, the
case with the extended CP can tolerate the maximum channel delay spread up to the
corresponding theoretical limits. Beyond those limits, the required CNR for the PDCCH
with the current DMRS pattern (i.e., 𝐷𝐷𝑥𝑥 = 4) immediately raises to a considerably high
level. In the following we provide an analysis of these results and discuss possible
solutions.
a) Analysis
Focusing on the numerology 0, the usage of extended CP does not help to increase the
PTM coverage and the required CNR to achieve the same BLER is increased with
extended CP. Two aspects can be highlighted:
• CNR loss with increased CP
The insertion of CP disperses the transmitter energy (the amount of consumed power
depends on how large the CP length), where the signal-to-noise (SNR) lost due to the
CP introduction indicates the loss of transmission energy. The loss factor is given as:
𝑇𝑇𝐶𝐶𝐶𝐶
CNR 𝑙𝑙𝑜𝑜𝑜𝑜𝑜𝑜 = −10𝑙𝑙𝑙𝑙𝑙𝑙10 (1 − ) (20)
𝑇𝑇𝑈𝑈 + 𝑇𝑇𝑐𝑐𝑐𝑐
55
5G-Xcast_D3.2
which is equivalent to the increase of noise variance (if fixed symbol energy.), thus yields
an increased channel estimation error at the pilot position. As we can see the comparison
in Figure 29, with a relatively short 2nd path delay, both normal and extended CP
scenarios can still reconstruct the channel within an acceptable offset range. However,
in Figure 30, with higher 2nd path delay (but still inside the CP duration), with normal CP,
the channel can be reconstructed, which is not the case for the extended CP. Combining
with the second aspect, the setting with the extended CP results in that the channel
almost unable to be reconstructed.
• Not enough pilot to catch the channel variant
When the delay of the second path increases, we obtained the simulation results for
estimated and ideal channel as well. Comparing the two figures at the left side of Figure
30, with normal CP, the interpolation almost perfectly captures the channel variant.
However, with extended CP, even with perfect channel estimation at the pilot position,
the interpolation cannot reconstruct the channel due to the low pilot granularity, and this
can cause the corresponding channel estimation error is further amplified by the
increased noise due to the increase of CP (see last point in the figure), as shown in the
right two figures in Figure 30.
Normal CP, perfect estimation at pilot position Normal CP, estimations at pilot position are with noise
1
1
0.8
0.8
Channel FIR (Real part )
0.6
0.6
0.4 0.4
0.2 0.2
0 0
100 110 120 130 140 100 110 120 130 140
Subcarrier Index Subcarrier Index
Estiamted Channel Real CE with
Ideal CE
at Pilot positioon DFT Interpolation
Extended CP, perfect estimation at pilot position Extended CP, estimations at pilot position are with noise
1.5
1
Channel FIR (Real part )
0.5
0.5
0
100 110 120 130 140 100 110 120 130 140
Subcarrier Index Subcarrier Index
Figure 29. Normal & extended CP with 0.1 ∗ 𝑇𝑇𝑐𝑐𝑐𝑐 as delay for the second path.
To elaborate more on the relationship between the pilot granularity in the frequency
domain and the supported maximum delay spread of the channel, we consider the
requirement of the DFT-based channel estimation/interpolation. As discussed in [49] and
its related works, one requirement should be fulfilled in order to ensure that the DFT-
based interpolation can work properly: The number of pilots should be much greater than
the channel delay spread (counted as the number of the channel delay taps in the time
domain). The frequency domain pilot granularity in the current PDCCH DMRS pattern
cannot meet this requirement in the certain scenarios (see the time delay of 0.6 ∗ Tcp in
the Figure 27 and 25), which causes the imperfection of the DFT-based channel
estimation, thus the performance degradation of the system.
56
5G-Xcast_D3.2
Normal CP,perfect estimation at pilot position Normal CP, estimations at pilot position are with noise
1
1
Channel FIR (Real part )
0
0
100 105 110 115 120 125 130 135 140 100 105 110 115 120 125 130 135 140
Extended CP, perfect estimation at pilot position Extended CP, estimations at pilot position are with noise
1.5
1
0.8 1
Channel FIR (Real part )
0.4
0
0.2
-0.5
0
-1
100 105 110 115 120 125 130 135 140 100 105 110 115 120 125 130 135 140
Subcarrier Index Subcarrier Index
Figure 30. Normal & extended CP with 0.6 ∗ 𝑇𝑇𝑐𝑐𝑐𝑐 as delay for the second path.
Carrier
700MHz
frequency
Channel
10MHz as baseline
BW
BS antenna
10.5dBi 13dBi 13dBi 15dBi
gain
57
5G-Xcast_D3.2
Receiving antenna
10m 1.5m
height
200/800 200/800
90 200/1800 90 200/1800
200/2800 200/2800
80 200/3800 80 200/3800
200/4800 200/4800
70 400/1600 70 400/1600
400/2600 400/2600
60 400/3600 60 400/3600
Coverage Probability (%)
400/4600 400/4600
50 50
40 40
30 30
20 20
10 10
0 0
-10 -5 0 5 10 15 20 25 30 35 40 -10 -5 0 5 10 15 20 25 30 35 40
Available SINR at receiver location (dB) Available SINR at receiver location (dB)
Figure 31 SINR vs Coverage Probability at the location of minimum capacity for the
network configurations shown. Fixed rooftop reception. HPHT1 (left) and HPHT2 (right).
