Interference Analysis of the Coexistence of 5G

Cellular Networks with Satellite Earth Stations in

3.7-4.2GHz
Ghaith Hattab, Prakash Moorut, Eugene Visotsky, Mark Cudak, and Amitava Ghosh
Nokia Bell Labs, Naperville, IL, USA
ghattab@ucla.edu, {prakash.moorut, eugene.visotsky, mark.cudak, amitava.ghosh}@nokia.com

Abstract—The explosive growth of mobile traffic has elevated adjacent 3.55-3.7 GHz band (3.5GHz band) which has
the need to explore additional spectrum for fifth-generation (5G) been recently allocated for wireless broadband applica-
networks in low bands (e.g., below 3GHz), mid bands (e.g., 3- tions by the FCC [6]. In addition, unlike the incumbents
6GHz), and high bands (mainly above 24GHz). To this end,
it is imperative to study the interference of 5G systems into in the 3.5GHz band, which are primarily for Federal
incumbent systems occupying these bands. In this paper, we study Government uses, there is greater opportunity in the 3.7-
the coexistence of 5G Massive MIMO systems with incumbents 4.2 GHz band to minimize encumbrances, thus creating
in the mid-band spectrum between 3.7GHz to 4.2GHz, which are a cleaner interference environment for 5G terrestrial
primarily fixed receive-only satellite earth stations (FESs) that are deployments.
endowed with highly directive antennas. In particular, we analyze
• Large contiguous spectrum: As stated earlier, the 3.7-
the aggregate interference experienced at FESs from the uplink
and downlink transmissions of the 5G system. We then identify 4.2GHz is adjacent to the 3.5 band, and thus combining
and present several interference mitigation techniques, including the two helps provide 650MHz of contiguous spectrum.
protection regions, power control, and frequency partitioning, • Global harmonization: The 3.7-4.2GHz is also being
to ensure harmonious coexistence. Finally, we present a case considered for future 5G deployment in other countries
study of 5G deployment in a dense area, where we use an actual
database of FESs for accurate analysis. Simulation results reveal and regions, and it has the potential to become a glob-
that the favorable propagation environment sub-6GHz limits ally harmonized range. Clearly, spectrum harmonization
the opportunity of co-channel deployment, making frequency enables economies of scale, expands global roaming, and
partitioning more desired for a practical coexistence. reduces equipment design complexity.
Index Terms—5G, coexistence, interference analysis, massive
Due to the high potentials the 3.7-4.2GHz band can bring to
MIMO, spectrum sharing, satellite earth stations.
cellular networks, in this paper we study the deployment of
I. I NTRODUCTION 5G in the mid-band spectrum, and its coexistence with current
legacy systems that operate in the band.
It is universally accepted that the deployment of fifth- Existing work pertaining shared spectrum access between
generation cellular networks requires expanding spectrum ac- 5G systems and incumbent systems has been attracting signif-
cess beyond the traditional licensed spectrum bands allocated icant attention. For instance, 5G deployment over millimeter
for mobile cellular networks. Hence, it is critical to identify frequencies has spurred several studies such as the coexistence
current underutilized channels in the low-, mid- and high- of 5G systems with earth stations transmitting to satellite
bands spectrum to provide coverage, gigabit connectivity, stations at 28GHz [7], with fixed service at 39GHz [8], and
and accommodate the explosive growth of mobile traffic [1]. with point-to-point fixed service for wireless backhaul at 70
While spectrum access at millimeter frequencies has attracted and 80GHz [9], [10]. For coexistence sub-6 GHz, the FCC has
significant attention [2]–[4], e.g., the Federal Communications developed a sharing framework at 3.55-3.7GHz with radar,
Commission (FCC) has opened 3.85GHz of licensed spectrum fixed satellite service earth stations (FESs), etc [6]. Other
for cellular services at 28GHz and 39GHz [4], 5G access works, e.g., [11], [12], have studied the interference from
sub-6GHz remains a key piece of next-generation networks. incumbent radar systems into cellular networks at 3.5 GHz.
Indeed, the FCC has recently opened an inquiry into potential In this paper, we focus on 5G deployment over the C-band,
wireless broadband access in the mid-band spectrum sub-6 i.e., 3-7-4.2 GHz, where the incumbent systems are primarily
GHz, and specifically at 3.7-4.2 GHz [5]. We also note that FESs that operate in a receive-only mode, i.e., downlink from
other countries and regions, e.g., China, Europe, Japan, Korea, the satellite to the FES.
etc., have shown interest for 5G deployments in the mid-band We have thoroughly analyzed the FESs that have been
spectrum. registered in the FCC’s International Bureau Filing System
The mid-band spectrum uniquely suits the deployment of (IBFS) database. Our analysis has reveled that both the number
cellular networks for the following reasons: of existing FESs and the number of grants obtained to install
• Favorable propagation characteristics: The 3.7-4.2 new ones have been constantly declining over time. Indeed,
GHz band has similar propagation characteristics as the analyzing the incumbents’ database [13], Fig. 1a shows that,

