Received May 13, 2021, accepted July 8, 2021, date of publication July 26, 2021, date of current version

August 17, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3099845

5G New Radio Key Performance Indicators

Evaluation for IMT-2020 Radio
Interface Technology
A. PHANI KUMAR REDDY 1 , (Graduate Student Member, IEEE),
M. SHEEBA KUMARI 2 , (Senior Member, IEEE), VISHAKHA DHANWANI 3 ,
AKHIL KALPESH BACHKANIWALA 4 , NAVIN KUMAR 5 , (Senior Member, IEEE),
K. VASUDEVAN 1 , (Senior Member, IEEE), SRINIVASAN SELVAGANAPATHY6 ,
SENDIL KUMAR DEVAR7 , PUNIT RATHOD8 , (Member, IEEE),
AND VINOSH BABU JAMES 8 , (Senior Member, IEEE)
1 Department of EE, Indian Institute of Technology Kanpur, Kanpur 208016, India
2 Saankhya Labs, Bengaluru 560001, India
3 Marvell India, Bengaluru 560103, India
4 BigCat Wireless India, Chennai 600042, India
5 Department of ECE, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Bengaluru 560035, India
6 Nokia Bell Labs, Bengaluru 560045, India
7 Ericsson India, Gurugram 122002, India
8 Qualcomm India, Bengaluru 560066, India

Corresponding author: A. Phani Kumar Reddy ([email protected])

This work was supported by the 5G India Forum of Cellular Operators Association of India (COAI) (https://www.5gindiaforum.in).

ABSTRACT The International Telecommunication Union (ITU) has recently announced the detailed 5G
specifications for International Mobile Telecommunications (IMT-2020). A number of candidate Radio
Interface Technologies (RITs) were being evaluated by the independent evaluation groups of the ITU.
Meanwhile, the roll out of fifth generation (5G) is now going on, and 5G services are offered by more
than 160 mobile network operators (MNO). This paper presents the evaluation of the proponent technolo-
gies, including the ones specified as 3rd generation partnership project (3GPP) 5G new radio (NR). The
entire 3GPP specifications were examined and evaluated through simulation using Matlab and a custom
simulator based on the Go-language. The simulator facilitated the comprehensive evaluation of the 5G NR
performance by using the IMT-2020 evaluation framework. Some of the submitted technologies displayed
certain discrepancies which were reported to ITU as well as discussed with proponents to improvise the
shortcomings. The detailed results and observations are presented in this paper.

INDEX TERMS 5G, data rate, energy efficiency, IMT-2020 evaluation, latency, mobility, new radio, spectral
efficiency.

I. INTRODUCTION Accordingly, the 3rd generation partnership project (3GPP)

The international telecommunication union radio communi- developed the 5G New radio (NR) RAT to satisfy the 5G
cation sector (ITU-R) has set the requirements for interna- requirements. It also informed that along with Long term
tional mobile telecommunications-2020 (IMT-2020) which evolution-M (LTE-M) and Narrow band-Internet of Things
defined the emergence of the fifth generation (5G) mobile (NB-IoT), they satisfy all the performance requirements
communication standards [1]. The IMT-2020 system require- of IMT-2020 (3GPP TR 38.913) [3] as a set of RAT’s.
ments for the radio access technologies (RAT) were finalized Additionally, proponents such as China, Korea, European
and adopted in the ITU-R report M.2410-0 [2] way back in telecommunications standards institute-Digital enhanced
2017. Following this acceptance, developers of RATs have cordless telecommunications (ETSI-DECT) Forum, and
been developing 5G technologies to meet these requirements. NuFront have also developed suitable technology and sub-
mitted to ITU-R as candidate technology for IMT-2020.
The associate editor coordinating the review of this manuscript and The standardization of 5G technologies for IMT-2020
approving it for publication was Young Jin Chun . typically requires the verification of these submissions.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
112290 VOLUME 9, 2021
A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

The submission is then evaluated by independent evaluation TABLE 1. List of acronyms.

groups (IEG) comprising researchers from academia and
industry and, they should be registered to ITU. 5G India
Forum (5GIF) is one of the IEGs who carried out the evalua-
tion of 3GPP NR for NuFront, enhanced ultra high throughput
(EUHT-5G) and many other proponents [4], [5]. Multiple
levels of process like, inspection, analysis, simulation, exper-
iments have been suggested by ITU. Complete guidelines,
specifications to be evaluated, etc. are found in [2].
In the standardization process, every candidate technology
is required to submit a self evaluation report to ITU-R. The
IEGs evaluate the self report and the technology for the
proposed radio interface technologies (RITs) and set of RITs
(SRITs) by following the guidelines of evaluation process as
listed by ITU-R in report M.2412 [6]. These evaluations are
then submitted to ITU-R and discussed in the working party
(WP) 5D meetings. It is expected that after a couple of such
meetings, the RITs and SRITs will be accepted as IMT-2020
(5G) standard.
The 5GIF IEG is one of the independent evaluation groups
registered with ITU-R for IMT-2020 candidate radio tech-
nology evaluation. This group was formed to evaluate the
IMT-2020 candidates from the perspective of Indian network
deployments. This is a group of operators, original equipment
manufacturers (OEM’s), universities and individual experts
participating in a collaborative manner, in the evaluation of
candidate IMT-2020 technologies of interest. This is a con-
tribution driven activity in which decision is made through a
consensus seeking approach. The technology and standards
working group of the 5GIF is involved in creating use cases
and specifications that are specifically designed for Indian
adoption. This involves the assessment of core innovations
coming from 5G standardisation bodies such as the 3GPP as
well as integration concepts and migration paths.
The remainder of the paper is organised as follows. The
technical aspects of 5G NR are briefly discussed in Section II.
Section III discusses the 5G NR simulator, Section IV
explains the evaluation process, and Section V presents and
discusses the findings and outcomes. Finally, Section VI
concludes the paper. Table 1 lists the acronyms used in the
document.

II. TECHNICAL DETAILS OF 5G NR

A. NR NUMEROLOGY
In 5G NR, the term numerology defines a combination of sub-
carrier spacing (SCS) and cyclic prefix. The specifications [7]
introduce the Greek letter µ to represent a given numerology
according to Table 2.
As can be seen from the Table 2, SCS as large as 240 kHz
is used. The reason for this is to transmit over a large band-
width without increasing the number of subcarriers. This is
important from the implementation standpoint because larger
number of subcarriers implies large fast Fourier transforms
(FFTs) that are complex/costly to implement. The SCS has

VOLUME 9, 2021 112291

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 2. Supported transmission numerologies.