The SINR achieved for 95% locations is summarised in the Table 25.
Table 25 Achievable SINR (dB) at 95% locations. Fixed rooftop reception.
CP/ 200/ 200/ 200/ 200/ 200/ 400/ 400/ 400/ 400/
Tu (µs) 800 1800 2800 3800 4800 1600 2600 3600 4600
HPHT1 0.35 6.55 8.86 10.06 10.99 11.52 13.76 16.42 17.6
HPHT2 -2.19 -1.71 3.42 4.57 5.32 -1.53 5.35 6.5 7.28
58
5G-Xcast_D3.2
Table 25 shows that, for HPHT1 and HPT2, increasing the CP to 400 µs would
significantly increase the achievable SINR compared with the existing 200 µs CP. It can
also be seen that simultaneously increasing the useful symbol duration (Tu) up to 4,600
µs would be similarly beneficial. Increasing Tu in order to reduce LTE’s conventional
Cp/(Tu + CP) ratio of 20% would also reduce the overhead given over to the CP, as
Table 26 shows.
Table 26 Carrier spacings and CP overhead for numerologies used in simulations
CP/ 200/ 200/ 200/ 200/ 200/ 400/ 400/ 400/ 400/
Tu (µs) 800 1800 2800 3800 4800 1600 2600 3600 4600
Carrier Spacing (Hz) 1250 556 357 263 208 625 385 278 217
CP/(Tu + CP)
20% 10% 6.7% 5% 4% 20% 13.3% 10% 8%
Overhead
60 400/2600
Coverage Probability (%)
400/2600
400/3600 400/3600
50 400/4600 50 400/4600
40 40
30 30
20 20
10 10
0 0
-10 -5 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45
Available SINR at receiver location (dB) Available SINR at receiver location (dB)
Figure 32: SINR vs Coverage Probability at the Location of Minimum Capacity for car
mounted reception. MPMT (left), LPLT (right).
The SINR achieved for 95% locations is summarised in the Table 27.
Table 27 Achievable SINR (dB) at 95% locations. Car mounted reception
CP/ 100/ 200/ 200/ 200/ 200/ 200/ 400/ 400/ 400/ 400/
Tu (µs) 400 800 1800 2800 3800 4800 1600 2600 3600 4600
MPMT 2.1 6.0 8.8 9.2 9.2 9.3 9.2 9.5 9.5 9.5
LPLT 11.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6
Table 27 shows that there may be some merit in introducing an additional numerology
for the 200 µs CP with a longer Tu (e.g. 1800 or 2800 µs) in order to improve the capacity
in this reception mode for MPMT networks. However, the Doppler performance of such
a numerology would have to be carefully considered.
The table also shows that there would be no benefit in increasing the CP or the Tu for
this reception mode from LPLT networks. Furthermore, shortening the CP would reduce
the achievable SINR – it would degrade from 15.6 dB to 11.6 dB.
59
5G-Xcast_D3.2
However, the degradation is tolerable as a usefully high SINR would still be achievable.
Most importantly, the 100µs would improve the Doppler performance of the system by
approximately a factor of two relative to the existing 200µs numerology. Thus the shorter,
100µs CP would be a good compromise between Doppler performance and coverage
for LPLT networks.
12
10
6
0 200 400 600 800 1000 1200 1400
Figure 33. CNR against user speed for 5G-Xcast PTM modes in TU-6 mobile channel.
Results in Figure 18 show that the use of a different numerology has a great impact on
the mobility tolerance. The MM has been designed with higher SCS and therefore the
maximum user speed permitted is much higher than the TB, originally designed for fixed
reception.
The mixed mode allows Doppler shifts of up to 1350 Hz with numerology 0, equivalent
to 2080 km/h in a frequency band of 700 MHz, typical from multicast and broadcast
scenarios. The other two numerologies, i.e. -1 and -2 reduce the maximum speed to 775
and 330 Hz, equivalent to 1200 and 510 km/h. Therefore, all numerologies in the MM
fulfil the mobility requirement at 700 MHz, if the MCS selected is robust enough. In
the 4 GHz frequency band, only the numerology 0 provides user speeds higher than
250 km/h. The rest of numerologies are well below this value.
Compared to the MM, the user speeds permitted with TB mode are considerably lower.
TB has been designed to support high-demanding coverage requirements in fixed
reception scenarios, which implies the use of long CPs that in turn require narrow a
carrier spacing. This considerably limits the use of 5G for mobility scenarios. The SCS
of 2.5 kHz represents a compromise between both options. In this case, the coverage
60
5G-Xcast_D3.2
is limited, but the mobility is still relatively high, i.e. 260 Hz equivalent to 400 km/h at 700
MHz. As main drawback, this is only valid for low throughputs that are not ideal for these
type of scenarios. The SCS 1.25 kHz and 625 Hz are not suitable for mobility
conditions, since these modes only permit user speeds of 150 and 80 km/h respectively
with MCS 3. If a less robust mode is selected, maximum speeds are even lower.
