You are on page 1of 6

LTE IoT Link Budget and Coverage

Performance in Practical Deployments


István Z. Kovács1, Preben Mogensen1,5, Mads Lauridsen5, Thomas Jacobsen5, Krzysztof Bakowski2,
Poul Larsen3, Nitin Mangalvedhe4, Rapeepat Ratasuk4
1
Nokia Bell Labs, Aalborg, Denmark
2
Nokia, Wroclaw, Poland
3
Nokia, Copenhagen, Denmark
4
Nokia Bell Labs, Arlington Heights, USA
5
Wireless Communication Networks, Aalborg University, Aalborg, Denmark
{istvan.kovacs, preben.mogensen}@nokia-bell-labs.com

Abstract—This paper provides an updated analysis of The standardization in 3GPP, over several successive
link-budget and coverage performance for LTE IoT releases, has targeted progressively improved support for
technologies: enhancements for machine type low-power consumption IoT device connectivity. The
communications (eMTC) and narrowband Internet of design targets for enhanced support in recent releases
Things (NB-IoT). Previous studies have used the 3GPP
evaluations assumptions and have demonstrated the
have included achieving a low device-cost (below 5
coverage capabilities of these technologies when they are USD), limited uplink latency (less than 10 s), support for
independently deployed. Some operators are, however, a massive number of devices (40 per household), long
likely to deploy dual-mode networks - eMTC and NB-IoT - battery life (10 years), and enhanced coverage (20 dB
with the intent of supporting different applications in the better than General Packet Radio Service or GPRS) [4].
two systems. With all conditions being equal for the two Two technologies have been specified, namely
systems in such a scenario, comparison of the eMTC and enhancements for machine type communications (eMTC,
NB-IoT technologies then requires that all assumptions are also known as LTE-M) and narrowband IoT (NB-
aligned. This paper extends the previous studies with the IoT)[2][5]. Release 13 introduced a new user equipment
assumptions aligned for eMTC and NB-IoT systems to
provide a fair coverage performance comparison. The
(UE) category for each of these technologies. According
study shows that compared to the 3GPP assumption, under to 3GPP the Cat-M1 UE (for eMTC) supports a
the aligned assumptions, both eMTC and NB-IoT can Maximum Coupling Loss (MCL) of 156dB with 1.4MHz
support up to 5dB and 3dB higher uplink coverage, bandwidth; the Cat-NB1 UE (for NB-IoT) supports a
respectively. We further show these link budget MCL of 164dB in 200kHz bandwidth [4][5]. The
improvements yield significantly reduced LTE IoT capabilities for these UE categories have been further
coverage outage in two simulated scenarios based on real enhanced in the recently completed Release 14
network deployments: a wide area rural case and an urban standardization in terms of peak data rates and
deep indoor case. The results further demonstrate the transmission latencies [2]. Detailed analysis to
potential of these LTE IoT technologies in realistic network
deployment conditions.
demonstrate that the coverage target is achieved for all
the physical channels is provided in [6] for eMTC and in
[7] for in-band NB-IoT. Although these LTE IoT
I. INTRODUCTION technologies are designed to meet certain coverage and
The Internet of Things (IoT) comprises a connected capacity requirements based on 3GPP scenarios, it is
network of different types of physical objects with important to evaluate performance in realistic
capability to communicate over the Internet. The deployment scenarios. A coverage and capacity analysis
deployment of IoT devices has increased in recent years for NB-IoT in a rural area was investigated in [8]. This
and, according to one forecast, the number of such simulation study showed that NB-IoT provide extended
devices in 2020 will be three times as high as the global support for cellular IoT under the 3GPP assumptions.