978-1-5386-4328-0/18/$31.00 ©2018 IEEE

 105 II. S YSTEM M ODEL

Number of stations (all types)

104 We consider a grid deployment of 5G base stations, hence-
forth denoted as gNBs. The gNBs are deployed on street
103
corners at height hg and tilt φstrg , with an inter-site distance
102 of dISD . Each gNB has a cross-polarized large-scale planar
101
antenna array of size Nh × Nv × 2. The gNB’s site is further
1970 1980 1990 2000 2010 2020 2030 2040
Expiry year divided into four sectors, where in each sector, the gNB uses
(a) a total of 2Nh × 2Nv beams for user association and data
2500 communication.
2000 For users, henceforth denoted as UEs, we assume they
Number of grants

1500
are randomly distributed in space, with 20% of them being
1000
deployed outdoors and the rest are indoors [14]. We assume
a fixed antenna height of 1.5m for outdoor users. For a user
500
located indoors, we assume it is uniformly distributed across
0
1975 1980 1985 1990 1995
Year
2000 2005 2010 2015 2020 the floors of the building, where each floor is assumed to have
a height of 3m. All UEs are assumed to operate in an omni-
(b)
directional mode.
Fig. 1: The utilization of the 3.7-4.2GHz spectrum is decreas- During cell selection and association, the UE measures the
ing: (a) The number of FESs for a given expiry year; (b) The received power of reference signals sent over different beams
number of new grants per year. from gNBs in vicinity of the UE. Then, the UE connects
to the beam with the highest received power. We further
assume a static time-division duplexing (TDD) as means of
for a given expiry date, the number of FESs in the United resource allocation. Hence, all gNB-UE links point in the same
States has been constantly decreasing, reaching few thousands direction at a given time, i.e., either all are UL or all are DL.
in 2017. Similarly, Fig. 1b highlights that the number of new For the incumbents in the mid-spectrum band at 3.7-
grants for earth stations is also decreasing. Such decrease is 4.2GHz, we only focus on FESs as the density of other
believed to be due licensees using alternative transmission incumbents is very low, e.g., there are approximately 100
options such as utilizing the higher frequency bands, e.g., point-to-point wireless backhaul nodes in the United states
Ku-band, or trending away from wireless altogether onto whereas the number of FESs is roughly 4,800 [5]. We use the
fiber technologies. For instance, licensees find that the Ku- actual database of these FESs to get their location, antenna
band provides more bandwidth to support high-resolution height, and antenna gain. The antenna dish can have different
content distribution, and the dishes are cheaper due to their tilt angles, depending on the location of the satellite at a given
smaller antenna size. This motivates the utilization of the mid- time. However, per FCC regulations [15], the FES must adhere
spectrum for current and next-generation cellular networks. to a certain elevation range. Thus, the data entry of each FES
Our contributions in this work are twofold. First, we present specifies the elevation angle to the eastern most and western
a framework for interference analysis, and particularly the most geostationary satellite orbital arc limits. In this paper, we
aggregate interference from 5G systems into FESs, where the pick the minimum of these limits as the worst case scenario.
former operates in both the uplink (UL) and the downlink
(DL). The aggregate interference is simulated using realistic III. I NTERFERENCE A NALYSIS F RAMEWORK
scenarios, where the FESs’ deployment and parameters are We study the aggregate interference of the 5G system into
extracted from the IBFS database [13], and building databases FESs when the former is either in the UL mode or the DL
are used for accurate link-budget analysis. Second, we study mode. Since gNBs use massive MIMO systems and FESs use
several passive and practical interference mitigation tech- antenna dishes with high directivity and antenna gain, it is
niques, including protection regions in the spatial and angular critical to consider the attenuation due to the misalignment
domains, uplink power control, and frequency partitioning. A of their respective antenna directions. To this end, the aggre-
case study is presented, where the 5G system is deployed in gate interference mainly depends on: (i) the propagation loss
a dense urban area. Simulation results show that the favorable between the interferer and the victim FES receiver, (ii) the
propagation environment sub-6GHz can be detrimental from effective radiated power from the interferer towards the FES,
a coexistence perspective, and thus frequency partitioning, and (iii) the attenuation in azimuth and elevation due to the
or clearing, is recommended to ensure that FESs are well antenna pattern of the FES antenna dish.
protected from any harmful interference.
The rest of the paper is organized as follows. The system A. Path Loss between the interferer and the FES
model is presented in Section II, while the interference anal- We consider a modified 3GPP path loss model, emulating
ysis is given in Section III. The mitigation techniques are a street canyon, i.e., we use a modified version of the 3GPP
discussed in IV and the simulation results are presented in 3D-UMi channel model [16]. Mathematically, let xi be the
Section V. Finally, the conclusions are drawn in Section VI. (x, y)-coordinates of the interferer, which can be a gNB or a
UE, and hi be its antenna height. Similarly, let xf and hf be
Side view Top view
the coordinates and height of the FES, respectively. Then, the
path loss between the interferer and the FES is expressed as
FES
LOS
PLi→f = 1(β=0) PLLOS (xi , xf , hi , hf ) BLDG Interferer