5.2.2 Latency
The next section describes the user plane latency for the MM and the TB, for all
configurations designed in Section 3.
5.2.2.1 Mixed Mode
As done in section 4, we assume UE capability 2. This analysis presents first an example
for numerology -2, extended CP (3 symbols) with non-slot based scheduling of 2 symbols
and probability of retransmission 𝑝𝑝 = 0.1. It is also necessary to define values for N1 and
N2, since they are only defined in 3GPP for positive numerologies. We assume N1=2 and
N2=2,5 for numerology -2 and N1=2,5 and N2=4 for numerology -1.
a) First transmission
The processing time in the gNB, 𝑡𝑡𝐵𝐵𝐵𝐵,𝑡𝑡𝑡𝑡 , in this case is 356,8 µs. To start transmitting the
content, the gNB needs to be aligned with the first possible symbol to transmit. The gNB
waits a minimum time 𝑡𝑡𝐹𝐹𝐹𝐹1 of 309,9 µs and a maximum time of 643,2 µs. On average,
the time needed is 476,6 µs. Then, the TTI is transmitted in 666,6 µs. Finally, the
processing time in the UE, 𝑡𝑡𝑈𝑈𝑈𝑈,𝑟𝑟𝑟𝑟 , is 285,4 µs. The total time needed for the data
transmission without HARQ, 𝑇𝑇1 , is 1,78 ms.
b) HARQ petition
If the data is not correctly received, then the UE sends the HARQ petition. The UE then
needs 285,4 µs to process the petition, and waits an average time 𝑡𝑡𝐹𝐹𝐹𝐹1 of 95,8 µs. The
time for the HARQ petition is 1 OFDM symbol, having that 𝑡𝑡𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 is 333,3 µs. The gNB
then processes the request in 356,8 µs. In total, the HARQ petition needs a time 𝑇𝑇2 of
1,1 ms.
c) HARQ retransmission
The gNB processes the retransmission in 356,8 µs and, on average, needs 286,5 µs to
be aligned with the TTI and retransmit. Then, the data is retransmitted in 666,6 µs, and
the UE processes the data again in 285,4 µs. The total time of retransmission 𝑇𝑇3 is 1,6
ms. The total user plane latency with a probability of retransmission 𝑝𝑝 = 0.1 is therefore
2,05 ms.
Figure 34. User plane latency for numerology -2, with non-slot based scheduling of 2
symbols. Minimum frame alignment. Comparison with PTP µ=0.
61
5G-Xcast_D3.2
The process can be easily extrapolated to all numerologies, as well as different slot
configurations. In this deliverable, we assume FDD for the latency calculation. The
results are summarized in Table 28.
Table 28. User plane latency (ms) with the Mixed Mode.
Slot HARQ
µ = -2 µ = -1 µ=0
configuration probability
𝑝𝑝 = 0 1,78 0,93 0,56
2 symbols 𝑝𝑝 = 0,1 2,08 1,09 0,66
𝑝𝑝 = 1 4,78 2,60 1,56
𝑝𝑝 = 0 - 1,43 0,82
4 symbols 𝑝𝑝 = 0,1 - 1,66 0,95
𝑝𝑝 = 1 - 3,76 2,14
𝑝𝑝 = 0 - - 1,19
7 symbols 𝑝𝑝 = 0,1 - - 1,34
𝑝𝑝 = 1 - - 2,77
62
5G-Xcast_D3.2
6 Conclusions
Air Interface Design of 5G Mixed Mode and Terrestrial Broadcast
The 5G-Xcast air interface has been divided into three modes, i.e. Single-Cell Mixed
Mode (SC-MM), Multi-Cell Mixed Mode (MC-MM) and Terrestrial Broadcast (TB).
6.1.1 Single-Cell Mixed Mode
This mode has been designed as a flexible single-cell solution that enables the dynamic
allocation of unicast and multicast services both in the downlink and the uplink. SC-MM
has a common air interface with 5G NR Rel’15 PTP in order to keep backwards
compatibility. However, some modifications have been defined to enable group
communications in both physical downlink and uplink channels:
- G-RNTI: SC-MM includes the use of the G-RNTI, a group identifier acquired during
RRC connections that helps UEs to scramble and descramble the CRC sequence.
G-RNTI is based on C-RNTI, but it enables group communications. The definition
of a common RNTI identifier permits the simultaneous discovery of scheduling
information by groups of users interested in the same content over the same cell.
- Multicast DCI format: the delivery of multicast content may introduce some changes
on the control information (DCI bits) sent to UEs via PDCCH. The definition of new
DCI patterns based on 5G unicast formats has been explored. This would imply the
delivery of the same DCI message to all UEs interested in a particular content.
This way, UEs would access to the same PDSCH resources thanks to the same
scheduling information.
- Feedback evaluation: SC-MM has also evaluated the use of feedback mechanisms
in multicast communications. In particular, uplink mechanisms such as AMC and
HARQ feedback have been considered. If few UEs request feedback transmissions,
link adaptation schemes and HARQ retransmissions can provide considerable
performance improvements. However, if many UEs send requests, feedback
schemes may introduce considerable overheads and even collapse the PUCCH.