population [1]. With two-thirds of all Internet traffic On the other hand, several chipset vendors have
predicted to involve wireless and mobile devices, the 3rd already developed a Cat-M1 modem that will support
Generation Partnership Project (3GPP) has taken the lead both Cat-M1 and Cat-NB1 through a software upgrade
[9][10]. The advantage of such a development is that the
in specifying air-interface standards for wide-area
UEs equipped with a dual-modem can be deployed in
wireless connectivity of Long Term Evolution (LTE) IoT different networks and hence it provides a lot of
devices [2]-[5]. flexibility for device manufacturers and possibly,
operators as well. Some operators are likely to deploy
dual-mode networks - eMTC and NB-IoT - with the
978-1-5386-3531-5/17/$31.00 ©2017 IEEE intent of supporting diverse IoT applications in their
networks. With all conditions being equal for the two TABLE I THE 3GPP VS. ALIGNED LINK-BUDGET ASSUMPTIONS
systems in such a scenario, comparison of the eMTC and 3GPP 3GPP Aligned
NB-IoT technologies requires that link budget eMTC NB-IoT eMTC & NB-IoT
assumptions are also aligned. UE NF (dB) 9 5 5
UE max TX Pw
This paper extends the previous studies with the 20 23 23
(dBm)
assumptions aligned for eMTC and NB-IoT systems to eNB NF (dB) 5 3 3
provide a fair coverage performance comparison. The eNB TX power 29 35 32 for eMTC
per 200KHz (w/o 3dB (w/ 6 dB 35 for NB-IoT
impact on coverage from aligning the assumptions for the (dBm) boosting) boosting) (w/ boosting)
two technologies is further investigated in two different DL & UL
practical scenarios: a wide area rural deployment, based interference 0 0 3
margin (dB)
on [8], and an urban deep indoor deployment, based on Max. DL TBS
[11]. Our study offers a direct comparison of eMTC and 680 680 680
(bits)
NB-IoT under these realistic deployment conditions. Max. UL TBS
680 1000 680
With competition among operators that are deploying (bits)
BLER target
these two LTE technologies, such a comparison becomes 10 10 10
(%)
especially relevant, allowing assessment of coverage modulation configurations [2] for a transport
performance in the relevant deployment scenarios. block size (TBS) of 680bits.
The remainder of this paper is organized as follows. • Signal-to-interference-plus-noise ratio (SINR)
The assumptions and methodology used in the simulation requirement. The common assumption for 10%
study are described in Section II. Simulation results are block error rate (BLER) at the first Hybrid
provided in Section III. The impact of higher layer system Automatic Repeat reQuest (HARQ) transmission
overheads and transmission overheads is discussed in is used.
Section IV. The paper is concluded in Section V. • UE number of antenna elements. One antenna
element was assumed for transmit and receive, the
II. ASSUMPTIONS AND METHODOLOGY same as in 3GPP evaluations.
A. Assumptions • eNB number of antenna elements. Two antenna
elements were assumed for transmit and receive,
To align with the earlier link budget studies for without diversity techniques, the same as in 3GPP
eMTC and NB-IoT reported in [6], we derive our results evaluations.
in terms of achievable physical layer throughput
The impact of higher layer system overheads and
(PHYTput) versus the coupling loss (CL) for the Physical transmission overheads are further discussed in Section
Downlink Shared Channel (PDSCH) and Physical IV.