a direct impact in the orthogonal frequency division multi-

plexing (OFDM) symbol duration.
FIGURE 2. Slot.
B. CHANNEL BANDWIDTH
5G NR channel bandwidths [7] and resource block (RB)
assignments enable higher spectrum usage as compared to TABLE 3. Number of slots per sub-frame.
LTE. For example: LTE 20 MHz channel vs. 5G NR 20 MHz
channel using 15 kHz SCS. LTE 20 MHz: 100RB × 12 ×
15 kHz = 18 MHz, which translates to 90% of the 20 MHz
channel being nominally occupied with OFDM subcarriers.
5G NR 20 MHz: 106RB × 12 × 15 kHz = 19.1 MHz, which
translates to 95% of the 20 MHz channel being nominally
occupied with OFDM subcarriers. It may be noted that 5G
NR protocol does support arbitrary channel bandwidths with
fewer RBs for applications requiring larger guard bands. For
are also supported by the subframe structure which can start
example, an operator could easily deploy 5G NR with only
at any symbol relative to the symbol 0 of each subframe.
100 RBs while using 15 kHz SCS over a 20 MHz spectrum
One of the motivations for a mini-slot is to support very
block, resulting in an increased guard band as compared to
low latency URLLC traffic. The mini-slots may transmit
that of the 5G NR example above.
overlapping slots while carrying eMBB data which can cause
Resource block is defined as 12 consecutive OFDM subcar-
errors in the eMBB data reception. However, this can be
riers in frequency, irrespective of the numerology. Therefore,
appropriately resolved with hybrid automatic repeat request
the resource block bandwidth changes with numerology.
(HARQ) retransmission [7]. Even though slots and mini-
slots are the basic scheduling units, 5G NR also supports
C. FRAME AND SUBFRAME STRUCTURE
scheduling of a partial slot.
In [7], 5G NR frame structure is defined. One radio frame IMT-2020 is intended to extend and enable a variety of
lasts 10 milliseconds and is divided into ten sub-frames of usage scenarios and applications that will proceed beyond
1 millisecond each as shown in Figure 1. The normal cyclic IMT-Advanced, as specified in recommendation ITU-R
prefix slots have 14 OFDM symbols and the extended cyclic M.2083 [8]. The following are three possible applications for
prefix slots have 12 OFDM symbols (supported only for the IMT-2020 [6]:
60 kHz SCS). The total number of slots per sub-frame are
• Enhanced Mobile Broadband (eMBB): In comparison to
shown in Figure 2 and Table 3. As seen from Figure 2, there
are varying length of slots to enable different applications current mobile broadband applications, this usage sce-
such as very low latency in the case of URLLC. nario would introduce additional application areas and
standards for better functionality and a more streamlined
user interface. This usage scenario includes a variety of
scenarios, such as wide-area coverage and hotspot, each
of which has its own set of specifications.
• Massive Machine Type Communications (mMTC): This
use case is defined by a large number of linked machines
exchanging a small amount of non-delay-sensitive
data.
• Ultra-Reliable and Low-Latency Communications
(URLLC): This use case has precise specifications for
throughput, latency, and availability. Wireless manage-
FIGURE 1. Frame and sub-frame.
ment of industrial processing or development systems,
The subframes in the case of 15 kHz can be aligned with remote medical surgery, delivery automation in a smart
the LTE subframe. Mini-slots with 2, 4, or 7 OFDM symbols grid, transportation protection, and so on.

112292 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

Standard is developed to work in different environment.

IMT-2020 [6] is expected to work in different scenario and
therefore, tested in the following environments:
1) Indoor Hotspot-eMBB: An indoor isolated area focused
on stationary and pedestrian users with a very large
traffic level at offices and/or shopping malls.
2) Dense Urban-eMBB: A dense urban area with large
user intensity and traffic loads, with an emphasis on
pedestrians and vehicular users.
3) Rural-eMBB: A rural community that supports pedes-
trian, vehicular, and high-speed vehicular users by pro-
viding a wider and consistent broad area coverage.
4) Urban Macro–mMTC: An urban macro ecosystem that
focuses on a large range of wired computer style
devices and aims to have continuous coverage. FIGURE 3. 5G new radio simulator setup.
5) Urban Macro–URLLC: An urban macro environment
aimed at providing ultra-reliable and low-latency com-
munications. of a scenario that is generated according to the parameters
InH_x (Indoor Hotspot), UMa_x (Urban Macro), UMi_x specified by the configuration setting. The individual link
(Urban Micro), and RMa_x (Rural Micro) are the channel quality is determined as per the network layout, pathloss
models in each setting. The test environment and correspond- and fading effects. The generate statistics are then utilized
ing channel model is given in Table 4. When a channel to infer performance. 5GIF simulator is calibrated based on
type is applied with _x suffix, in this paper, we refer to all the SINR and coupling loss CDF for 19 cell cluster. The
versions of the model, e.g., InH_x means both, InH_A and input simulation configuration files can be found at GitHub
InH_B. Similarly, if a version is referred to as ‘‘model A’’ repository.1
or ‘‘model B,’’ it refers to all instances of that variant,
e.g., ‘‘model A’’ refers to all InH_A, UMa_A, UMi_A, and
A. KEY FEATURES
RMa_A variants.
1) The simulator is capable to simulate various
TABLE 4. Mapping of channel models (reproduced from [6]). environments and scenarios such as indoor hotspot
environment, dense urban environment, rural macro
environment, urban macro environment: URLLC sce-
nario, urban macro environment: mMTC environment.
Other environments and scenarios can also be simu-
lated by setting the relevant parameters in the con-
figuration files. Two network layouts are supported:
Indoor hotspot - Rectangular network layout with 3
Sectors (Transmission and reception point (TRxP))
each; Macro - Hexagon network layout with 3 sectors
(TRxP) each.
III. 5G NR SIMULATOR
2) The simulator is capable to provide visual represen-
5GIF has developed a simulator for analyzing, interpreting
tation of the coverage of gain provided by a number
and visualizing the performance of the major key perfor-
of transceiver units (TxRU) on each TRxP of a base
mance indexes of 5G NR. The simulator is based on an
station in horizontal and vertical direction. For any
abstract modeling approach which separates the simulation
required configuration, both 2D and 3D antenna radi-
into both system level and link level. The system level sim-
ation patterns can be obtained and analysed provid-
ulations involve network evaluations, the generated statistics
ing gain for all the azimuth and zenith combinations.
of which is used by link level simulation to obtain statistics
Figure 4 shows a 2D plot of antenna array radiation
for individual BS-UE pair.
pattern and Figure 5 shows a 3D plot of antenna array
Simulator modules include configuration setting, network
radiation pattern for rural eMBB configuration A show-
layout, pathloss, fading effect, channel model and frame
ing that the beam is narrow along vertical direction and
generation. Figure 3 shows the configuration settings, pro-
wide along horizontal direction which is required in
cessing and output arrangement for the 5GIF simulator. The
rural environment. The plots demonstrate the benefit
configuration settings are categorized into those given by the
when the angle of azimuth is 0◦ and the height angle
proponent technology, customizable simulation parameters,
and those specified by ITU-R for M.2412 [6] evaluation
purposes. Processing involves randomly sampled realizations 1 https://github.com/5gif

VOLUME 9, 2021 112293

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

FIGURE 6. Antenna sector plot for RMa configuration A.

FIGURE 4. 2D Antenna array radiation pattern for rural eMBB config. A.

FIGURE 7. Pathloss vs. distance for channel model A.

FIGURE 5. Antenna array radiation pattern for 8 × 1 elements with scan

angle = 0◦ and etilt angle = 100◦ .

is 100◦ , which corresponds to the electric tilt and hori-

zontal scan used for beamforming.
3) The simulator provides the primary metrics like cou-
pling gain cumulative distribution function (CDF)
plots, SINR CDF plots, BLER vs SINR plots, etc.
4) Inside a cell, a visual representation of the SINR dis-
tribution and antenna gains distribution can be created. FIGURE 8. Pathloss vs. distance at 0.7 GHz, 7.0 GHz and 27.0 GHz.
The beamforming capacity to expand the propagation
range is visualised in Figure 6. Figure 6 shows positive
results in areas outside of the base station with adequate LMLC-NLOS scenario at 0.7 GHz frequency. Figure 8
coverage for nearby consumers and negative gains in shows pathloss vs. distance for rural channel model A
locations farther from the base station with no coverage for LOS and NLOS at 0.7 GHz, 7.0 GHz and 27.0 GHz
within the cell. frequency.
5) Pathloss and shadow fading (large-scale fading) met- 6) The simulator supports wrap around capability that is
rics can be generated. Figure 7 shows pathloss vs. to have virtual base stations apart from the 19 base
distance for rural channel model A for LOS, NLOS and stations to remove the edge cell effect.

112294 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 5. Summary of evaluation methodologies (reproduced from [6]). has 12 carriers and each carrier in turn is 30 kHz, yielding
a carrier bandwidth of 273 × 12 × 30 kHz = 98.28 MHz.
Similarly, row 3 yields a carrier bandwidth of 135 × 12 ×
60 kHz = 97.20 MHz.
In addition, NR can aggregate up to 16 such component
carriers. The peak spectral efficiency for a particular compo-
nent carrier (jth CC) can be obtained from equation based on
the specification and discussion in 3GPP [11].