+ 1(β=1) PLNLOS (xi , xf , hi , hf ) (1)

+ X (β) + PLin (din ) + PLi2o , FES
LOS BLDG
݄୆୐
݄෨
݀୧՜୆୐ ݄୧
where PLLOS is the line-of-sight (LOS) path loss, PLNLOS Interferer
||‫ܠ‬୧ െ ‫ ܠ‬୤ ||
is the non-LOS (NLOS) path loss, X is log-normal shadow-
ing, PLin (din ) is the indoor path loss when the interferer
FES
is located indoors at distance din from the nearest wall, NLOS
PLi2o is the indoor-to-outdoor penetration loss, β ∈ {0, 1} BLDG Interferer
is a binary variable that indicates whether the interferer-FES
link is LOS or NLOS, and 1(·) is an indicator function.
The LOSp and NLOS path losses depend on the distance Fig. 2: Computing whether the interferer-FES link is LOS or
ri→f = kxi − xf k2 + (hf − hi )2 , and they are essentially NLOS.
a multi-slope path loss with two different path loss exponents,
depending on whether the distance is less or larger than a
cut-off distance [16]. The standard deviation of log-normal attenuation patterns, respectively, and Ai,FTBR is the front-to-
shadowing depends on whether the link is LOS or NLOS. back ratio (FTBR) loss. For the element pattern we consider
The penetration loss is assumed to be PLi2o = 20dB [16], the following parabola [16]
whereas the indoor path loss is set to PLin (din ) = 0.5 din dB 
θ
2 
φ
2
az el
[16]. Ai,EP (θ) = 12 , Ai,EP (φ) = 12 , (4)
θ3dB φ3dB
Our model differs from the 3GPP channel model in com-
puting whether the link is LOS or NLOS. Specifically, in where θ3dB and φ3dB are the 3dB beamwidth in azimuth and
the 3GPP model, this is determined probabilistically using a elevation, respectively. The beam patterns are based on a rect-
non-increasing function with distance. In this paper, however, angular array with equally spaced antennas in both dimensions.
we use an actual building database to determine whether the The beam attenuation depends on the azimuth off-axis angles
off str
link is LOS or NLOS. This is done as follows. Assuming θi→f and θi→f which can be computed, respectively, as follows
the xy-plan represents the ground, we check first whether the  
connecting line between the interferer and the FES, i.e., the
off
θi→f = cos−1 (ui→f )T ubeam
i , (5)
line between xi and xf , intersects with the 2D polygon that
and
defines the building’s boundaries. If the line does intersect  
str
θi→f = cos−1 (ui→f )T ustr
i , (6)
with the polygon, we then check whether the polygon, of
height hBL , intersects the connecting line in 3D, i.e., the line where ui→f is a unit vector in the direction of the interference
connecting (xi , hi ) and (xf , hf ). Thus, the link is NLOS if axis, ubeam is a unit vector in the azimuth direction of the
i
h̃ + hi ≤ hBL , where interfering beam, and ustr is a unit vector of the azimuth
i
 
hf − hi
 antenna orientation. Similarly, the elevation off-axis angles
−1
h̃ = di→BL × tan tan , (2) φoff str
i→f and φi→f can be computed as
kxi − xf k
 