Uplink Shared Channel (PUSCH) [2]. In our analysis of
the eMTC and NB-IoT link-budget performance we have B. Methodology
adopted the common assumptions summarized in Table
I, and commented below: The full set of transmission formats specified in 3GPP
• In-band carrier deployment. NB-IoT supports in- Release 13-14 [2] are used to search for the optimal PHY
band, guard-band and stand-alone carrier transmission configuration in terms of modulation and
deployment while eMTC can only be deployed in- coding schemes (MCS) and transmission format. We use
band. the link level performance curves in Figure 1 (downlink)
• UE receive noise factor. For dual-mode UE and Figure 2 (uplink) to estimate the required SNR values
implementation the same noise factor is assumed corresponding to a 10% BLER at the first HARQ
for both eMTC and NB-IoT. transmission. These required cell-edge SNR values are
• UE transmit power. The selected value represents then translated to MCL values based on the assumptions
the maximum possible under the assumption of listed in Table I and the methodology outlined in [3]. For
dual-mode UE implementation. In the 3GPP a fair comparison between eMTC and NB-IoT results, we
assumption a conservative 3dB back-off was have chosen to use only the common set of QPSK MCS
assumed, which is not needed in dual-mode UEs. [2]. This means that we only consider the Coverage
• Evolved NodeB (eNB) receive noise factor. For Extension Mode B (CEmodeB) operating mode for
dual-mode eNB implementation the same noise eMTC. The associated number of transmission
factor is assumed for both eMTC and NB-IoT. repetitions (up to 2048) was used for both eMTC and NB-
• eNB transmit power. The selected 46dBm total IoT. Further, the standard specifies the available
transmit power for a 10MHz carrier bandwidth is transmission formats in terms of number of Physical
a typical transmit power level in the 800-900MHz Resource Blocks vs. TBS index for eMTC; and in terms
frequency bands. of number of sub-frames (for downlink) or resource units
• Modulation and coding scheme (MCS) selection (for uplink) vs. TBS index for NB-IoT [2].
procedure. We used the set of 3GPP Release 14 In this study, the optimal downlink and uplink
transmission formats (including number of repetitions)
Fig. 3. Estimated signal penetration loss for a 800MHz carrier
Fig. 1. Downlink link-level BLER performance curves (analytical frequency at different indoor levels (-1 to -4), based on data from [11]
approximations) for the QPSK MCS with the target 10% BLER
highlighted
By dividing the area into 100m x 100m pixels and
identifying the locations of residential homes, the path
loss between all base stations and potential end users has
been simulated. A spatial resolution of 100m x 100m was
selected and estimated to provide sufficient accuracy due
to the low population density in the rural area considered.
In addition, shadow fading and indoor penetration loss of
10dB, 20dB, or 30dB has been added. The coverage
evaluation has been performed, for the are covered by 319
base stations, with average inter-site distance (ISD) =
2.8km, and for a spatially filtered version of the base
station locations to examine the effect of not deploying
NB-IoT and/or eMTC on all sites. The spatial filtering is
based on a minimum ISD of 4km, which results in a
Fig. 2. Uplink link-level BLER performance curves (analytical network of 170 base stations with an average ISD =
approximations) for the QPSK MCS with the target 10% BLER 6.2km.
highlighted
Based on the rural network scenario the average
coupling loss between every end user device and the
for each CL are selected for each technology such as to
serving base station is simulated, as described in [8]. The
achieve the maximum PHY throughput for a TBS of
serving base station is defined to have the strongest
680bits plus 32bits cyclic redundancy check overhead.
received power (lowest coupling loss). The effect of
For uplink transmission, the UEs are assumed to use the
inter-cell interference is considered in the added
maximum available total transmit power of 23dBm
interference margin of 3dB (see Table I). The outage
without any back-off margin.
probabilities are based on the uplink link budget results
We analyze the downlink and uplink link budget and
(see Section III.A), and are identified by determining the
radio coverage results for representative CL values. The
number of users whose coupling loss exceeds the MCL
selected values are the 3GPP specified MCL values:
for a given technology.
144dB for LTE, 156dB for eMTC, 164dB for NB-IoT;
the 170dB CL value is the maximum achievable D. Deep indoor scenario
downlink coverage with the aligned assumptions in Table The underground parking scenario investigated in
I. [11] can be considered as a typical use case for LTE IoT
C. Rural network scenario such as deployment of sensor devices or radio
connectivity for remote car services. While it is likely that
The rural network scenario covers approximately
network operators would deploy indoor small cells or
8000km2 in the North Jutland region of Denmark. The
indoor remote radio heads to provide at least voice
area is sparsely populated with 10 cities covering
connectivity in these scenarios, the LTE IoT coverage
150km2. The region has been implemented in a calibrated
could still be provided from an outdoor macro cell, which
radio propagation simulator, including a terrain profile
services a much larger geographical urban area.
and 319 sub-1GHz base stations and their configuration
We investigate this latter case and quantify the outage
of a local operator.
