SUSLOV

K.COM/ANDREY
©SHUTTERSTOC

ENHANCEMENT OF
SATELLITE-TO-PHONE
LINK BUDGET
An Approach Using Distributed Beamforming

Zhuoao Xu , Yue Gao , Gaojie Chen , Ryan Fernandez ,

Vedaprabhu Basavarajappa, and Rahim Tafazolli

S
mall satellites in low-Earth orbit (LEO) attract or voice in a few Kbps, cannot be transmitted directly
much attention from both industry and academia. from a LEO satellite to a standard mobile cellular phone
The latest production and launch technologies due to the insufficient link budget. The current LEO con-
constantly drive the development of LEO constel- stellation network has to use an extra ground device to
lations. However, wideband signals, except text messages receive the signal from the satellite first and then for-
ward the signal to the user equipment (UE). To achieve
Digital Object Identifier 10.1109/MVT.2023.3320403 direct network communications between LEO satellites
Date of current version: 16 October 2023 and UE, in this article, we propose a novel distributed

This work is licensed under a Creative Commons Attribution 4.0 License.

DECEMBER 2023 | IEEE VEHICULAR TECHNOLOGY MAGAZINE For more information, see https://creativecommons.org/licenses/by/4.0/ ||| 85
beamforming technology based on the superposition initially, their service will offer text messages only [4].
of electromagnetic (EM) waves radiated from multiple Therefore, how to realize a future LEO satellite network
satellites that can significantly enhance the link budget. for direct access from a mobile cellular phone has be-
EM full-wave simulation and Monte Carlo simulation come a hot issue.
results are provided to verify the effectiveness of the Recent work by Mohammed [5] proposed a cell-free
proposed method. The simulation results show a nearly massive multiple-input, multiple-output (MIMO)-based
6-dB enhancement using two radiation sources and an architecture and discussed various aspects of ultra-
almost 12-dB enhancement using four sources. The dense LEO satellite network design, but it mainly fo-
received power enhancement could be doubled com- cused on a joint optimization framework for the power
pared to the diversity gain in multiple-input, single-output allocation and handover management processes. The
(MISO). Furthermore, other practical application chal- MIMO applicability to ultrahigh-frequency (UHF) satel-
lenges, such as synchronization and the Doppler effect, lite communications in the geostationary orbital space
are also presented. was analyzed in [6]. The results show that applying the
MIMO technique can increase the channel capacity in
Introduction narrowband UHF satellite communications. The authors
In the 1990s, LEO (lower than 2,000 km above Earth’s of [7] considered a land mobile satellite MIMO, where two
surface) constellations, such as Iridium and Globalstar, geosynchronous Earth orbit satellites simultaneously
emerged and aimed to provide portable satellite phone communicate with a mobile user terminal. Its simulation
service with global coverage. Still, most of them could results show that dual-satellite MIMO communications
not sustain their development due to their big rival, the can achieve better bit error rate performance under
3G terrestrial cellular network, which could provide bet- the same signal-to-noise ratio condition compared to
ter performance with lower costs. single-satellite communications. However, the diversity
However, over the past few years, LEO constellations or multiplexing gain provided by MISO is proportional
have staged a comeback because of the rapid develop- only to the number of coordinated satellites N [8]. To fur-
ment of satellite technologies, increased demand, and ther improve the received signal strength, we propose a
much lower launch costs. Compared to the old LEO new distributed beamforming technology. Compared to
constellations, new LEO constellations have much bet- the maximum diversity gain of N in MISO, the received
ter performance, benefitting from digital communication power could be enhanced by N 2 times at maximum. As
payloads, advanced modulation schemes, multibeam known to all, beamforming is an effective technology
antennas, and more sophisticated frequency reuse schemes in modern communication systems that can concen-
[1]. Among them, the most crucial factor is launch tech- trate energy and extend the communication range [9].
nology. Rocket reuse and launching multiple satellites It has evolved to distributed transmit beamforming [10].
with one rocket have significantly reduced launch costs However, the distributed beamforming applied to LEO
[2]. New LEO constellation representatives Starlink and constellations is significantly distinct from that for ter-
OneWeb launched their first test satellites in 2018 and restrial networks. The main reason is that satellites are
2019, respectively. so far away from one another that they cannot form a
Standard mobile cellular phones do not have enough main beam as in conventional arrays. Another reason
reception gain, and LEO satellites’ transmit power and is that satellites keep moving along their orbits, causing
antenna gain are limited by size and weight, resulting their distribution to change at any time.
in an inadequate link budget for direct wideband com- The rest of this article is organized as follows. The
munication between LEO satellites and UE. Beam link “Novel Distributed Beamforming for LEO Satellite-Based
budgets for the user downlink for three LEO constella- Networks” section gives a link budget calculation based
tions were given in [1]. Take the Starlink constellation on the 5G UE operating frequency. The distributed
as an example. Although it reduces the orbital altitude beamforming technology is then proposed to compen-
from 1,150 km to 550 km, the received power is still sate for the insufficient link budget. Both the structure
less than −110 dBm, following the link budget calcula- of the distributed array and the theory are presented in
tion process in [3]. Given this, the current solution ad- detail. Then, in the “Simulation Results and Discussion”
opted by Starlink is to use an extra device composed section, simulations are given to verify that the received
of a phased antenna array with a high gain of more power could be enhanced through distributed beam-
than 30 dBi. The device receives signals from the LEO forming and that the beam coverage patterns could be
satellite and then forwards the signals to UE, which designed by changing relevant parameters. Afterward,
limits the portability of LEO satellite networks. A few we introduce a few challenges that need to be addressed
advanced companies, such as Lynk, AST, and SpaceX, in the future in the “Challenges” section. Finally, the con-
are exploring a direct way to connect cellular phones clusion and future work are given in the “Conclusion and
with LEO satellites. However, with limited bandwidth, Future Work” section.