6.1.2 Multi-Cell Mixed Mode
It has been proposed as a scalable PTM solution to deliver unicast and multicast content
dynamically in multiple cells. It is an extension of SC-MM. Among all possible multiple
cell coordination mechanisms, 5G-Xcast has considered the inclusion of a SFN
operation mode. To enable SFN transmissions, the following air interface parameters
have been proposed:
- Common cell scrambling sequence: used to determine control and data location
as well as DMRS values. In SFN deployments, the same content is transmitted from
different sites. Therefore, MC-MM initialises the same scrambling sequence for all
the serving cells. This configuration provides SFN gains in multi-cell deployments,
mitigating interferences between adjacent cells.
- Negative numerologies and extended CP: the MC-MM solution has defined the
use of negative numerologies combined with extended CPs to increase the coverage
63
5G-Xcast_D3.2
in SFN scenarios. These numerologies have been proposed for the same PDSCH
channel. In 5G-Xcast, numerologies 0, - 1 and - 2 have been considered as
compatible solutions with the mini-slot frame structures proposed in this channel. The
lower the numerology, the larger the ISD. The largest ISD is 30 km, achieved with
numerology -2. This represents an important increase compared to PTP (1.4 km).
- New extended CPs and longer subframes: three extended CP combinations have
been proposed to cover nation-wise ISDs, i.e. 100, 200 and 400 µs. The specific
case of 400 µs expands the subframe duration to 2 ms, but is crucial to enable up to
120 km of coverage, covering the IMT-2020 requirement for broadcast coverage of
100 km.
64
5G-Xcast_D3.2
Additionally, the BICM performance of both PDSCH and PDCCH channels has been
evaluated through link-level simulations. It has been shown that the PDSCH performance
in 5G for AWGN when using a single antenna port and layer is significantly better than
4G LTE, achieving gains from 0.5 to 1 dB, depending on the MCS employed. The use
of Low-Density Parity Check (LDPC) codes provides significant performance gains.
However, there is room for improvement since the gap to ATSC 3.0 is still high, especially
for high CNR values because of the use of non-uniform constellations [54-57].
The use of MIMO drastically increases the BICM spectral efficiency. In fact, NR exceeds
the SISO capacity if the CNR is high enough. The throughputs provided in this case are
extremely high compared to SISO for the same minimum CNR requirement. The
performance of NR has been also evaluated for the IMT-2020 evaluation scenarios.
Indoor hotspot, dense urban and rural environments have been considered.
The main conclusion from PDCCH results is that a higher aggregation level (AL)
generally gives more protection to the codewords, which is reflected on the required
CNR, by trading more occupied bandwidth. The minimum CNR in this case is obtained
with AL 8, obtaining -10 dB. This value is lower than any possible value obtained with
PDSCH, which facilitates the demodulation in real scenarios.
65
5G-Xcast_D3.2
schemes are usually related to mobility scenarios. Results have shown that all possible
numerologies and DMRS signals fulfil the requirement at 700 MHz. On the other
hand, numerology 1 with more than 2 DMRS symbols is at least required at 4 GHz
with this modulation and coding scheme. The requirement of 500 km/h is not fulfilled in
any case with numerology 0. The use of a higher MCS 15 further reduces these limits.
The maximum Doppler speed value is reduced to half compared to MCS 3.
Following these results, the PDCCH has been evaluated for numerology 0, the most
demanding configuration in terms of mobility. Results have shown that the PDCCH
meets the mobility requirements for any AL and for all considered frequency bands.
This affirmation can be extended to all numerologies available in 5G NR PTP.
The results on user plane latency have shown that this KPI is highly dependent on the
numerology and the number of HARQ retransmissions. Without retransmission, the
analysis has provided a minimum user plane latency of 0.23 ms, with numerology 2.
This number increases to 0.27 ms if the probability of a retransmission is 0.1. With 1
complete retransmission, the user plane latency goes up to 0.66 ms. In any case, the
IMT-2020 requirement of 1 ms is met with 5G PTP.
66
5G-Xcast_D3.2
Regarding the PDCCH, it has been shown that with numerology 0, both normal and
extended CPs can work properly in SFN scenarios. In particular, the larger the time
delay, the higher the required CNR for normal CP, due to the increased ISI. However,
the required CNR for the extended CP is increased even within the CP duration.
Considering the different values of 𝜇𝜇, due to the increased useful symbol duration, the
case with the extended CP can tolerate the maximum channel delay spread up to the
corresponding theoretical limits.
6.3.2 Ultra-Reliable Low Latency Communications (URLLC)
This section assumes similar mobility results for control than those obtained for PTP and
focuses on data. The new numerologies defined for both MC-MM and TB modes have
been evaluated in terms of user speed and Doppler resilience. Real channel estimation
with MCS 3 is used in all cases. All numerologies in the MC-MM fulfil the mobility
requirement at 700 MHz, for the MCS under consideration, having 510 km/h as the
more limited requirement with numerology -2. In the frequency band of 4 GHz, only the
numerology 0 provides user speeds higher than 250 km/h.