values achieved in the indoor locations. Figure 3 shows
TABLE II DOWNLINK COVERAGE RESULTS (PHY DATA RATES) B. Outage performance in rural network scenarios
WITH 3GPP VS. ALIGNED LINK-BUDGET ASSUMPTIONS
Figure 4 shows the outage probability (defined in
3GPP Aligned 3GPP Aligned
MCL eMTC eMTC NB-IoT NB-IoT Section II.C) as obtained for the rural network using the
144 dB 165 kbps 330 kbps 158 kbps 79 kbps original base station deployment, with the average ISD =
156 dB 10 kbps 21 kbps 10 kbps 5.0 kbps 2.8km. The results are given for both the 3GPP and
161 dB 3.4 kbps 5.0 kbps 2.4 kbps 1.2 kbps aligned versions of eMTC and NB-IoT performance
164 dB 1.7 kbps 3.4 kbps 1.2 kbps 0.8 kbps
167 dB 0.8 kbps 1.7 kbps 0.8 kbps 0.4 kbps assumptions, and include both outdoor and indoor users.
170 dB 0.3 kbps 0.8 kbps 0.4 kbps 0.2 kbps The coverage for outdoor and light indoor users, with
10dB penetration loss, is very good, having less than
TABLE III UPLINK COVERAGE RESULTS (PHY DATA RATES) 0.1% outage for both technologies, independent of the
WITH 3GPP VS. ALIGNED LINK-BUDGET ASSUMPTIONS
assumptions used. The deep indoor users, who have 30dB
3GPP Aligned 3GPP Aligned penetration loss, experience a significant outage
MCL eMTC eMTC NB-IoT NB-IoT reduction from 28% to 14% and from 8% to 4% for
144 dB 10 kbps 41 kbps 19 kbps 10 kbps
156 dB 0.6 kbps 0.8 kbps 1.2 kbps 0.8 kbps eMTC and NB-IoT, respectively, when the aligned
161 dB n/c 0.3 kbps 0.7 kbps 0.6 kbps assumptions are used instead of the earlier 3GPP target
164 dB n/c n/c 0.4 kbps 0.3 kbps coverage assumptions.
167 dB n/c n/c 0.2 kbps 0.2 kbps Figure 5 shows the outage probability for the same
170 dB n/c n/c n/c n/c
area, for the average ISD = 6.2km. The increased ISD,
compared to Figure 4, does not affect the outdoor users,
the measured signal losses at the different indoor
but the all indoor users would experience higher outage
levels; the arrows at level -1, -2, -3 and -4 indicate the
compared to the deployment ISD = 2.8km.
average estimated signal loss values. The large-scale
signal variations across measured locations at each indoor
level (-1 to -4) have been estimated as being log-normal
distributed and employed in the outage calculations. The
indoor outage probabilities are based on the uplink link
budget results (see Section III.A), and are identified by
determining the probability of indoor locations with
coupling loss exceeding the MCL for a given technology.

III. RESULTS

A. Link budget performance


The link-budget results for the downlink and uplink
shared channels are summarized in Table II and Table III,
respectively. We show results for the achievable MCL
values within the extended radio coverage range as
discussed in Section II.B. Fig. 4. Uplink outage probability results for a rural macro network
deployment [8] with average inter-side distance of 2.8 km
The first main observation is that the uplink coverage
is still the limiting factor, and under the aligned
assumptions, eMTC can support up to 161dB MCL in
uplink i.e., 5dB MCL improvement compared to the
3GPP target of 156dB. Secondly, compared to the case of
using the 3GPP assumptions, for the same coupling loss
values, the eMTC performance has improved in uplink
due to the lower noise factor values and higher transmit
power level; while the NB-IoT performance has
decreased due to the interference margin. Third, the
results for NB-IoT indicate that even devices with 167dB
coupling loss should be able to receive and transmit data.