86 ||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | DECEMBER 2023

Novel Distributed Beamforming for LEO
Satellite-Based Networks LEO constellations have staged a
comeback because of the rapid development
Link Budget of satellite technologies, increased
We provide the link budget in Table 1 to achieve direct 5G demand, and much lower launch costs.
communication between the LEO satellite and UE. Based
on the current 5G network and smartphones working at
3.5 GHz, the received power of the signal transmitted from The current coverage scheme is appears in Figure 1(a),
a single satellite is calculated as −100.4 dBm. However, where each satellite has its coverage area and works
from 3rd Generation Partnership Project Technical Specifi- independently. When the UE is nearly moving out, the
cation (3GPP) 38.101-1 [11], it is known that the minimum coverage of the present satellite is limited by the eleva-
reference sensitivity for operating band n78 is −96.5 dBm. tion angle; the satellite should smoothly hand over the
Thus, to achieve direct LEO satellite network access from UE to the incoming satellite. To achieve seamless cov-
UE, the solution must at least provide a 4-dB enhance- erage, inevitably, there will be some overlap at the
ment. As the demand for bandwidth increases, more edge of the beam, but the interference within the over-
received power will be required. For example, 100-MHz lapping area is unwanted. Inversely, for our distributed
bandwidth corresponds to a 13.5-dB power improvement. beamforming technology, constructive interference is
exactly what is desired. The authors of [12] present
System Model of Distributed Beamforming research on how EM waves interact and exchange ener-
As we know, a LEO megaconstellation includes tens of gy. From the article, it can be found that beams merely
thousands of small satellites that run in fixed orbits interfere constructively or destructively in the over-
with different altitudes, and a phased-array antenna is lapped area and remain in their original propagation
commonly used on the satellites, which can provide a after passing it. Under this circumstance, satellites
high gain and electrically steerable beam scanning. need to steer their radiation beams properly for accu-
rate coverage. In Figure 1(b), the dashed line repre-
sents beams radiated from satellites propagating
Table 1 The link budget. toward the same coverage area.
Parameter Value Figure 2 depicts the EM waves superposition principle.
Distance 550 km
Assume that satellite 1 (Sat1) and satellite 2 (Sat2) run in
the same orbit, communicate with the same UE on the
Operating frequency 3.5 GHz
ground, and are located on either side of the centerline.
Free space path loss 158.1 dB
Effective isotropic radiated power 36.7 dBW
Transmit antenna gain 37.1 dBi
Receive antenna gain 0 dBi
E1 E1z
Atmospheric losses and rain attenuation 5 dB Sat2 Orbit
E2y
β1
Losses from transmitter 2 dB H2 Sat1
β2 z E1y
Losses from receiver 2 dB
E2z α2 H1
E2 R1
R