−1 hf − hi
where di→BL is the 2D distance between the interferer and the φoff
i→f = φ beam
i + tan , (7)
kxi − xf k
polygon. This is visualized in Fig. 2.
and  
−1 hf − hi
B. Radiated power from interferer towards the FES φstr str
i→f = φi + tan , (8)
kxi − xf k
If the interferer uses directional beams, e.g., the gNB in the
DL, then the antenna and beam alignment between the gNB where φbeam
i is the angle of the unit vector, vibeam , in the
and the FES must be taken into consideration. To this end, the elevation direction of the beam and φstr
i is the antenna tilt. The
total radiated power can be computed as off-axis angles are illustrated in Fig. 3a. Finally, we remark
that for omni-directional transmission, the total radiated power
Ei→f = Gi,max + Pi − Aaz off el off
i,BP (θi→f ) − Ai,BP (φi→f ) reduces to Ei→f = Pi .
 az str
(3)
− min Ai,EP (θi→f ) + Ael str
i,EP (φi→f ), Ai,FTBR , C. Attenuation due to FES Antenna Pattern
where Gi,max is the maximum array gain, Pi is the transmitted While the antenna dish has a very high gain, it also has a
power, Ai,BP (·) and Ai,EP (·) are the beam and antenna narrowbeam, and thus any misalignment can significantly help
 FES
FES
୭୤୤
‫ܝ‬୧՜୤ ߶୧՜୤
ୱ୲୰
ߠ୧՜୤ ‫ܝ‬ୱ୲୰ interference
୧
‫ܝ‬ୠୣୟ୫ ݄୤ ୱ୲୰
axis ߶୧՜୤
௜
݄୧
Interferer ୭୤୤
ߠ୧՜୤ ‫ܞ‬୧ୱ୲୰ ‫ܝ‬୧՜୤
‫ܞ‬୧ୠୣୟ୫
||‫ܠ‬୧ െ ‫ ܠ‬୤ ||
‫ܝ‬ୠୣୟ୫
௜
Azimuth Elevation

(a)
Interferer
‫ܝ‬୤
୭୤୤
ߠ୤՜୧ FES ୭୤୤
߶୤ୱ୲୰ ߶୤՜୧
‫ܝ‬୤՜୧

interference
݄୤ axis
݄୧
Interferer

||‫ܠ‬୧ െ ‫ ܠ‬୤ || Fig. 5: Beams that are aligned in the direction of FESs are
Azimuth Elevation switched off.
(b)
Fig. 3: Off-axis azimuth and elevation angles: (a) From inter- IV. PASSIVE I NTERFERENCE M ITIGATION
ferer to FES; (b) From FES to interferer.
In this paper, we focus on passive interference mitigation
techniques as they do not require coordination with incumbent
50

40
systems. Specifically, we study: (i) angular domain exclusion
ITU-R S.465-6
zones, (ii) uplink power control, and (iii) frequency partition-
Antenna Gain (dBi

30 FCC regulations
20
ing via guard bands.
10

-10 A. Angular-domain exclusion zones

-20
0 10 20 30 40 50 60
Off-axis elevation/azimuth angles (degrees)
70 80 90 The interference seen can become high if the interferer’s
used beam is in the direction of the FES, (cf. (3)). To this
Fig. 4: Comparison between the ITU-R recommended pattern end, the BS can shut the aligned beams in the direction of
and FCC regulations. FESs to protect them from interference. More formally, we
propose to switch the l-th interferer beam off if
   
attenuate the interfering signals. The antenna pattern can be bi,l = 1 cos−1 (ui→f )T ubeam ≤ψ , (12)
i
expressed as
off
where ψ is a angular protection threshold. Note that the same
) + Af (φoff