(1 − OH (j) )
SEpj = (j)
BW
N BW(j),µ·12
 
(j)
× vLayers · Q(j)
m ·f (j)
· Rmax · µ (1)
Ts

948 (j)
where Rmax = 1024 . For the jth CC, vLayers represents the
(j)
maximum number of layers, Qm is the maximum modulation
order, f (j) is the scaling factor which takes values 1 and 0.75
at least. f (j) is signalled per band combination as per User
equipment (UE) capability signalling, µ is the numerology as
µ
defined in TS 38.211 [7], Ts is the average OFDM symbol
µ 10−3
duration in a subframe for µ, i.e. Ts = 14·2 µ where the
BW(j)
IV. EVALUATION METHODOLOGY normal cyclic prefix is assumed. NPRB is the maximum PRB
According to the guidelines described in M.2412 [6], the eval- allocation in bandwidth with µ, as given in section 4.5.1 of
uation of candidate technologies consists of multiple steps (TR 38.817 − 01) [12], where BW(j) is the UE supported
such as, inspection, analytical evaluation and experimental maximum bandwidth in the given band combination. OH(j)
verification of their self-evaluation submissions. The system is the overhead calculated as the average ratio of the number
performance is evaluated by considering the following key of resource elements (REs) occupied by L1/L2 control, syn-
parameters as per the specified evaluation methodology is chronization signal, physical broadcast channel (PBCH) and
given in Table 5. reference signals, etc. with respect to the total number of REs
µ
in effective bandwidth time product α (j) · BW(j) · (14 × Ts ),
V. RESULTS AND OBSERVATIONS α is the normalized scalar considering the downlink/uplink
(j)
Our team has evaluated most of these parameters following (DL/UL) ratio; for FDD α (j) = 1 for DL and UL; and for TDD
the guideline as described above. In this section, we discuss and other duplexing α (j) for DL and UL is calculated based on
them briefly including the results. the frame structure. 50 percent of guard period (GP) symbols
A. PEAK SPECTRAL EFFICIENCY are called DL overhead, and 50 percent of GP symbols are
Peak spectral efficiency is an important parameter in any considered UL overhead for GP. Given the maximum number
wireless standard. It defines how many number of bits per sec- of Tx/Rx elements in ITU-R configurations, the maximum
ond is sent in a given one Hz of bandwidth. It depends on number of TXRU allowed is upto 8 layers. Spectral efficiency
many factors, such as, channel bandwidth, sub-carrier spac- is calculated for both, DL and UL.
ing, total number of subcarrier available in OFDM symbol,
physical resource block (PRB) and so on [10]. Table 6 shows 1) DOWNLINK
the maximum number of PRBs usable for a given SCS and For frequencies in FR1, e.g. the 3.5 GHz band is considered
the channel bandwidth as defined in 3GPP RAN4. for early IMT-2020 deployments. This band is TDD band.
In FR2, 26 GHz, 28 GHz and 39 GHz bands are supported
TABLE 6. Max. number of PRBs for FR1 and FR2. in 3GPP NR specifications. 3GPP NR candidate supports
various TDD slot patterns. Table 7 shows parameters for a DL
centric configuration DDDSU (where D, S, and U stand for
downlink, special, and uplink slots) (i.e. Five slots – 3 slots
with all DL-only symbols, special slot and one slot with all
UL-only symbols). The special slot (S) – has 11 DL symbols,
1 GP (Guard), 2 UL symbols.
Different SCS and bandwidth parameters for NR TDD
Each PRB can have 12 subcarriers and will span a band- DL peak spectral efficiency is shown in Table 8 where the
width of 12 × SCS. For example in Table 6, for FR1 fre- DL dominant frame structure ‘‘DDDSU’’ (DL:UL = 4:1) is
quency range, row 2 has 273 PRBs. Each resource block chosen and the results are summarized.

VOLUME 9, 2021 112295

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 7. Assumptions for TDD DL peak spectral efficiency (DDDSU).

TABLE 8. Peak spectral efficiency (bits/s/Hz) for NR TDD DL (Frame of resource elements, overhead, etc. Our evaluation for this
structure: DDDSU, DL:UL = 4 : 1).
parameters for both DL and UL is presented.

1) DOWNLINK
For DL peak data-rate, the overheads due to synchroniza-
tion signal block (SSB), tracking reference signal (TRS),
2) UPLINK physical downlink control channel (PDCCH), phase tracking
Similarly, different SCS and bandwidth parameters for NR reference signal (PT-RS), channel state information reference
TDD UL peak spectral efficiency were evaluated for the same signal (CSI-RS), are considered. Typical values for these are
dominant frame structure ‘‘DDDSU’’ shown in Table 9. shown in Table 12.
The achievable peak spectral efficiency is shown Table 10 Bandwidths of the order of 400 MHz are required to
and, peak spectral efficiency is shown in Table 11. achieve peak data rates of 20 Gbits/s. The peak data rate is
evaluated as [16], [17] :
B. PEAK DATA RATE
5G specification defines peak data rate for DL and UL. Again DRdl = (repmat(Nslots/s , Nrows , size(BWSC , 2)))


it depends on various factors, such as bandwidth, number × NRE/slot ∗ (1 − OHdl ) ∗ Nlayers ∗ Modformat ∗ CR (2)

112296 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 9. Parameter assumptions of NR TDD UL peak spectral efficiency.

TABLE 10. Peak spectral efficiency for NR TDD UL (bit/s/Hz) (Frame length of the corresponding dimension of A, NRE = Number
structure: DDDSU).
of resource elements, OHdl = overhead on the DL.
For a 400 MHz wide component carrier, the peak data rate
is 17.49 Gbits/s. Aggregating two such component carriers
consume a bandwidth of 800 MHz and gives a peak data-
rate of about 35 Gbits/s, well beyond the passing criterion of
TABLE 11. Peak spectral efficiency observations.
20 Gbits/s shown in Table 13.

TABLE 13. Downlink peak data-rate in Gbps (1 CC).

TABLE 12. Peak data rate evaluation assumptions (FR2).

The NR capability of maximum aggregated system band-

width is presented in Table 8.1.1 − 1. of (TR 37.910) [18].
It is observed that the maximum aggregated bandwidth for
FR1 is 800 MHz to 1600 MHz and that of FR2 is 3200 MHz
to 6400 MHz.

2) UPLINK
The UL evaluation parameters are listed in Table 14. The
overheads due to demodulation reference signal (DM-RS),
where DRdl = data rate on the DL, B = repmat(A,m,n) PT-RS, sounding reference symbol (SRS), and physical UL
produces a large matrix B with m-by-n tiling of copies of A, control channel (PUCCH) are considered. The ITU peak data
s = size(A) returns a row vector with elements containing the rate targets are fulfilled with a carrier aggregation of two

VOLUME 9, 2021 112297

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 14. Evaluation assumptions for peak data-rate for uplink. bands. Similar case when using carrier aggregation to derive
user-experienced data-rate.
3GPP self-evaluation report provides support for up
to 16 CC aggregation and the user experienced data
rate for maximum available bandwidth. This is provided
in Table 18. User experienced data rate observations shown
in Table 19.

TABLE 18. Downlink maximum user experienced data rate for different
possible aggregated bandwidth.
TABLE 15. Uplink peak data-rate in Gbps (per CC).

TABLE 16. Peak data rate observations.

TABLE 19. User experienced data rate observations.

400 MHz wide carrier component, (Table 15). Also, the peak
data rate observations shown in Table 16.

C. USER EXPERIENCED DATA RATE

User experience data rate is another important service. It is Based on assessment of user experienced data rate, the fol-
evaluated in dense urban eMBB test environment for con- lowing points can be observed:
fig. A (4 GHz). Table 17 lists the DL 5% spectral efficiency 1) Multiple carrier aggregation configurations are sup-
evaluated for config. A for different bandwidth and antenna ported and can be used to improve spectrum utilization
configurations and, the corresponding user experienced data and hence user experienced data rate by using higher
rate for both UL and DL. bandwidth carriers to reduce guard bands and over-
heads.
TABLE 17. Spectral efficiency evaluation of TDD DL for different system 2) The maximum possible user experienced data rate for
bandwidths (FR1). 3GPP for 16 CC configuration is 998.2 Mbps in DL and
686.4 Mbps in UL in FR1, for the given dense urban
IMT-2020 evaluation configuration.
3) By employing carrier aggregation, it can be seen that
the minimum bandwidth required in case of DL can be
approximated to 180 MHz (100 × 0.624 + 2 × 40 ×
0.568 = 107.84 Mbps) when using 64T4R with one
100 MHz carrier and two 40 MHz carrier which are
available for use in the n77 band (3300 − 4200 MHz).
4) In case of UL user experienced data rate, by using
carrier aggregation it can be seen that the minimum
bandwidth required can be approximated to 120 MHz
The user-experienced data-rate in the case of one frequency (100 × 0.429 + 20 × 0.386 = 50.62Mbps) when using
band and one layer of transmission reception points (TRxP) 4T64R* with one 100 MHz carrier and one 20 MHz
[2], [6], is computed as in (3) carrier which are available for use in the n77 band
(3300 − 4200 MHz).
Ruser = W · SE5% (3)
This assures that Indian operators are well positioned
where SE5% is the 5th percentile user spectral efficiency and to address the national digital communications policy
W denotes the channel bandwidth. In the case bandwidth is NDCP [19] requirement using this candidate technology
aggregated across multiple bands (one or more TRxP lay- (IMT-2020/14), using a minimum bandwidth of 180 MHz in
ers), the user-experienced data-rate will be summed over the n77 Band.