User speeds permitted with the TB mode are considerably lower, since the designed
SCS are initially designed for fixed reception and therefore considerably narrower. The
SCS of 2.5 kHz is the only option where the mobility is still relatively high, i.e. 400 km/h
at 700 MHz. The SCS 1.25 kHz and 625 Hz are not suitable for mobility conditions,
since these modes only permit user speeds of 150 and 80 km/h in this representative
band.
The results on user plane latency for the Mixed Mode show that the minimum user
plane latency is 0.56 ms, with numerology 0. This number increases to 0.66 ms if the
probability of a retransmission is 0.1. With 1 complete retransmission, the user plane
latency goes up to 1.56 ms. Therefore, the requirement of 1 ms is only fulfilled if no
retransmission takes place.
With the TB mode, there is no retransmission. The lowest possible value has been
achieved with the narrowest SCS of 2.5 kHz. The latency obtained is 1.21 ms, which
does not fulfil the URLLC requirement, but it is below the 4 ms of eMBB use cases.
67
5G-Xcast_D3.2
For NR, in the current version of the 3GPP document TS 38.212 [41], the available
formats specifically for scheduling of PDSCH are:
• Format 1_0: used for the scheduling of PDSCH in one DL cell.
• Format 1_1: used for the scheduling of PDSCH in one cell.
Taking format 1_0 with CRC scrambled by C-RNTI, the information and corresponding
occupied bit positions are summarized as follows:
Table 31. DCI Format 1_0 content.
This format is for the unicast purpose, and some of the fields do not have strong relations
with PTM, such as fields related to PUCCH and for the HARQ process. Moreover, there
is no format that supports CRC (Cyclic Redundancy Check) scrambled by m-RNTI
(MBMS-RNTI) or g-RNTI (group-RNTI).
68
5G-Xcast_D3.2
a) DCI coding
After deciding the DCI format, for example, the DCI includes 𝐴𝐴 bits of information bits.
To form the PDCCH payload, the DCI has to undergo coding processes, including:
• CRC attachment: The parity bits are generated by one of the six available cyclic
generator polynomials [41], where the generator polynomial g CRC24C (D) is used,
which makes the DCI bits now has a length of (A + 24) bits.
• Code block segmentation and code block CRC attachment: depending on the
length of the input bit sequence i.e.A, it will be segmented into at most 2 code
blocks, and each of them will be attached with 24 bits of CRC parity check bits.
• Channel Coding: the details of polar encoding can be found in [41]. Given the
length of the polar encoded bits 𝑁𝑁, where 𝑁𝑁 = 2𝑛𝑛 , and the value of n is a
positive integer between 5 and 9 (including 5 and 9). Therefore, the maximum
length of the encoded bits is fixed at 29 = 512.
• Rate Matching: defined per coded block and consists of sub-block interleaving,
bit collection, and bit interleaving. But according to the 3GPP document, the flag
of doing bit interleaving is set to be 0. Denoting the length of DCI message after
rate matching as 𝐸𝐸.
b) PDCCH processing
Before being mapped onto the resource elements, the coded DCI bits have to be further
processed, including:
• Multiplexing and Scrambling: blocks of coded bits for each control channel
are multiplexed and scrambled in order to create a block of data.
• Modulation: the available modulation for control channel bits is only QPSK.
• Layer Mapping and Precoding: these processes mapped the complex-valued
modulation symbols for each of the codewords onto one or several layers and
transmitted by different antenna ports. The mapping logistics are covered in
[41].
69
5G-Xcast_D3.2
In Figure 36, the total available REs inside one CCE of this CORESET is given by:
𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = 𝐴𝐴𝐴𝐴 ∗ (𝑅𝑅𝑅𝑅𝑅𝑅 𝑝𝑝𝑝𝑝𝑝𝑝 𝐶𝐶𝐶𝐶𝐶𝐶) ∗ (𝑅𝑅𝑅𝑅 𝑝𝑝𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅) = 1 ∗ 6 ∗ 12 = 72
which means the total available bits position in one CCE equals to 72 ∗ 2 = 144 bits
(QPSK modulation) i.e., the length of bits after rate matching.
FEC
TB Encoder
CB Scrambler MAP RE Mapper OFDM
Layer
Precoding
Mapper
FEC
TB Encoder
CB Scrambler MAP RE Mapper OFDM
The channel FEC encoder is formed by segmentation, CRC attachment, LDPC coding
and rate matching. The transmission of the input data bits A is divided into the following
stages:
70
5G-Xcast_D3.2
71
5G-Xcast_D3.2
A.2.4 Scrambling
Input bits are scrambled prior to modulation for protection against burst errors. Bits are
~
multiplied by a scrambling sequence b (i ) = (b(i) + c(i)) mod 2 .
A.2.5 Modulation
Bits are transformed into complex-valued symbols using Quadrature Amplitude
Modulation (QAM). Constellation orders used are QPSK, 16QAM, 64QAM and 256QAM.
The selection of the constellation depends on the MCS index provided.