Gf→i = Gf,max − min Af (θf→i f→i ), Af,FTBR , principle can be applied to sectors instead of beams, where
(9) the sector’s boresight is used instead of the direction of the
where Af,FTBR is the FTBR loss, and Af (·) is the antenna beam. We emphasize that the unit vector ui→f is known in
attenuation, which is symmetric in both azimuth and elevation. advance using the database, and hence no prior coordination
In this work, we consider the ITU-R antenna pattern recom- is needed with the FES. This is illustrated in Fig. 5.
mendation, which is defined for receiving FESs [17]. Note
that the pattern used adheres to the FCC regulations [15], as B. Uplink power control
shown in Fig. 4. The azimuth and off-axis angles with respect
The angular-domain exclusion zone is applicable when
to the interferer are computed similar to the off-axis angles
the interferer uses directional beams. For an omni-directional
with respect to the interferer. Fig. 3b shows the relevant off-
mode, one can use power control in vicinity of FESs. To this
axis angles.
end, one simple algorithm is to set a target signal-to-noise ratio
(SNR) in regions in vicinity of the FES. The BS for instance
D. Aggregate Interference
can broadcast, during cell selection, reselection, or handover,
The interference generated from one interferer is expressed, the maximum allowable target SNR. The UE then estimates
in dBm, as the received signal and transmits at a power level that would
meet the target SNR. Specifically, the transmit power is
Ii→f = Ei→f +Gf→i − PLi→f , (10)
Pu = min{P̄γ , Pu,max }, (13)
and thus the aggregate interference, in dBm, is
! where P̄γ is set such that a target UL SNR is, on average,
agg.
X
Ii→f /10 equal to γ. Alternatively, the UE can deploy a fractional power
Ii→f = 10 log10 10 . (11)
i
control algorithm, where the power control factor (PCF) is set
by the BS. To this end, let ζ denote the linear-scale path loss 106

of the downlink, then, the UE can transmit at power

4.638
PuPCF = min{P̄o , (ζ) , Pu,max }, (14)
FES
4.6375
where  ∈ [0, 1] is the PCF, and P̄o is an open-loop transmit

y-coord
power. 4.637

C. Guard bands UE
4.6365

A popular interference mitigation technique is frequency gNB

partitioning, where different systems use channels that are 4.636

separated by guard bands. While spectrum sharing and co- 4.474 4.476 4.478 4.48 4.482
x-coord
4.484 4.486 4.488
105

channel deployment have become integral to cellular networks,

frequency partitioning could be preferred towards the coexis- (a)
tence of multiple systems, particularly since FESs do not use
the entire band, i.e., the spectrum is underutilized. Indeed, a
random sampling of the amount of spectrum used by approx-
imately a thousand FESs has been done by the Broadband FESs
Access Coalition. The sampling has shown that only 23MHz
of the 500MHz spectrum has been used at each earth station
[18], potentially showing that frequency partitioning can be an gNB
attractive solution for 5G deployments. Assuming that the 5G
system enforces a spectrum mask similar to the one used for
the Citizens Broadband Radio Service (CBRS) at 3.5GHz, then (b)
the total radiated power, in dBm/1MHz, for a give frequency
Fig. 6: The Chicago Loop area: (a) Layout of 8 FESs; (b)
separation from the 5G channel edge, f (MHz), is
 Example of FESs on buildings.
 −13, 0 ≤ f ≤ 10
Ei→f = −25, 10 < f ≤ 20 (15)
−40, 20 ≤ f