Received power calculated −100.4 dBm

2

α1
y
Ground
UE
x

Single Beam θ2
Overlapped Beams θ1
LEO Satellite
Coverage Area
Earth’s Center

(a) (b)
Figure 2 The superposition of EM waves radiated from LEO satel-
lites. The black solid curve represents the ground, the gray dashed
Figure 1 The beam coverage scheme. (a) Satellites work indepen- curve above it represents the satellite operating orbit, and the
dently and have even coverage. (b) The same coverage is obtained black point below represents Earth’s center. Sat1: satellite 1; Sat2:
by distributed beamforming. satellite 2.

DECEMBER 2023 | IEEE VEHICULAR TECHNOLOGY MAGAZINE ||| 87

In a spherical coordinate system with Earth’s center as combination of received powers from multiple sources.
the origin, the central angles of Sat1 and Sat2 are i 1 and For example, two satellites can provide a maximum gain
i 2, respectively; a 1 denotes the angle between Sat1’s line of 3 dB with MISO.
with Earth’s center and its line with the UE, and so does Assuming there are N satellites running in the same
a 2 . As known to all, radiating EM waves are transverse direction and working constructively, and the other
EM waves. Thus, the electric (E) field vector, the magnet- conditions are the same as above, then (1) should be
ic (H) field vector, and the wave vector (W) are perpen- modified as
dicular to one another. Considering that there is also a
N N
S av = 1 # / (E iy + E iz) # / H i dt
T
rectangular coordinate with the UE as the origin, E- fields
T 0 i=1
and H-fields can be decomposed along the y- and z-axes i=1

and the x- and z-axes independently; b 1 and b 2 represent NE 20

# (cos b 1 + cos b 2 + g + cos b N ) (- e z) 
Z0
the decomposition angles of E 1 and E 2 . According to the 2 2
N E0
geometry, it can be easily obtained that b 1 = a 1 + i 1; simi- # (- e z) . (2)
Z0
larly, b 2 = a 2 + i 2 . Besides, R 1 and R 2 are the distance
from Sat1 and Sat2 to the UE. The received power would ideally be increased by N 2
with distributed beamforming, according to (2), while
Performance Analysis only a factor of N can be achieved by MISO. If the in-
As described in Figure 2, suppose the two satellites above creased received power is expressed in decibels, the for-
the UE work coherently. In other words, two beams of the mer would be twice as large as the latter. As the number
same frequency arrive at the receiver simultaneously and of cooperating satellites increases, the received power
in phase; then, the E-field components along the x-axis will be significantly boosted. Finally, the received signal
are constructively superimposed. At the same time, the would become strong enough to achieve direct network
E-field components along the z-axis should be subtracted communication between LEO satellites and UE. Howev-
here, while they should be summed when they are in the er, considering the satellite spacing and inclination an-
same space divided by the xoz-plane. The superposition gle, the maximum value N 2 may not always be available
of the H-field is similar except for the substitution of x in reality, and it will decrease when the satellite spacing
for y. But in Figure 2, due to the two H-fields having the and inclination angle increase.
same direction, they can be directly added together. The orbits of satellites could cross one another. In
After superimposing the E- and H-fields in their decom- order not to lose generality and convenience, two inter-
position surfaces, synthesized E and H vectors can be secting orbits are considered. Suppose that one orbit is
obtained. The average energy flux (power per unit area) along the x-axis, and the other orbit intersects it with an
can be further calculated by the average Poynting vector angle of p; M of a total N satellites move in the intersect-
calculation, as follows: ing orbit. The average Poynting vector can be calculated
as follows:
2 2
S av = 1 # / (E iy + E iz) # / H i dt
T
N N
T 0 i=1 i=1 S av = 1 #0 T / (E ix + E iy + E iz) # / (H ix + H iy) dt
E 20 T i=1 i=1
= [(cos b 1 + cos b 1 cos T{ + cos b 2 cos T{ + cos b 2) 
Z0 6N - MN (1 - cos p )@ E
2 2
0
(- e z) + (sin b 1 + sin b 1 cos T{ - sin b 2 cos T{ # (- e z) . (3)
Z0
- sin b 2) (- e y)]
2E 20
# [(cos b 1 + cos b 2) (- e z) + (sin b 1 - sin b 2) (- e y)] From the result, it can be known that the maximum
Z0
4E 0 2 enhancement provided by N satellites in two intersecting
(1) # (- e z)
Z0 orbits is N 2 - MN (1 - cos p ) . Similar to the previous sce-
nario, the maximum is also obtained when T{ and b are 0.
where T represents a time period, both E 1 and E 2 have In particular, when half of N satellites move in the perpen-
the same E-field effective value 2 E 0, Z 0 is the imped- dicular orbit, the maximum enhancement will be N 2 /2.
ance of free space, and T{ is the phase difference of inci-
dent beams caused by the routes. It can be seen from (1) Simulation Results and Discussion
that in order to approach the maximum received power,
both T{ and b should be as close to 0 as possible. Differ- EM Simulations for Superposition
ent orbital altitudes make it possible for satellites to get EM simulations are given to verify the superposition
closer when a satellite runs right above another. Ideally, of EM waves radiated from distributed sources far apart.
two satellites would yield four times as much received The radiation beam of the antenna on a satellite is usual-
power as one satellite, meaning a 6-dB enhancement. By ly narrow so that energy can be concentrated in the
contrast, the received power obtained by MISO is the expected direction. For the convenience of verification, a