A.2.6 Layer mapping
The complex-valued modulation symbols are next mapped onto one or several
transmission layers and precoded for transmission on the antenna ports.
A.2.7 Resource Element (RE) mapping and waveform
The data symbols are located in the available elements in the resource grid for each
antenna port. Finally, the OFDM signal is generated by means of an inverse Fast Fourier
Transform and the Cyclic Prefix is inserted.
72
5G-Xcast_D3.2
Highest
modulation 8 256 QAM
(𝑗𝑗)
order 𝑄𝑄𝑚𝑚
No capability
Scaling factor of mismatch
modulation between
1
f ( j) baseband and
RF.
Max. coding
rate 984/1024 = 0.9258 Maximum CR
Rmax
µ According to
0,1,2,3
[12]
10 −3 Depending on
Tsµ Tsµ =
the numerology.
14 ⋅ 2 µ
73
5G-Xcast_D3.2
- For TDD DL, frame structure: DDDSUDDDSU (6D: Downlink, 2U: Uplink, 2S: Mixed Downlink
aaa and Uplink) and SFI = 31 with a slot structure allocating 14 OFDM symbols as: 11 DL, 1 GP and
aaaaai 2 UL. Half of the GP symbols are considered as DL resources.
74
5G-Xcast_D3.2
B.2 Uplink
Uplink Configuration
Parameters Details
FDD FR1 TDD FR1 TDD FR2
Total number of
aggregated Maximum value
16
carriers allowed
J
(𝑗𝑗)
𝛼𝛼𝑈𝑈𝑈𝑈 1 0.6375 0.6375 Note 1
Max. number of
layers Maximum value
4
v (Layers
j) allowed for UL
Highest
modulation 8 256 QAM
(𝑗𝑗)
order 𝑄𝑄𝑚𝑚
No capability
Scaling factor of mismatch
modulation between
1
f ( j) baseband and
RF.
Max. coding rate
984/1024 = 0.9258 Maximum CR
Rmax
µ According to
0,1,2,3
[12]
10 −3 Depending on
Tsµ Tsµ =
the numerology.
14 ⋅ 2 µ
- 0.0834 for BW
50 MHz, SCS 15
- 0.1194 for BW 50 - 0.1163 for BW
kHz
MHz, SCS 15 kHz 200 MHz, SCS 60
- 0.0815 for BW kHz
- 0.1163 for BW 100
OH ( j ) 100 MHz, SCS 30 Note 2
MHz, SCS 30 kHz
kHz - 0.1155 for BW
400 MHz, SCS 120
-0.0826 for BW kHz
- 0.1174 for BW 100
100 MHz, SCS 60
MHz, SCS 60 kHz
kHz
75
5G-Xcast_D3.2
76
5G-Xcast_D3.2
77
5G-Xcast_D3.2
78
5G-Xcast_D3.2
To evaluate the selected cases, 5G-Xcast adopts the following 3GPP parameter
configuration:
• SSB/PBCH: Allocation in slot 0 in each second frame. 1 slot per 20 ms.
• CORESET configuration: Full BW allocation, 2 control symbols.
• PDSCH configuration:
o Time domain: Mapping type A: Start symbol 2, duration 12 symbols.
o Frequency domain: Full BW allocation.
o Scheduling in all slots except SSB/PBCH slot. 19 out of every 20
subframes contain data.
o HARQ assumptions: RV sequence {0, 2, 3,1}. 4 HARQ processes.
• DMRS configuration: 2 DMRS symbols.
15
Throughput (Mbps)
10
0
-2 0 2 4 6 8 10 12 14 16 18 20 22
CNR (dB)
Huawei
35 Ericsson
Samsung
30 Intel
5G-Xcast
25
Throughput (Mbps)
20
15
10
0
-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
CNR (dB)
As shown in Figure 41 and Figure 42, the 5G-Xcast calibration results are fully aligned
with those obtained by different companies in 3GPP RAN WG4. As a result of this
calibration, the reliability of the simulation results provided in this deliverable is ensured.
79
5G-Xcast_D3.2
References
[1] S. Parkvall, E. Dahlman, A. Furuskar and M. Frenne, "NR: The New 5G Radio
Access Technology," IEEE Communications Standards Magazine, vol. 1, no. 4,
pp. 24-30, December 2017.
[2] 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, pp. 53-59, March 2018.
[3] D. Ratkaj and A. Murphy, Eds., “Definition of Use Cases, Requirements and
KPIs,” Deliverable D2.1, 5G-PPP 5G-Xcast project, October 2017.
[4] 3GPP TR 22.261 v16.3.0, “Service requirements for next generation new
services and markets”, April 2018.
[5] 3GPP TR 38.913 v14.3.0: “Study on Scenarios and Requirements for Next
Generation Access Technologies”, August 2017.
[6] D. Vargas and D. Mi, Eds., “LTE-Advanced Pro Broadcast Radio Access Network
Benchmark,” Deliverable D3.1, 5G-PPP 5G-Xcast project, November 2017.
[7] M. Fuentes et. al., “Physical Layer Performance Evaluation of LTE-Advanced Pro
Broadcast and ATSC 3.0 Systems”, accepted for IEEE Transactions on
Broadcasting, 2018.