can significantly reduce the interference. However, the spatial
V. S IMULATION R ESULTS zone needs to have a large radius, making it a limiting factor
We study the aggregate UL and DL interference on FESs for 5G deployment over dense urban areas. Nevertheless, such
deployed in Chicago Loop, as shown in Fig. 6. The gNB is approach may be proven effective for suburban and rural areas.
assumed to be at height hg = 6m with an array of size 8×4×2 To reduce the protection radius in dense areas, an angular
that has an element pattern beamwidth of θ3dB = φ3dB = 65◦ , protection angle can be used in combination with the spatial
downtilt of φstr ◦ zone. For example, the median aggregate interference for
g = −6 , and Ag,FTBR = 30dB.The transmit
powers are assumed to be Pg = 46dBm/100MHz and Pu = a protection radius of 500m reduces from −155.6dBm/Hz
33dBm/100MHz. We consider a 25% instantaneous load in (−163.7dBm/Hz) to −175.6dBm/Hz (−180.8dBm/Hz) when
the available UL slots and 25% and 50% in the DL slots. The we combine it with a protection angle of 50◦ for 50% (25%)
carrier frequency is assumed to be 3.95GHz. For the FESs, we load.
extract their parameters from the database, and assume that For power control in the UL, the median aggregate interfer-
the noise temperature is computed at 100K. The subsequent ence is −183.9dBm/Hz when the target UL SNR is γ = 5dB.
results are averaged out over 100 spatial realizations of UEs. Similarly, for fractional power control, setting  = 0.8 makes
Fig. 7 shows the cumulative distribution function (CDF) of the median aggregate interference to be −183dBm/Hz and
the individual interference generated by UEs and gNBs into the median UL SNR to be 4.6dB. Reducing the aggregate
the FESs. It is clear that the DL incurs higher interference interference further is still achievable via power control at the
due to the higher transmit power and beamforming gains of expense of affecting the 5G system performance.
the gNB. Similarly, increasing the load from 25% to 50% Since co-channel operation of 5G and FESs in close prox-
increases the interference, without significant changes in the imity could be problematic, we study the case where the two
right-tail distribution. It is observed that, overall, the majority systems are not using the same spectrum blocks. Our results,
of interferers do not individually incur significant interference. shown in Fig. 9, illustrate that guard bands can ensure that
Nevertheless, the aggregate interference is dominated by few the aggregate interference level at FES is well below the noise
UEs and few gNBs, which can cause interference above the floor in the DL and the UL. For this reason, while the band is
noise floor. being cleared, the 5G systems could share the spectrum with
Fig. 8 shows the DL median aggregate interference in the FESs via frequency coordination during a transitional sharing
presence of two types of protection zones: one over space and period, particularly because the FES system does not use the
one over the angular domain. It is evident that exclusion zones entire 500MHz available [18].
 1
Individual Downlink Interference a balanced network coverage and capacity. To this end, it is
0.9 critical to study the coexistence of 5G systems with existing
0.8
incumbent FESs that operate in the mid-band. In this paper,
0.7
25% load we have analyzed the aggregate interference on FESs from a
0.6 ISD=200m: 50%
CDF 0.5
ISD=200m: 25%
ISD=400m: 50%
5G system operating in the UL and the DL. We have relied on
0.4
ISD=400m: 25%
ISD=400m: 25% a realistic deployment of the 5G and incumbent systems. We
ISD=600m: 25%
0.3 have also presented several passive mitigation techniques that
50% load
0.2
do not require coordination between both systems. We then
0.1 Noise floor
presented a case study on the interference seen at FESs in a
0
-250 -240 -230 -220 -210 -200 -190 -180 -170 -160 -150 -140
Interference per gNB (dBm/Hz) dense urban area. Our simulation results have demonstrated
that co-channel deployment may become problematic when
(a)
Individual Uplink Interference
FESs are in close proximity with the 5G system. Thus, we
1
have studied the case where frequency partitioning or clearing
0.9

0.8
is used, i.e., the 5G and incumbent systems use different
0.7 spectrum blocks, particularly because FESs do not occupy the
0.6 entire band. It is shown that guard bands ensure harmonious
CDF

0.5
coexistence in the UL and the DL.
0.4 ISD=200m
ISD=400m
0.3 ISD=600m R EFERENCES
0.2

0.1
[1] J. G. Andrews, S. Buzzi, W. Choi et al., “What will 5G be?” IEEE J.
Noise floor
Sel. Areas Commun., vol. 32, no. 6, pp. 1065–1082, Jun. 2014.
0
-250 -240 -230 -220 -210 -200 -190 -180 -170 -160 -150 -140 [2] M. Xiao, S. Mumtaz, Y. Huang et al., “Millimeter wave communications
Interference per UE (dBm/Hz)
for future mobile networks,” IEEE J. Sel. Areas Commun., vol. 35, no. 9,
(b) pp. 1909–1935, Sep. 2017.
[3] “IMT vision- framework and overall objectives of the future develop-
Fig. 7: CDF of individual interference: (a) DL; (b) UL. ment of IMT for 2020 and beyond,” ITU-R, M.2083-0, Sep. 2015.
[4] “Use of spectrum bands above 24 GHz for mobile radio services,”
Federal Communications Commission (FCC), Tech. Rep. GN Docket
Spatial Exclusion Zones Angular Exclusion Zones
-140 -140 No. 14-177, July 2016.
-145
50%
25%
[5] FCC, “Expanding Flexible Use in Mid-Band SpectrumBetween 3.7 and
-145
24 GHz,” Aug. 2017.
Aggregate Interference (dBm/Hz)