88 ||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | DECEMBER 2023

dipole antenna is adopted here without loss of generality. Coverage Pattern
In particular, the simplest case with only two radiation The receiver gains from constructive interference;
sources is considered. accordingly, the surroundings should lose some power
As in Figure 3, two dipole antennas are displayed sym- due to the destructive interference. Since the distance
metrically along the centerline. Both have an inclination between satellites is much longer than the wavelength,
angle with a value of 30°. Considering that the main beam there may be frequent switching between constructive
radiation direction is perpendicular to the antenna, the and destructive interference within the coverage. There-
main beam convergence location can be easily obtained fore, knowing the beam coverage pattern generated by
according to the geometric structure. The operating fre- interference is vital. We provide pattern simulations
quency is set to 3.5 GHz, corresponding to a wavelength of
85.7 mm in a vacuum. Generally, a distance of more than 10
wavelengths can be regarded as far enough for antennas Z
to work independently. Therefore, the two half-wavelength Antenna 1 Antenna 2
dipoles are displayed 10 wavelengths apart. Besides, each X Y
source’s input power and initial phase can be modified
based on requirements. Given the geometry of the model,
constructive interference will appear along the centerline
when the initial phases and power are set the same.
In addition, a Poynting vector represented by arrows
is plotted on the green rectangular sheet. The arrow size
and color indicate its value, and an arrow’s point indi-
cates its direction. It can be easily found that radiation
power gets weaker as it propagates away. Significantly,
the synthetic Poynting vector around the centerline
points straight down because the transverse compo-
nents cancel one another. The red dashed line repre-
sents the destructive interference, where the dominant
longitudinal components cancel one another. From the
point of view of vector superposition, the synthetic vec-
tor would become more prominent when the angle be-
tween vectors is less than 90°. As a result, the received
power can be enhanced due to the constructive interfer-
ence brought by distributed beamforming.
As mentioned above, the same setup and symmetrical
position of the two sources ensures that their radiation
waves constructively interfere with each other along Figure 3 Two EM waves interfere constructively, represented by a
Poynting vector.
the centerline. The synthetic Poynting vector on the line
is extracted from z = -200 mm mm to z = -770 mm to
compare with that of a single antenna and that of MISO.
Figure 4 illustrates the power received in three ways: 6
one satellite, two satellites by MISO, and two satellites by 5
Enhancement (dB)
Received Power

distributed beamforming. As a reference, the received 4

power of the single radiation source is normalized. The 3
received power obtained through MISO is proportional 2
to the number of sources, as depicted in the “Novel Dis- 1
tributed Beamforming for LEO Satellite-Based Networks” 0
section, increasing the power twice, or 3 dB. By contrast, –1
–2
the power achieved by distributed beamforming is even –200 –300 –400 –500 –600 –700 –800
lower than that of the single antenna at the beginning, Superposition Point Along the z-axis (mm)
but it surpasses that of MISO as the observation point
Distributed Beamforming
gets close to the target point (around z = -742 mm). Un- MISO
der the condition of a 30° inclination, the received power Single Antenna
enhancement is around 4.7 dB at the target. Through
the EM simulations, the possibility of increasing the re- Figure 4 A comparison of the received power at different posi-
ceived power by distributed beamforming is verified. tions along the centerline by three methods.