112298 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

D. AREA TRAFFIC CAPACITY TABLE 21. Downlink area traffic capacity (Mbit/s/m2 ) in indoor
hotspot-eMBB at 4 GHz, Ch.Model-A.
Area traffic capacity is evaluated based on achievable average
spectral efficiency, TRxP density and the bandwidth. Let
W denote the channel bandwidth and ρ the TRxP density
(TRxP/m2 ). The area traffic capacity Carea is related to aver-
age spectral efficiency SEavg as in (4)
Carea = ρ × W × SEavg (4)
TABLE 22. Area traffic capacity observations.
In the case multiple bands are aggregated, the area traffic
capacity will be summed over the bands.
Area traffic capacity in indoor hotspot eMBB for config. A,
based on the average spectral efficiency is evaluated.

TABLE 23. Mobility interruption time observations.

FIGURE 9. Indoor hotspot site layout (reproduced from [6]).

proposed technology including the functions and the timing
involved’’. Mobility interruption time can be evaluated using
two schemes supported by 3GPP NR: Beam mobility and
TABLE 20. TRxP density.
Carrier aggregation (CA).

1) BEAM MOBILITY [2], [6]

In the beam mobility scenario, when moving within the same
Indoor hotspot site layout is shown in Figure 9 as defined cell, the transmit-receive beam pair of the user equipment
in [6], The TRxP density is calculated using (5) and the needs to be changed. gNB configures different beams for the
values: for 12 TRxP is 0.02 and 36 TRxP is 0.06 (Table 20), UE at different slots during UE mobility for DL data transmis-
where the total area of the network layout is 120 × 50 = sion. UE and gNB allocate different beams between them for
6, 000m2 . continuous DL transmission. Since there are different beams,
Number of TRxP even if one link fails, the other link maintains a connection as
ρ(TRxP/m2 ) = (5) beam pair switching happens at different slots. For UL data
Total Area of the network layout
transmission, physical UL shared channel (PUSCH) is sent
For config. A Indoor hotspot-eMBB, the DL area traffic using the beam configured by SRI (SRS resource indicator)
capacity (Mbit/s/m2 ) is shown in Table 21. Area traffic capac- by gNB. The UL communication is done by selecting a side
ity observations shown in Table 22. beam for data transmission by selecting different slots.
Based on area traffic capacity assessment following points
can be observed: 2) CA MOBILITY [2], [6]
1) Three component carriers of 100 MHz are needed to be When moving within the same Primary Cell (PCell) with
aggregated in n77 from the Indian perspective to satisfy CA enabled, the set of configured Secondary Cells (SCells)
the dense indoor area traffic capacity requirement. of the UE may change. The SCell addition procedure and
2) The available bandwidth in the sub-6 GHz mid band SCell release procedures can occur. During these procedures,
(3300 − 3600 MHz) is less than the minimum required UE can always exchange user plane packets with the gNB
300 MHz threshold, but the requirements can be met by during transitions. The data transmission between the UE and
employing a higher density of TRxP per Cell. the PCell is kept during the transition. Based on the above
analysis and procedures supported by 3GPP NR, the UE can
E. MOBILITY INTERRUPTION TIME always exchange user plane packets with gNB during the
Mobility interruption time is defined according to [2], [6] as mobility transitions. Therefore, 0 ms mobility interruption
‘‘The procedure of exchanging user plane packets with base time is achieved by NR for this scenario. Mobility interrup-
stations during transitions shall be described based on the tion time observations shown in Table 23.

VOLUME 9, 2021 112299

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

F. SPECTRAL EFFICIENCY 4) SPECTRAL EFFICIENCY CALCULATION (TDD/FDD)

1) 5th PERCENTILE USER SPECTRAL EFFICIENCY The spectral efficiency of different duplexing schemes can be
The 5th percentile user spectral efficiency is evaluated by calculated according to Report ITU-R M.2412 [6].
system level simulation using the evaluation configuration For DL average spectral efficiency and 5th percentile spec-
parameters of Indoor Hotspot-eMBB, Dense Urban-eMBB, tral efficiency,
and Rural-eMBB test environments. In case of FDD, the simulation bandwidth is 10 MHz for
(j)
Let user i in drop j correctly decode Ri (T ) accumulated DL and 10 MHz for UL. The DL average spectral efficiency
bits in [0, T]. For non-scheduled duration of user i zero is given by
bits are accumulated. During this total time user i receives PN
Ri (T )
accumulated service time of Ti ≤ T , where the service time SEavg = i=1 (8)
is the time duration between the first packet arrival and when T ·W ·M
the last packet of the burst is correctly decoded. In case of full where W is the DL bandwidth of 10 MHz; Ri (T ) denotes the
buffer, Ti ≤ T . Hence the rate normalised by service time Ti number of correctly received bits of user i, and the overhead
(j) of DL control and DL reference signals on the DL bandwidth
and channel bandwidth W of user i in drop j, ri is given in (6)
of 10 MHz is taken into account when deriving Ri (T ); and
(j) T is the simulation time. Similar notations are applied to
(j) Ri (T )
ri = (6) 5th percentile user spectral efficiency.
Ti · W
For TDD, the simulation bandwidth is 20 MHz for DL and
Running N drops simulations leads to N drops × N values UL. The DL average spectral efficiency is given by (8), where
(j)
of ri which the lowest 5th percentile point of the CDF is used W is the effective DL bandwidth that accounts for the time
to estimate the 5th percentile user spectral efficiency. frequency resource used for DL transmission (including GP
symbols); Ri (T ) denotes the number of correctly received bits
2) AVERAGE SPECTRAL EFFICIENCY of user i, and the overhead of DL control, DL reference signal
Let Ri (T ) denote the number of correctly received bits by on the DL effective bandwidth is taken into account; and
user i (i = 1, · · · , N ) (downlink) or from user i (uplink) T is the simulation time. Similar notations are applied to
in a system comprising a user population of N users and 5th percentile user spectral efficiency.
M Transmission Reception Points (TRxPs). Further, let W For UL average spectral efficiency and 5th percentile spec-
denote the channel bandwidth and T the time over which the tral efficiency, similar way is employed to derive the evalua-
data bits are received. The average spectral efficiency may be tion results for these two metrics.
estimated by running system-level simulations over P number
of drops N drops. Each drop gives a value of N i 5) SPECTRAL EFFICIENCY CALCULATION
i=1 R (T )
denoted as: R (T ), · · · , R drops (T ) and the estimated average
1 N (OH & GUARD-BAND)
spectral efficiency resulting is given by (7) To reflect the benefit of reduced guard band ratio and over-
head for larger bandwidth in NR, i.e. when the system band-
PNdrops width is larger than simulation bandwidth (10 MHz in FDD
j=1 Rj (T )
SE
c avg = and 20 MHz in TDD), the spectral efficiency can be derived
Ndrops T · W · M
PNdrops PN from (9)
i
j=1 i=1 R (T ) (1 − gb(NRB )) (1 − OH(NRB ))
= (7) SE0 = SEavg × × (9)
Ndrops T · W · M (1 − gb(NRB0 )) (1 − OH(NRB0 ))
where gb(N ) and OH (N ) is the guard band ratio and the
where SE
c avg is the estimated average spectral efficiency and
overhead at given number of RB- N , respectively, and SEavg is
will approach the actual average with an increasing number
(j) calculated by (8) For FDD, NRB0 = 52 for 10 MHz simulation
of Ndrops and Ri (T ) is the simulated total number of correctly
bandwidth and 15 kHz subcarrier spacing. For TDD, NRB0 =
received bits for user i in drop j.
51 for 20 MHz simulation bandwidth and 30 kHz subcarrier
spacing. The overhead reduction for the larger bandwidth
3) DUPLEXING SCHEME
mainly comes from the PDCCH. In addition, SSB and TRS
In NR design, the flexible duplexing scheme is available, e.g., overhead will be reduced slightly. By assuming M0 OFDM
• Different transmission directions in either part of a symbols for PDCCH at the bandwidth BW0 , the number of
paired spectrum, OFDM symbol for PDCCH at bandwidth BW could be
• TDD operation on an unpaired spectrum where the
transmission direction of most time resources can be M = BW0 /BW × M0 (10)
dynamically changing. For example, if we assume M0 = 2 for 20 MHz bandwidth
In this document, the FDD is considered for evaluation con- system, then M = 1 for 40 MHz bandwidth system. The
figurations with 700 MHz and TDD is used for configurations value of M could be a non-integer since NR supports PDCCH
with 4 GHz, 30 GHz sharing with PDSCH. The guard band ratio and PDCCH