[8] 3GPP RP-180672: “WID on LTE-based 5G Terrestrial Broadcast”, June 2018.
[9] 3GPP RP-180669: “SID on NR Mixed Mode Broadcast/Multicast”, June 2018.
[10] 3GPP TR 38.817-01 v1.0.0: “General Aspects for UE RF for NR”, March 2018.
[11] 3GPP TS 38.213 v15.1.0: “NR; Physical Layer Procedures for Control”, April
2018.
[12] 3GPP TS 38.211 v15.1.0: “NR; Physical Channels and Modulation”, April 2018.
[13] 3GPP TS 38.331 v15.1.0: “NR; Radio Resource Control (RRC); Protocol
Specification”, April 2018.
[14] 3GPP R2-166852: “SC-PTM feedback scheme for link adaptation and
retransmission”, October 2016.
[15] X. Zhu, D. Yang, J. Wang and X. Zhang, "Small area broadcast in LTE," 2015
International Conference on Wireless Communications & Signal Processing
(WCSP), Nanjing, China, 2015, pp. 1-5.
[16] F. Tesema and V. Pauli, Eds., “RAT Protocols and Radio Resource Management
in 5G-Xcast”, Deliverable D3.4, 5G-PPP 5G-Xcast project, November 2018.
[17] 3GPP TR 36.890 v13.0.0: “Study on Single-Cell Point-to-Multipoint Transmission
for E-UTRA (Release 13)”, July 2015.
[18] 3GPP R1-1712879: “Forward Compatibility Consideration for ECP Design and
NR-MBMS”, August 2017.
[19] 3GPP TR 38.912 v14.1.0, Release 14, “5G; Study on new radio access
technology”, October 2017.
[20] M. Joham, W. Utschick, and J. A. Nossek, “Linear Transmit Processing in MIMO
Communications Systems,” IEEE Transactions on Signal Processing, vol. 53, no.
8, pp. 2700-2712, Aug 2005.
[21] V. Stankovic and M. Haardt, “Generalized Design of Multi-User MIMO Precoding
Matrices,” IEEE Transactions on Wireless Communications, vol. 7, no. 3, pp. 953-
961, March 2008.
[22] T. K. Y. Lo, “Maximum Ratio Transmission,” IEEE Transactions on
Communications, vol. 47, no. 10, pp. 1458-1461, Oct 1999.
[23] T. Yoo and A. Goldsmith, “Optimality of Zero-forcing Beamforming with Multiuser
Diversity,” in IEEE International Conference on Communications (ICC), vol. 1,
May 2005, pp. 542-546 Vol. 1.
[24] E. Bjornson, M. Bengtsson, and B. Ottersten, “Optimal Multiuser Transmit
Beamforming: A Diffcult Problem with a Simple Solution Structure [Lecture
80
5G-Xcast_D3.2
Notes],” IEEE Signal Processing Magazine, vol. 31, no. 4, pp. 142-148, July
2014.
[25] Y. Sun and K. J. R. Liu, “Transmit Diversity Techniques for Multicasting Over
Wireless Networks,” in IEEE Wireless Communications and Networking
Conference, vol. 1, March 2004, pp. 593-598.
[26] Q. Spencer, A. Swindlehurst, and M. Haardt, “Zero-forcing Methods for Downlink
Spatial Multiplexing in Multiuser MIMO Channels,” IEEE Transactions on Signal
Processing, vol. 52, no. 2, pp. 461-471, Feb 2004.
[27] Q. Spencer, C. Peel, A. Swindlehurst, and M. Haardt, “An Introduction to The
Multi-User MIMO Downlink,” IEEE Communications Magazine, vol. 42, no. 10,
pp. 60-67, Oct 2004.
[28] Y. C. B. Silva and A. Klein, “Linear Transmit Beamforming Techniques for the
Multigroup Multicast Scenario,” IEEE Transactions on Vehicular Technology, vol.
58, no. 8, pp. 4353-4367, Oct 2009.
[29] W. Weichselberger, M. Herdin, H. Ozcelik, and E. Bonek, “A Stochastic MIMO
Channel Model with Joint Correlation of Both Link Ends,” IEEE Transactions on
Wireless Communications, vol. 5, no. 1, pp. 90-100, Jan 2006.
[30] M. Tomlinson, “New Automatic Equaliser Employing Modulo Arithmetic,”
Electronics Letters, vol. 7, no. 5, pp. 138-139, March 1971.
[31] N. D. Sidiropoulos, T. N. Davidson, and Z.-Q. Luo, “Transmit Beamforming for
Physical-layer Multi- casting,” IEEE Transactions on Signal Processing, vol. 54,
no. 6, pp. 2239-2251, June 2006.
[32] E. Karipidis, N. D. Sidiropoulos, and Z. Luo, “Quality of Service and Max-Min Fair
Transmit Beam- forming to Multiple Cochannel Multicast Groups,” IEEE
Transactions on Signal Processing, vol. 56, no. 3, pp. 1268-1279, March 2008.
[33] M. Schubert and H. Boche, “Solution of the Multiuser Downlink Beamforming
Problem with Individual SINR Constraints,” IEEE Transactions on Vehicular
Technology, vol. 53, no. 1, pp. 18-28, Jan 2004.