DECEMBER 2023 | IEEE VEHICULAR TECHNOLOGY MAGAZINE ||| 89

obtained under different input parameters. As mentioned is reasonable. Furthermore, it can be observed from the
in the previous part, the initial variables include the figure that by adopting distributed beamforming, the
transmitting power, initial phase, orbit height, and incli- maximum received signal gain of nearly 6 dB can be
nation angle of the beam relative to the vertical line direc- achieved with two transmitting sources, which is larger
tion. The vertical line direction represents the direction than the 3 dB of diversity gain.
in which the satellite points toward Earth’s center. In the second case, four satellites are used to imple-
Assuming that the single beamwidth of the satellite is ment distributed beamforming; then, the spot beam
2.5°, and given that the operating satellite height is 550 km, coverage pattern appears. More than two satellites gen-
it can be concluded that the coverage area of the single erally create a 2D distribution, resulting in spot beams.
beam is a circle at the nadir with a radius of 12.5 km. The When all the inclination angles of the beams from the
power density at the boresight is the largest and drops to four directions are 0.1°, interference spots will be gener-
half the maximum value at the edge, which is −3 dB. As ated whose diameters are around 24 m. As depicted in
the beam slants away from the nadir, the coverage pat- Figure 6, four transmitting sources could improve nearly
tern gradually changes from a circle to an ellipse. More- 12 dB by distributed beamforming, which is larger than
over, as the transmitting beam is steered, the power is that of MISO (6 dB). The two cases above present feasible
adjusted to maintain a constant power flux density at beam coverage patterns with multiple satellites working
Earth’s surface, compensating for variations in antenna coherently. The pattern can be designed based on spe-
gain and path loss associated with the steering angle. cific requirements by changing the initial parameters.
Here, we present only two representative beam cover-
age cases. In the first case, only two satellites are used Challenges
to implement distributed beamforming; then, interfer- To achieve portable communication between LEO satel-
ence fringes appear. In Cartesian coordinates, if the line lites and UE directly, some other significant challenges
between the two satellites is parallel to the x-axis, then must be further addressed.
the interference fringes will be parallel to the y-axis. The
fringe width depends on the angle of the incident wave at Synchronization
the receiver. When the angle between the two beams is In practice, the radiated EM waves must meet some syn-
large, the fringes are narrow and dense. On the contrary, chronous characteristics to realize the constructive inter-
as the angle becomes smaller, the fringes become wide ference of satellite radiation beams. The first is frequency
and sparse. As detailed in Figure 5, when the inclination synchronization, which requires excellent stability and
angles of both transmitting beams are 0.1°, the interfer- accuracy of satellite-mounted radio-frequency modules.
ence fringes are broad, and the width is about 12 m. In The crystal oscillators and front-end devices on different
addition, the lateral distance between the two satellites satellites need to ensure that the transmitted EM waves
is approximately 1.92 km, which is too close for two sat- have the same frequency, which is the premise that multi-
ellites in the same orbit to achieve. But considering the ple EM waves can be superimposed to form a standing
high density of LEO megaconstellation deployments and wave. The second is phase synchronization. The incident
satellites that could cooperate with others running in dif- EM waves at the receiver need to have the same phase to
ferent orbits, this assumption about the inclination angle maximize the effect of field strength superposition.

6
6
Power Enhancement (dB)

5
5
Power Enhancement (dB)

6 6
4 4
4
Received

4
Received

2 3 2 3
dB

dB

0 2 0 2
10 40
5 20 40 1
10 1 0 20
0
0 –20 0
–5 y-axis (m) –40 –20 x-axis (m) 0
y-axis (km) –10 x-axis (km) 0 –40
–10

(a) (b)

Figure 5 The coverage pattern generated by two satellites working coherently.