112300 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

overhead reduction model for larger bandwidth based on (9) TABLE 27. Channel model B uplink spectral efficiency evaluation (FR2).
is considered in DL.
For frequencies in FR1, the 4 GHz band is considered for
early IMT-2020 deployments, this band is a TDD band. In the
FR2, 30 GHz bands are considered for deployment.

6) DOWNLINK SE
The DL spectral efficiency evaluation results for NR are given
in Tables 24-25. 2) InH Config. B (30 GHz) UL Avg Spectral Efficiency
meets requirements in case where the minimum num-
TABLE 24. Channel model A downlink spectral efficiency evaluation for
different bandwidths (FR1).
ber of TxRU at UE are 8 and that of BS are 32.
3) It has being observed from the SER of 3GPP that DU
config. B DL & UL both do not meet the 5th Per-
centile Spectral Efficiency requirements due to higher
losses in the mmWave (30 GHz) which not being able
to cover the cell edge users at ISD 200 m (3GPP
TR 37.910) [18].
Summary of spectral efficiency observations are given
in Table 28.

TABLE 28. Spectral efficiency observations.

TABLE 25. Channel model B downlink spectral efficiency

evaluation (FR2).

7) UPLINK SE
The UL spectral efficiency evaluation results for NR are given
in Tables 26-27.

TABLE 26. Channel model A uplink spectral efficiency evaluation (FR1).

G. SPECTRAL EFFICIENCY- SUPPLEMENTARY EVALUATION

1) FIXED WIRELESS ACCESS (FWA)
5G FWA can be used to offer an easy and affordable alter-
native to wired broadband in the lower bands of the wire-
less spectrum. FWA allows service companies to provide
high-speed connectivity to suburban and remote areas where
the expense of laying fibre is a limiting factor. It provides
ultra-high-speed broadband services to both home and enter-
prise customers using standardised 3GPP technologies and
Evolved packet core (EPC) networks. Option 3x gNBs that
help FWA and other early 5G implementations run in NSA
From Tables 24-27 and assessment of spectral efficiency mode alongside the current 4G eNodeB. When applying 5G
following points can be observed: FWA for the first time, option 3 eliminates rollout uncertain-
1) 5G NR meets the requirements of IMT-2020 since ties and variables.
InH-eMBB Config. A, Config. B, DU-eMBB Config. 5G FWA will have operation frequency capacities that
A satisfy the Spectral Efficiency requirements. are comparable to fibre optics at mmWave. 5G FWA will

VOLUME 9, 2021 112301

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

FIGURE 10. 3GPP gNodeB used for FWA applications.

be a cost-effective alternative to fixed-line DSL, wire, and

fibre in all markets with NR in the mmWave. They have FIGURE 11. Throughput comparison of PF and WDSL.
thin beams, which allows for a higher user density without
competing and interfering with other users. This satisfies the
last-mile requirement by supplying the bandwidth needed to 2) The gNodeB monitors the moving average throughput
accommodate high-definition television networks and high- Tk,n (t) for UE-k.
speed Internet access for suburban and rural residents. This 3) UE-k ∗ and PRB-n that fulfil the optimal relative chan-
increases prospects for both emerging countries with poor nel quality condition are prioritised in the t th time slot
broadband penetration and industrialised countries with slow by the scheduling process:
DSL lines.  α
At millimetre wavelengths, 5G FWA would be able to ∗ Rk,n (t)
k = arg max  β (11)
have data bandwidth equal to fibre optics (mmWave). In all k=1,2,...,K Tk,n (t)
markets with NR in the mmWave, 5G FWA would provide a 4) The choice of values for α and β decide the nature of
cost-effective option to fixed-line DSL, wire, and fibre. the scheduler.
It’s worth mentioning that the IMT-2020 KPI are geared
• α = 1 and β = 0, represents a max-rate scheduler.
for wireless applications. The KPIs for wireline and wireless
• α = 0 and β = 1, represents a round-robin
networks are vastly different. Fixed line networks aim for
scheduler.
fixed data speeds, while wireless systems aim for spectral
• α = 1 and β = 1, represents a proportionally fair
quality values. This puts an unfair pressure on the wireless
scheduler.
scheduler to service if an FWA targets those use cases. WP5D
was given a performance comparison using a Wireless DSL 5) For the WDSL scheduler, we employ α = 1 and β = 5.
(WDSL [20]) scheduler attributed to telecom centres of excel- 6) The gNodeB updates Tk,n (t) of the k th - UE in the
lence (TCoE) India during the IMT-Advanced standardisation t th slot using the exponential moving average filter:
 
process. It used a rather basic modification to the proportion- 
 1 − t1c Tk,n (t)
ally fair (PF) scheduler, changing the fairness exponent (β) 


from 1 to 5. Since, there are no follow-up research about + 1 Rk,n (t),

k∗ = k
t
Tk,n (t + 1) =  c  (12)
whether this has to be a comparative method, this method 1

 1 − tc Tk,n (t)
gives little insight into how the wireless device performs 


as timing limits are placed on the same IMT assessment · · · ,

k ∗ 6= k
process. The relation with various fairness coefficients (β) 7) Individual PRBs are treated as separate entities by the
is shown in Figure 11. While the PF scheduler strives for scheduling algorithms, which update the system after
a balance between fairness and overall system throughput, each time slot.
the WDSL scheduler strives to provide a minimum rate
While the PF scheduler aims for a combination of fairness
guarantee to the users admitted into the system as shown
and total device throughput, the WDSL scheduler aims to give
in Figure 11.
users admitted to the system a minimum rate guarantee.
a: SCHEDULER DESCRIPTION b: PERFORMANCE COMPARISON
In this segment, we include a quick overview of the scheduler The simulation setup follows the rural config. C scenario
so that interested readers can catch up on our explanation. The in V-F. The only tweak to the analysis is in rerunning the
MAC employs the following scheduling algorithm: simulation with the new value for β for the PF scheduler. The
1) The gNodeB obtains input on the instantaneous Chan- cell capacity with different values of β is listed in Table 29.
nel quality indicator (CQI) for each UE-k in time slot t If the simulation were a real deployment scenario, then
in terms of a requested data rate Rk,n (t) for any PRB-n. with the WDSL scheduler about 8 Mbps data rate per user

112302 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 29. Cell capacity (Mbps) with different β values.