[34] M. Sadeghi, E. Bjornson, E. G. Larsson, C. Yuen, and T. L. Marzetta, “Max-Min
Fair Transmit Precoding for Multi-Group Multicasting in Massive MIMO,” IEEE
Transactions on Wireless Communications, vol. 17, no. 2, pp. 1358-1373, Feb
2018.
[35] D. Christopoulos, S. Chatzinotas, and B. Ottersten, “Weighted Fair Multicast
Multigroup Beamforming Under Per-antenna Power Constraints,” IEEE
Transactions on Signal Processing, vol. 62, no. 19, pp. 5132-5142, Oct 2014.
[36] Y. Gao and M. Schubert, “Group-oriented Beamforming for Multi-stream
Multicasting Based on Quality- of-service Requirements,” in IEEE International
Workshop on Computational Advances in Multi-Sensor Adaptive Processing
(CAMSAP), Dec 2005, pp. 193-196.
[37] O. T. Demir and T. E. Tuncer, “Antenna Selection and Hybrid Beamforming for
Simultaneous Wireless Information and Power Transfer in Multi-Group
Multicasting Systems,” IEEE Transactions on Wireless Communications, vol. 15,
no. 10, pp. 6948-6962, Oct 2016.
[38] D. Vargas, D. Gozalvez, D. Gomez-Barquero and N. Cardona, "MIMO for DVB-
NGH, the next generation mobile TV broadcasting," in IEEE Communications
Magazine, vol. 51, no. 7, pp. 130-137, July 2013.
[39] D. Gómez-Barquero et al., "MIMO for ATSC 3.0," in IEEE Transactions on
Broadcasting, vol. 62, no. 1, pp. 298-305, March 2016.
[40] D. Vargas, “Transmit and Receive Signal Processing for MIMO Terrestrial
Broadcast Systems,” Ph.D. dissertation, Universitat Politecnica de Valencia, May
2016.
[41] 3GPP TS 38.212 v15.1.1: “NR; Multiplexing and Channel Coding”, April 2018.
[42] 3GPP TS 38.214 v15.1.0: “NR; Physical Layer Procedures for Data”, April 2018.
81
5G-Xcast_D3.2
[43] ITU-R M2412-0: “Guidelines for Evaluation of Radio Interface Technologies for
IMT-2020”, October 2017.
[44] ITU-R WP5D: “Initial Description Template of 3GPP 5G Candidate for Inclusion
in IMT-2020”, January 2018.
[45] 3GPP TR 38.901 v14.3.0: “Study on Channel Model for Frequencies from 0.5 to
100 GHz”, January 2018.
[46] ITU-R M.2410-0: “Minimum Requirements Related to Technical Performance for
IMT-2020 Radio Interface(s)”, November 2017.
[47] 3GPP TR 37.910 v1.0.0, “Study on Self Evaluation towards IMT-2020
Submission (Release 15)”, September 2018.
[48] 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.
[49] L. Zhang, Y. Wu, W. Li, Z. Hong, K. Salehian, H. M. Kim, S. I. Park, J. Y. Lee, P.
Angueira, J. Montalban, and M. Velez, “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.
[50] 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.
[51] 3GPP TR 36.776 v0.0.3 “Study on LTE-based 5G terrestrial broadcast (Release
16)” November 2018.
[52] R1-1811588, Scenarios and simulation assumptions for the LTE based terrestrial
broadcast gap analysis, EBU, BBC, IRT; 3GPP RAN WG1 #94-Bis, Chengdu,
China, October 2018.
[53] R1-1812430, Evaluation Results for LTE-Based 5G Terrestrial Broadcast, EBU,
BBC, IRT; 3GPP RAN WG1 #95, Spokane, USA, November 2018.
[54] L. Fay, L. Michael, D. Gómez-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.
[55] L. Michael and D. Gómez-Barquero, "Bit-Interleaved Coded Modulation (BICM)
for ATSC 3.0," IEEE Trans. Broadcast., vol. 62, no. 1, pp. 181-188, March 2016.
[56] K. J. Kim et al., "Low-Density Parity-Check Codes for ATSC 3.0," IEEE Trans.
Broadcast., vol. 62, no. 1, pp. 189-196, March 2016.
[57] N. S. Loghin, J. Zöllner, B. Mouhouche, D. Ansorregui, J. Kim and S. I. Park,
"Non-Uniform Constellations for ATSC 3.0," IEEE Trans. Broadcast., vol. 62, no.
1, pp. 197-203, March 2016.
[58] 3GPP R4-1812164, “Normal PDSCH demodulation alignment simulation results,”
Intel Coorporation, October 2018.
[59] 3GPP R4-1812461, “Simulation results for NR UE PDSCH in FR1,” Samsung,
October 2018.
[60] 3GPP R4-1813632, “Simulation results for NR PDSCH demodulation
performance requirements,” Huawei, HiSilicon, October 2018.
[61] 3GPP R4-1813439, “Simulation results for NR UE PDSCH demodulation tests,”
Ericsson, October 2018.
82