In the next Section, for the outage analysis we
compare the achieved deployment outage levels, for each
technology, and between the case of using the target
3GPP MCL values and the newly identified uplink MCL
values in Table III.
Fig. 5. Uplink outage probability results for a rural macro network
deployment [8] with average inter-site distance of 6.2 km
The deep indoor users are heavily affected by the
reduced number of base stations, but likewise also
experience significantly better coverage under the aligned
assumptions. The outage is reduced from 47% to 30% for
eMTC, and from 20% to 11% for NB-IoT. However, an
outage above 10% may still be too high for some use
cases and therefore the scenario with the larger average
ISD = 6.2km might not be the optimal solution in terms
of cost vs. coverage for indoor users.

C. Outage performance in deep indoor scenario


The focus in this scenario is on the achievable indoor
coverage only, thus the results do not include any outdoor
user locations, where close to 100% coverage can be
achieved with both technologies.
Fig. 6. Uplink outage probability results for the urban macro cell
We show in Figure 6 the uplink outage probability coverage in deep-indoor underground locations
results for the analyzed deep indoor locations,
corresponding to the measurements in Figure 3. As hence the effective data rates usable at application level
expected, due to increasing signal penetration losses vs. become extremely low. This clearly limits the type of
indoor level, the outage increases significantly, and for applications services which can be provided over the LTE
both technologies. IoT radio links.
For the first two indoor levels (level -1 and -2), with Imperfect link adaptation i.e. the selection of a sub-
the 3GPP target coverage a maximum 10% outage is optimal transmission format and/or segmentation of the
possible for eMTC only down to level -1 and with NB- payload packets at MAC level, can result in additional re-
IoT only down to level -2. In contrast, with the aligned transmissions or simply failed transmissions, which
assumptions, both eMTC and NB-IoT can provide should also be accounted for in the transmission latency
coverage down to indoor level -2, and the outage levels and the service outage performance evaluation.
are significantly reduced to below 3%. The outage in The energy consumption (battery life time) of the
deeper indoor locations, on level -3 and level -4, remains LTE IoT devices is a critical aspect, often neglected in
very high for any practical deployment purposes. This is typical network deployment studies. The 3GPP
expected, as the average coupling loss in these locations evaluations have demonstrated battery life times of at
is above 165dB. least 10 years, under certain traffic model assumptions.
However, in real network deployment scenarios such as
IV. DISCUSSIONS presented in this work, all the other system aspects
Our analysis considers many of the practical system mentioned before could result in a much faster battery
aspects, as listed in Section II.A. Nevertheless, further discharge time.
considerations are needed to assess the LTE IoT service
V. CONCLUSIONS
performance.
The LTE physical downlink/uplink control channels In this paper, we have analyzed the link-budget and
are designed to be quite robust (due to redundancy) and coverage performance for the two 3GPP LTE IoT
require low data rates. Nevertheless, any failure in the technologies: eMTC and NB-IoT introduced in Release
control signaling results in loss of radio connectivity and 13. Previous studies have used the 3GPP evaluations
no data transmission is possible. The channel estimation assumptions and have demonstrated the coverage
errors, especially at very low SNR, can have also capabilities of these technologies when they are
significant impact because they affect both the physical independently deployed. Some operators are, however,
control and data channels performance. This aspect is likely to deploy dual-mode networks – eMTC and NB-
particularly important for the low-complexity user IoT – with the intent of supporting different applications
devices. in the two systems. A comparison of the eMTC and NB-
The data transmission latency includes delays and IoT technologies then requires that all assumptions be
processing time on the physical (PHY) layer and the aligned.
medium access control (MAC) layers. The data This paper extends the previous studies with the
transmission latency also includes the radio resource revised assumptions, aligned between eMTC and NB-IoT
control (RRC) signaling transmissions. The very low systems, to provide a fair coverage performance
PHY data rates achievable at large coupling loss values comparison.