can be guaranteed. However, from the operator perspective,

it only achieved about half of the call capacity.
FIGURE 13. A typical cell coverage using PC3 and PC2 devices.
2) UPLINK PERFORMANCE WITH HIGH POWER UE
Higher frequency signals can’t travel far, so cellular carriers
like Sprint worked within 3GPP as means to achieve higher
output power, specifically in the uplink (uplink defines the
cell range). Devices supporting a new power class, Power
class 2 (PC2) were chosen. PC2 was originally developed
to improve high-performance user equipment (HPUE) and
improve the 2.5 GHz LTE TDD coverage. With 3GPP NR
standardization, this functionality is being extended to several
more frequency bands in Rel-15 specifications. PC2 allows
for output power levels of 26 dBm which is twice the previous
maximum output power of PC3 (23 dBm). The higher output
power to PC2, compensates for higher propagation losses at
higher TDD frequencies, allowing carriers to retain cell cov-
erage without investing in costly infrastructure. PC2 devices
could use the same architecture as PC3 UEs, but with differ- FIGURE 14. UE’s reporting below MCS8 employing PC2 mode.

ent PAs and filters. Due to the additional power headroom

available with the higher uplink transmit power, such devices
help increase cell-edge spectral efficiency by using higher
order modulation and transport block size (Figure 12, 13).
It can also help improve overall cell-edge efficiency, espe-
cially where downlink performance is constrained by uplink
acknowledgement speed. Given that there will be some con-
nection imbalance during 5G NSA deployments, PC2 for dual
connectivity UE (one LTE band + one NR band) would be the
most realistic and suitable choice for improving uplink cover-
age for 5G NR NSA deployments. When compared to legacy
systems, HPUE improves out of service (OoS) and radio link
failures (RLF) dramatically with expanded coverage.

FIGURE 15. CDF of user SE with and without HPUE.

We assume that the UE’s are capable of PC2 and allow

the UE’s reporting below a certain Modulation and coding
scheme (MCS) value to employ PC2 (Figure 14).

b: PERFORMANCE COMPARISON
FIGURE 12. Extended coverage of PC2 devices over PC3. The simulation setup follows the same rural config. C sce-
nario in V-F. The only tweak to the analysis is in rerunning
the simulation with the link adaptation, where UE’s reporting
a: SCHEDULER DESCRIPTION below a certain MCS index were changed from PC3 (without
To understand the value proposition of HPUE to devices, HPUE) capability to PC2 (with HPUE). The CDF of spectral
we devise a simple modification to the existing IMT-2020 efficiency values seen under these scenarios is plotted below
rural low-mobility large-cell (LMLC) test scenario. for reference.

VOLUME 9, 2021 112303

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

It can be inferred from the plots that the SE of those UE’s

with very low MCS increase, whereas those with higher rates
did not change significantly. This is one move in the right
direction by 3GPP whereby the operators now have a chance
to deploy PC2 (HPUE) devices in their network to improve
cell edge or outage issues, without focusing on the need for
additional infrastructure.
Out of some supplementary studies on features supported
by the 3GPP technologies and their application to networks,
two of the studies sound promising.
1) The WDSL scheduler provided an insight into under-
standing a KPI not currently covered in IMT-2020.
If the operator were to trade off individual user perfor-
mance for cell capacity, then there is a huge trade off.
2) Similarly, a feature called HPUE defined in 3GPP FIGURE 16. Rural (4 GHz) ZoD (degree) mean value.
allows for UEs deployed in certain TDD configurations
to employ 26 dBm power amplifiers (PC2). HPUE
becomes an additional tool in the hands of operators in
addressing the coverage problem, without adding new
infrastructure.
H. MOBILITY
Mobility is defined as ‘‘The maximum mobile station speed
at which a defined Quality-of-service (QoS) can be achieved
(in km/h)’’ [2], [6].
1) MEAN VALUE OF ZOD SPREAD
The mean value of Zenith angle Of Departure (ZoD) in degree
is shown is Table 30, The CDF of mean value of ZoD spread
for line-of-sight (LOS) and non-line-of-sight (NLOS) for
Rural and Dense Urban test environment are plotted from
Figure 16 to Figure 18, respectively. FIGURE 17. Rural (700 MHz) ZoD (degree) mean value.

TABLE 30. Mean value of ZoD spread.

2) SINR DISTRIBUTION
The pre-processing SINR CDFs for eMBB test environment
are shown in Figures 19-22. From the Figures 19-22, different
test environments 50 percentile point of the CDF are listed
in Table 31.
3) LINK PROPERTIES
The results and observations of NR’s link-level mobil- FIGURE 18. Dense urban (4 GHz) ZoD (degree) mean value.
ity assessment for various test environments are shown
in Table 32 and Table 33 depending on the evaluation.
detailed specifications of these test configurations can be
I. RELIABILITY found in [6].
1) SYSTEM LEVEL SIMULATIONS For config. A, the total gain (including antenna gain) is
The system-level simulations (SLS) assumptions given presented in Figure 23 for UMa channel models A and B. The
in Table 34 are the results for the two test- resulting SINR (cell utilization 1) illustrated in Figure 24 is at
configurations A and B (4 GHz and 700 MHz) respectively; full load. The cell-edge (5th percentile) SINR is found to be

112304 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

FIGURE 19. Rural-eMBB (700 MHz) UL SINR distribution test environment. FIGURE 22. Indoor hotspot- eMBB (4 GHz) UL SINR distribution test
environment.

TABLE 31. The 50%-tile point of SINR CDF for different test environments.

TABLE 32. The uplink link level evaluation results for different test
environments for NR.

FIGURE 20. Rural- eMBB (4 GHz) UL SINR distribution test environment.

TABLE 33. Mobility observations.

FIGURE 21. Dense urban-eMBB (4 GHz) UL SINR distribution test

environment.

1.98 dB (on the DL) and 0.81 dB (on the UL) for channel
model UMa A, and 1.98 dB (DL) and 1.77 dB (UL) for
channel model UMa B as shown in Figure 25. SINR at full load (cell utilization 1) is given in Figure 27.
For config. B, the total gain (including antenna gain) is The cell-edge (5th percentile) SINR is found to be 0.16 dB
given in Figure 26. for UMa models A and B. The resulting (on the DL) and 0.83 dB (on the UL) for channel model UMa

VOLUME 9, 2021 112305

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 34. Assumptions of the system-level simulations.

FIGURE 25. 5th percentile SINR distribution for URLLC configuration A.

FIGURE 26. URLLC configuration B total gain.

FIGURE 23. URLLC configuration A total gain.

FIGURE 27. URLLC configuration B SINR distribution.

FIGURE 24. URLLC configuration A SINR distribution.

A and −0.06 dB (DL) and 0.65 dB (UL) for channel model

UMa B as shown in Figure 28. two separate datasets are used. For PDCCH, a DCI of
40 bits is presumed, without the CRC. PUCCH format 0 car-
2) LINK LEVEL SIMULATIONS ries a 1-bit UCI with a length of 2 OFDM symbols (OS)
The assumptions for link-level simulations (LLS) are and frequency hopping for PUCCH. Figure 29 shows the
described in Table 35. For the data and control channels, BLER for the control channels as a function of SNR, while

112306 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

FIGURE 28. 5th percentile SINR distribution for URLLC configuration B. FIGURE 30. LDPC BLER 4 OS-data for QPSK (1st attempt).

TABLE 35. Assumptions on the link-level simulations.

FIGURE 31. LDPC BLER 7 OS-data for QPSK (1st attempt).

TABLE 36. Success probabilities for calculating total reliability.

TABLE 37. Required number of PRBs for 32B packet and 1 OFDM symbol
overhead, at different coding rates.

FIGURE 29. Short PDCCH and PUCCH sequence selection BLER as a

function of SNR.

Figure 30 and Figure 31 displays the BLER for the data chan-
nels as a function of SNR.
success rate pt = 1 − , where  is the residual error rate.
J. TOTAL RELIABILITY With some exceptions, it is assumed that the retransmissions
The success probabilities are written on the channel level are uncorrelated, which is reasonable to assume if they are
according to Table 36, and expressions found for the total done on a different frequency allocation.

VOLUME 9, 2021 112307

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 38. Maximum number of transmissions including re-transmissions in FDD within 1 ms.

1) DL DATA, HARQ-BASED
The total reliability after N transmissions on the DL can be
described as in (13)
N X
n  
X n−1
pt = [(1 − p1 )p4 ]n−i
n−i
n=1 i=1
i−1
Y 
× p1 p2,i p1 p3 (1 − p2,j ) (13)
j=1

where for any positive integer p2,k , k is the probability of a

data block being correctly received after exactly k transmis-
sions are soft-combined. In this expression, the DL control
transmissions are seen as uncorrelated with each other and FIGURE 32. 4 OS – DL data total reliability with 1 − 3 HARQ transmissions.
with data. This is an approximation, but can be motivated by,
for example, moving the DL control between attempts. The
data attempts are correlated with each other.