Aggregate Interference (dBm/Hz)

-150 -150 [6] ——, “Amendment of the Commission’s Rules with Regard to Com-
-155 -155
mercial Operations in the 3550-3650 MHz Band,” Apr. 2015.
[7] S. Kim, E. Visotsky, P. Moorut et al., “Coexistence of 5G with the
-160 -160
incumbents in the 28 and 70 GHz bands,” IEEE J. Sel. Areas Commun.,
-165 -165 vol. 35, no. 6, pp. 1254–1268, Jun. 2017.
-170 -170
[8] J. Kim, L. Xian, A. Maltsev et al., “Study of coexistence between 5G
small-cell systems and systems of the fixed service at 39 GHz band,”
-175 -175
in Proc. IEEE MTT-S Int. Microwave Symp, May 2015, pp. 1–3.
-180 -180 [9] G. Hattab, E. Visotsky, M. Cudak et al., “Coexistence of 5g mmwave
0 500 1000 0 20 40 60 80
Protection Radius (m) Protection Threshold (degrees) users with incumbent fixed stations over 70 and 80 GHz,” in Proc. IEEE
Globecom Workshops (GC Wkshps), Dec. 2017, pp. 1–5.
Fig. 8: Median DL aggregate interference in the presence of [10] ——, “Interference mitigation techniques for coexistence of 5G
spatial and angular protection zones (dISD = 600m). mmwave users with incumbents at 70 and 80 GHz,” Submitted
to IEEE Trans. on Wireless Commun., 2017. [Online]. Available:
https://arxiv.org/abs/1801.05405
-180
Downlink
-185
Uplink [11] M. Ghorbanzadeh, E. Visotsky, P. Moorut et al., “Radar inband and
ISD=200m out-of-band interference into LTE macro and small cell uplinks in the
-185 ISD=400m
-190
ISD=600m 3.5 GHz band,” in Proc. IEEE Wireless Communications and Networking
Aggregate Interference (dBm/Hz)

Aggregate Interference (dBm/Hz)

-190 50% load

-195
Conf. (WCNC), Mar. 2015, pp. 1829–1834.
-195
[12] A. Khawar, A. Abdelhadi, and T. C. Clancy, “Coexistence analysis
-200 between radar and cellular system in LoS channel,” IEEE Antennas
-200
Wireless Propag. Lett., vol. 15, pp. 972–975, 2016.
-205
-205 [13] FCC International Bureau Application Filing and Reporting System.
-210 [Online]. Available: http://licensing.fcc.gov/myibfs/
-210
25% load
[14] 3GPP, “Study on scenarios and requirements for next generation access
-215
-215 technologies,” 3rd Generation Partnershio Project (3GPP), Tech. Rep.,
-220 -220
2017.
0 5 10 15 20 0 5 10 15 20
Guard band (MHz) Guard band (MHz) [15] “Title 47 of the code of federal regulations,” Federal Communications
Commission (FCC), Tech. Rep. Section 101.115, July 2017.
Fig. 9: Median aggregate interference with variations of the [16] 3GPP, “Study on 3D channel model for LTE,” TR 36.873, Sep. 2017.
[17] ITU-R, “Receiving reference earth station antenna pattern for earth
guard band. stations in FSS in the frequency range from 2 to 31 ghz coordinated
after 1993,” Tech. Rep. S.465-6, 2016.
[18] B. A. Coalition, “Petition for rulemaking to amend and modernize parts
VI. C ONCLUSION 25 and 101 of the commission’s rules to authorize and facilitate the
deployment of licensed point-to-multipoint fixed wireless broadband
5G spectrum access in the mid-band spectrum presents an service in the 3700 – 4200 mhz band,” Tech. Rep., 2017.
opportunity for enhancing spectral utilization and providing