(see Table II and Table III), imply extended transmission The first main observation is that the uplink coverage
times for a fixed payload size. This is further augmented is still the limiting factor. Nevertheless, under the aligned
by the need for transmission repetitions (up to 2048), assumptions, eMTC can support up to 161dB MCL in
uplink, compared to the 156dB 3GPP target; and NB-IoT [11] H.C. Nguyen, L.C. Gimenez, I. Z. Kovács, I. Rodriguez, T. B.
Sorensen, and P. Mogensen, “A Simple Statistical Signal Loss
uplink coverage is extended to 167dB MCL, compared to Model for Deep Underground Garage, ” IEEE Proc. Veh. Tech.
the 164dB 3GPP target. Conf., Fall 2016.
The impact on coverage performance from aligning
the assumptions for the two technologies is investigated
in two simulated scenarios based on real network
deployments: a wide-area rural deployment and an urban
deep indoor deployment. The rural outage probability,
including indoor locations, is significantly improved
when the aligned assumptions are used. In deep indoor
locations, the coverage improvements are visible on the
upper indoor levels only.
These results demonstrate the potential of the LTE
IoT technologies when dual-mode networks and/or dual-
mode UE are assumed in realistic network deployment
conditions.
Future work needs to model and evaluate the
remaining system aspects discussed in Section IV.
Finally, extensive comparison of simulation results with
in-field measurement using live NB-IoT and/or eMTC
deployments is required to fully validate the modeling
assumptions.
REFERENCES
[1] Cisco, “Cisco Visual networking index: Global mobile data
traffic forecast update, 2015-2020,” White paper, 2016.
[2] 3GPP TS 36.213, “Technical Specification Group Radio Access
Network; Evolved Universal Terrestrial Radio Access (E-
UTRA); Physical layer procedures (Release 14)”, V14.2.0,
March 2017.
[3] 3GPP TR 45.820, “Technical Specification Group GSM/EDGE
radio Access Network; Cellular System Support for Ultra Low
Complexity and Low Throughput Internet of Things; (Release
13)”, V13.1.0, November 2015.
[4] N. Mangalvedhe, R. Ratasuk, and A. Ghosh, “NB-IoT
deployment study for low power wide area cellular IoT,” Proc.
IEEE 27th Annual International Symposium on Personal, Indoor,
and Mobile Radio Communications (PIMRC), September 2016.
[5] A. Rico-Alvariño et al., “An Overview of 3GPP Enhancements
on Machine to Machine Communications,” IEEE Comm.. Mag.,
Vol. 54, Issue 6, pp. 14–21, June 2016.
[6] Sierra Wireless et al., “Coverage Analysis of LTE-M Category-
M1 (White Paper),” Version 1.0, January 2017. Accessed May
19, 2017:http://hub.sierrawireless.com/coverage_analysis_lte_m
[7] R. Ratasuk, N. Mangalvedhe, Y. Zhang, M. Robert, and J.-P.
Koskinen, “Overview of narrowband IoT in LTE Rel-13,” Proc.
IEEE Conf. Standards for Commun. and Networking (CSCN),
October 2016.
[8] M. Lauridsen, H.C. Nguyen, B. Vejlgaard, I.Z. Kovács, P.
Mogensen, and M. Sørensen, ”Coverage comparison of GPRS,
NB-IoT, LoRa, and SigFox in a 7800 km2 area”, IEEE Proc. Veh.
Tech. Conf. Spring 2017.
[9] J. Atkinson, “Qualcomm Clinches Module Design Wins for its
LTE Cat M1 and NB-1 Modem,” Wireless Magazine, November
2016. Accessed May 19, 2017: http://www.wireless-
mag.com/News/43416/qualcomm-clinches-module-design-
wins-for-its-lte-cat-m1-and-nb-1-modem.aspx
[10] Sequans Communications, “Sequans Introduces Monarch SX:
LTE-M/NB-IoT System-on-Chip for Integrated IoT Devices”,
February 2017. Accessed May 19, 2017:
http://www.sequans.com/press-release/sequans-introduces-
monarch-sx-lte-mnb-iot-system-on-chip-for-integrated-iot-
devices/

You might also like