2) UL DATA, CONFIGURED GRANT

With configured grant-based UL scheduling, the SR step and
the first DL control can be removed, and the total reliability
can be described as in (14)
N
X n−1
Y
pt = p2,1 + (1 − p2,1 ) p1 p2,n (1 − p1 p2,i ) (14)
n=2 i=2

Here, the PDCCH reliability starts from the first retrans-

mission, assuming perfect energy detection performance on
the PUSCH resource. FIGURE 33. 7 OS – DL data total reliability with 1 − 3 HARQ transmissions.

3) RELIABILITY ESTIMATE URLLC CONFIGURATION B, UMA B

By observing at the lower percentiles of the SINR distribu-
tions for URLLC config. B, UMa B, the channel BLER can
be found at the corresponding DL and UL SINR points. The
total error rates for DL and UL data respectively can then be
computed. The results are shown in Figure 32-34.
AL16 is assumed for PDCCH and 1% D2A level for
PUCCH. On the UL, SPS is assumed with a configured
resource every TTI. For both DL and UL, 1 − 3 transmission
attempts (including HARQ retransmissions) are considered.
The data transmissions are assumed to be correlated and are
soft-combined.

a: PACKET SIZE FIGURE 34. 4 OS UL data total reliability with 1 − 2 HARQ transmissions.
The ITU specifies a packet size of 32 bytes to meet the latency
and reliability targets. With QPSK modulation and a coding
rate from MCS1 to MCS5, along with one OFDM symbol b: TOTAL LATENCY
overhead, the required number of PRBs is given in Table 37. UP latency was evaluated in [9] for a sequence of trans-
Here, the CRC is not considered and TBS is 32B. missions. It was found that DL and configured-grant UL

112308 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

TABLE 42. Summary of all KPI’s.

FIGURE 35. 7 OS UL data total reliability with 1 − 2 HARQ transmissions.

TABLE 39. Reliability observations.

TABLE 40. Bandwidth.

TABLE 41. Bandwidth observations.

transmissions with 7 OS and 30 kHz SCS are possible within

the latency bound of 1 ms, as shown in Table 38. Thus,
the ITU reliability of 10−5 error within 1 ms can be met.
Reliability observations are shown in Table 39. Based
on the assessment of Reliability following points can be
observed.
1) The cell-edge SINR for URLLC config. A is approx-
imately 1.98 dB (DL) and 0.81 dB (UL) for channel
model UMa A and 1.93 dB (DL) and 1.77 dB (UL) for
channel model UMa B.
2) The cell-edge SINR for URLLC config. B is approx-
imately 0.16 dB (DL) and 0.83 dB (UL) for channel 3) With 1 transmission using MCS1, the reliability target
model UMa A and −0.06 dB (DL) and 0.65 dB (UL) of 10−5 error can be met on the DL and the UL (with a
for channel model UMa B. configured grant).

VOLUME 9, 2021 112309

 A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

4) With MCS1 and a 7 OS mini-slot, 46 PRBs are required [9] A. P. K. Reddy, N. Kumar, S. S. A. Tirumalasetty, S. Srinivasan, and
for a 32B packet. J. V. B. James, ‘‘Latency analysis for IMT-2020 radio interface technology
evaluation,’’ in Proc. IEEE 3rd 5G World Forum (5GWF), Sep. 2020,
5) With 30 kHz SCS and 7 OS mini-slot, 1 transmission pp. 613–618.
can be made in FDD mode within 1 ms. [10] Combined Updates (NSA) From RAN4 #86Bis and RAN4 #87,
document TS 38.104, R4-1806932, 3GPP, Ericsson, Jun. 2018.
[Online]. Available: https://portal.3gpp.org/ngppapp/CreateTDoc.
K. BANDWIDTH aspx?mode=view&contributionUid=R4-1806932
Based on the (Section 5.3.2) [15] bandwidth evaluation and [11] Reply to LS on NR UE Category, document R1-1721732,
RAN1, 3GPP, Ericsson and Intel, Dec. 2017. [Online]. Available:
observations were tabulated in Tables 40-41. https://www.3gpp.org/ftp/TSG_RAN/WG1_RL1/TSGR1_91/Docs/R1-
1721732.zip
VI. CONCLUSION [12] General Aspects for User Equipment (UE) Radio Frequency
(RF) for NR, document TR 38.817-01, Version 16.1.0, 3GPP,
Key performance metrics corresponding to the evaluation of Sep. 2019. [Online]. Available: https://portal.3gpp.org/desktopmodules/
the 3GPP 5G NR IMT-2020 radio interface technology as Specifications/SpecificationDetails.aspx?specificationId=3359
well as their related findings are discussed in this paper. The [13] NR; User Equipment (UE) Radio Access Capabilities,
document TS 38.306, Version 16.1.0, 3GPP, Jul. 2020. [Online].
core criteria for IMT-2020 technological efficiency specifi- Available: https://portal.3gpp.org/desktopmodules/Specifications/
cations were fulfilled by the 5G NR technology, according to SpecificationDetails.aspx?specificationId=3193
our findings. However, certain small variations are found in [14] NR; Radio Resource Control (RRC); Protocol Specification,
document TS 38.331, Version 16.1.0, 3GPP, Jul. 2020. [Online].
few situations such as, peak spectral efficiency, peak data rate, Available: https://portal.3gpp.org/desktopmodules/Specifications/
user experienced data rate, and area traffic capacity, though SpecificationDetails.aspx?specificationId=3197
still meeting the requirements. This is most certainly due to [15] NR; Base Station (BS) Radio Transmission and Reception, document TS
38.104, Version 16.4.0, 3GPP, Jul. 2020. [Online]. Available:
such biases or the lack of adequate information in the pro- https://portal.3gpp.org/desktopmodules/Specifications/Specification
ponents self-evaluation report. We were able to recommend Details.aspx?specificationId=3202
that the 3GPP 5G NR technology be accepted as a valid IMT- [16] Summary on Discussion on IMT-2020 Evaluation for Peakdata
Rate and Peak Spectral Efficiency, document R1-1809934,
2020 technology based on these assessments. Summary of all 3GPP, Huawei and HiSilicon, Aug. 2018. [Online]. Available:
KPI’s with requirement met status is given in Table 42. https://www.3gpp.org/ftp/TSG_RAN/WG1_RL1/TSGR1_94/Docs/R1-
1809934.zip
[17] Way Forward on NR Peak Data Rate Formula, document R1-1805641,
ACKNOWLEDGMENT 3GPP, Intel, Samsung, MediaTek, Huawei, HiSilicon, Apple, Vivo,
The authors would like to express their gratitude to and OPPO, Apr. 2018. [Online]. Available: https://www.3gpp.org/
Mr. Vikram, Ms. Vertika, other office bearers of Cellular ftp/TSG_RAN/WG1_RL1/TSGR1_92b/Docs/R1-1805641.zip
[18] Study on Self Evaluation Towards IMT-2020 Submission,
Operators Association of India (COAI), and the industry document TR 37.910, Version 16.1.0, 3GPP, Oct. 2019. [Online].
mentors who supported them during the 5GIF IMT-2020 Available: https://portal.3gpp.org/desktopmodules/Specifications/
independent evaluation group activity. SpecificationDetails.aspx?specificationId=3190
[19] Department of Telecommunications, Ministry of Communication,
GoI (Government of India), Poicy. (2018). National Digital
REFERENCES Communications Policy 2018. [Online]. Available: https://dot.gov.
[1] International Telecommunication Union Radiocommunication Sector. in/sites/default/files/EnglishPolicy-NDCP.pdf
(Jul. 2020). IMT for 2020 and Beyond. [Online]. Available: https://www. [20] Evaluation of IMT-Advanced Candidate Technology Submissions in Doc-
itu.int/en/ITU-R/study-groups/rsg5/rwp5d/imt-2020/Pages/default.aspx uments IMT-ADV/4 and IMT-ADV/8 by TCOE India, document IMT-
[2] Minimum Requirements Related to Technical Performance for IMT- ADV/16, Telecom Centres of Excellence, Jul. 2010. [Online]. Available:
2020 Radio Interface(s), document M.2410, International Telecommuni- https://www.itu.int/md/R07-IMT.ADV-C-0016/en
cation Union Radiocommunication Sector, Nov. 2017. [Online]. Available:
https://www.itu.int/pub/R-REP-M.2410-2017
[3] Study on Scenarios and Requirements for Next Generation
Access Technologies, document TR 38.913, Version 16.0.0, 3GPP,
Jul. 2020. [Online]. Available: https://portal.3gpp.org/desktopmodules/
Specifications/SpecificationDetals.aspx?specificationId=2996
[4] A. K. Bachkaniwala, V. Dhanwani, S. S. Charan, D. Rawal, and
S. K. Devar, ‘‘IMT-2020 evaluation of EUHT radio interface technology,’’
in Proc. IEEE 3rd 5G World Forum (5GWF), Sep. 2020, pp. 631–636.
[5] V. Dhanwani, N. Kumar, A. K. Bachkaniwala, D. Rawal, and S. Kumar,
‘‘Assessment of candidate technology ETSI: DECT-2020 new radio,’’ in
Proc. IEEE 3rd 5G World Forum (5GWF), Sep. 2020, pp. 625–630. A. PHANI KUMAR REDDY (Graduate Student
[6] Guidelines for Evaluation of Radio Interface Technologies for IMT-2020, Member, IEEE) received the B.Tech. degree in
document M.2412, International Telecommunication Union Radiocommu- electronics and communication engineering from
nication Sector, Nov. 2017. [Online]. Available: https://www.itu.int/pub/R-
Yogi Vemana University (YVU), Kadapa, India,
REP-M.2412
in 2015. He is currently pursuing the integrated
[7] NR; Physical Channels and Modulation, document TS 38.211, Version
M.Tech. and Ph.D. degree with the Department of
16.2.0, 3GPP, Jul. 2020. [Online]. Available: https://portal.3gpp.org/
desktopmodules/Specifications/SpecificationDetails.aspx?specificationId Electrical Engineering, Indian Institute of Tech-
=3213 nology Kanpur (IITK), Kanpur, India. His research
[8] Framework and Overall Objectives of the Future Development of interests include wireless communications sys-
IMT for 2020 and Beyond, document M.2083, International Telecom- tems, signal processing, massive MIMO, and
munication Union Radiocommunication Sector, Sep. 2015. [Online]. 3GPP standardization. He received the University Gold Medal from YVU
Available: https://www.itu.int/dms_pubrec/itu-r/rec/m/R-REC-M.2083-0- for his outstanding academic performance while pursuing the B.Tech. degree
201509-I!!PDF-E.pdf in engineering.

112310 VOLUME 9, 2021

A. P. K. Reddy et al.: 5G New Radio Key Performance Indicators Evaluation

M. SHEEBA KUMARI (Senior Member, IEEE) K. VASUDEVAN (Senior Member, IEEE) received
received the B.E. degree in electronics and com- the Bachelor of Technology degree (Hons.)
munication engineering from the NMAM Insti- from the Department of Electronics and Electri-
tute of Technology, Mangalore, in 2000, and the cal Communication Engineering, IIT Kharagpur,
M.Tech. degree in digital electronics and commu- Kharagpur, India, in 1991, and the M.S. and
nication from Dayanand Sagar College of Engi- Ph.D. degrees from the Department of Elec-
neering, Bengaluru, in 2007. She is currently trical Engineering, IIT Madras, in 1996 and
pursuing the Ph.D. degree in millimeter wave 2000, respectively. From 1991 to 1992, he was
channel modeling for 5G networks. She has over with Indian Telephone Industries Ltd., Bengaluru,
14 years of academic experience and five years of India. He was a Postdoctoral Fellow at the Mobile
research experience. Her current research interests include 5G cellular com- Communications Laboratory, EPFL, Switzerland, from December 1999 to
munications, mmWave technology, wireless channel modeling, and machine December 2000, and an Engineer at Texas Instruments, Bengaluru, from
learning applications in communication systems. January 2001 to June 2001. Since July 2001, he has been a Faculty at the
Electrical Department, IIT Kanpur, where he is currently a Professor. His
research interests include the area of communications and signal processing.

SRINIVASAN SELVAGANAPATHY received the

M.S. degree in software systems from BITS Pilani.
He has overall 20 years of experience in cellular
technology with expertise spanning across product
VISHAKHA DHANWANI received the B.Tech. development, system engineering, and standard-
degree in electronics and communication engi- ization research. In 2009, he joined Nokia and has
neering from The LNM Institute of Information specialized in the design of cellular systems for
Technology (LNMIIT), Jaipur, Rajasthan, India, in machine type communications and the Internet of
2020. She is currently working as a 5G NR L1 Things. He is currently a Senior Standardization
DSP Firmware Engineer at Marvell India. Prior Specialist at Nokia. He has over the years actively
to joining Marvell, she worked on analysis of contributed to the standardization work related to massive MTC commu-
submissions by 3GPP NR and DECT-2020 NR nications, small-cells, multi-connectivity, and mobility enhancement topics
candidate technology for IMT-2020 with the COAI in 3GPP. He is also a RAN2 delegate with a focus on radio protocol aspects
as a member of 5G India Forum Independent Eval- of 5G and the IoT topics in 3GPP.
uation Group. Her research interests include antenna beamforming, massive
MIMO, L1 signal processing, and 3GPP standardization.
SENDIL KUMAR DEVAR received the Ph.D.
degree in electrical engineering from the Indian
Institute of Technology Madras, India. He has
over 12 years of working experience in wireless
standards. He currently leads the standardization
and spectrum at Ericsson, India. He actively con-
tributed in the IMT-2020 evaluation process and in
AKHIL KALPESH BACHKANIWALA received the development of rural test environment require-
the B.Tech. degree in communication and com- ments in IMT-2020. He also contributes to spec-
puter engineering from The LNM Institute trum studies for the World Radio Conference. Prior
of Information Technology (LNMIIT), Jaipur, to Ericsson, he worked with research organizations, including Samsung
Rajasthan, India, in 2020. He is currently working Research, Centre of Excellence of Wireless Technology, and Telecom Centre
as a Junior Design Engineer at BigCat Wireless of Excellence. He had contributed to standardization development in both
Pvt., Ltd., India. Prior to joining BigCat Wireless 3GPP2 and 3GPP for HRPD, HSDPA, LTE, and 5G.
Pvt., Ltd., he worked on the analysis of 3GPP
NR and EUHT candidate technology for IMT- PUNIT RATHOD (Member, IEEE) received the
2020 as a member of 5G India Forum Independent Ph.D. degree from the Indian Institute of Technol-
Evaluation Group. His current research interest includes wireless systems ogy Bombay, India. He has more than 15 years
engineering. of wireless research and development experience.
He has a diverse experience of working as an
academic researcher and a co-founder of a startup
with deployed equipment in public-safety agencies
in India and multinational companies. He has con-
tributed to national and international standardiza-
tion efforts at 3GPP, IEEE, and ITU-R. His recent
work has been focused on spectrum and standardization as a Lead-Technical
NAVIN KUMAR (Senior Member, IEEE) received Standards with Qualcomm Standards and Industry Organization.
the Ph.D. degree in telecommunication engineer-
ing from the University of Minho, Aveiro, and VINOSH BABU JAMES (Senior Member, IEEE)
Porto of Portugal, Europe, in 2011. He has over received the Ph.D. degree in electrical engineering
24 years of working experience in government, from the Indian Institute of Technology Madras,
IT, and academia. He is currently an Associate India. He has over 18 years of working experi-
Professor with the Electronics and Communica- ence in wireless systems engineering. He currently
tion Department, Amrita School of Engineering, leads the standards engagement at Qualcomm,
Amrita Vishwa Vidyapeetham, Bengaluru, India. India.
He has published over 90 peer-reviewed journals
and IEEE conference proceedings. His current research interests include 5G
mmWave, THz technology, and optical wireless communication.

VOLUME 9, 2021 112311