90 ||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | DECEMBER 2023

Otherwise, the superimposed signal may be worse than feature of LEO constellation networks, offering a range of
the single one. Therefore, the initial phase of the trans- benefits and challenges that must be carefully considered
mitter needs to be carefully set after considering many in the design and operation of such systems.
factors. The third is time synchronization. Satellites are
always in motion, constantly changing their positions, so Doppler Effect
the distance between each satellite and the UE is differ- Since LEO satellites run in low-altitude orbits, they have to
ent. If satellites send signals to the UE simultaneously, the travel at high speed, which will change the frequency of
receiver will receive them at different times, or the signal the signal at the receiver, namely, the Doppler effect.
may even be too weak to be received successfully. Thus, When the transmitter and receiver are getting close to
the transmitted signals must be set with further delays to each other, the frequency increases; conversely, when
ensure they arrive at the receiver simultaneously. they move far away, the frequency decreases. The signifi-
cant Doppler frequency shift increases the difficulty of
Intersatellite Links receiver demodulation and degrades communication per-
Intersatellite links (ISLs) are communication links between formance. The authors of [14] present the analytic deriva-
two or more satellites in orbit around Earth. In a LEO con- tion of the Doppler shift about the signal transmitted from
stellation network, ISLs are a critical technology that LEO satellites to UE. There are three main solutions for
enables the exchange of data and control signals among Doppler frequency-shift estimation and compensation.
satellites in the constellation [13]. SpaceX successfully test- One of them is using the geometric analysis method
ed its laser ISLs in late 2020 and has launched satellites fea- directly to calculate the relative velocity of communica-
turing laser ISLs since September 2021. Leveraging the tion satellites and UE and then calculate the Doppler fre-
development of ISLs and LEO constellations, the proposed quency shift. Another is using the Kalman filter
approach is more feasible and does not incur much addi- frequency-shift estimation algorithm. The third one uses
tional cost. ISLs are highly robust and can be independent- the maximum-likelihood estimation algorithm to calculate
ly networked without relying on the terrestrial network, the Doppler frequency-shift factor, which is fed back to
expanding the coverage of the communication system. the frequency compensation module. Then, the frequency
Additionally, their features, such as high data rates and low compensation module precompensates the transmitted
latency, can help to address the synchronization problem signal frequency to achieve the purpose of frequency syn-
mentioned above. However, the acquisition, pointing, and chronization between the transmitter and the receiver.
tracking mechanisms among satellites are pretty compli- These Doppler shift estimation and compensation meth-
cated, and the laser link is greatly affected by space illumi- ods can reduce the Doppler shift value and improve com-
nation and other factors. For megaconstellations, the munication quality. But the reality is much more complex
routing of ISLs is a highly complex problem due to the rela- than the model assumes, and the specific situation should
tive position of LEO satellites changing all the time. In addi- be analyzed according to the actual constellation.
tion, signal processing on satellites dramatically increases
the complexity and development difficulty, reducing the Constellation Design and Beam Switching
adaptability of satellites to technology upgrades and Constellation design is often ambiguous for researchers
updates. This is regarded as the biggest issue that hinders because the design is constantly being adjusted as
the development of laser links. Overall, ISLs are a key requirements change. Ever y company modifies its

12
12
15 10
Power Enhancement (dB)

10
Power Enhancement (dB)

15
8
10 8
Received

10
Received
dB

6
5 5
dB

6
4 0
4
0 50
10 50
2 2
10 0
0 0
0
y-axis (km) –10 0 y-axis (m) x-axis (m) 0
–10 x-axis (km) –50 –50
(a) (b)
Figure 6 The coverage pattern generated by four satellites working coherently.

DECEMBER 2023 | IEEE VEHICULAR TECHNOLOGY MAGAZINE ||| 91

 network from everywhere in the world with only a cell
The high running speed of LEO satellites phone. The whole structure of the distributed array was
yields a Doppler shift, which may severely also described in detail to explain how it works. In addi-
affect beamforming performance. tion, EM simulations verified that extremely distant
sources can still interfere with one another. By utilizing
constructive interference, distributed beamforming
constellation parameters more or less in terms of the could obtain a higher enhancement of received power
number of satellites and orbit altitudes. Also, different than in other ways. Two representative beam coverage
companies come up with different constellations. For cases were presented to show the patterns obtained
example, even though SpaceX’s, OneWeb’s, and Telesat’s under different input parameters.
constellations have inclined orbits combined with polar Finally, to tackle the challenges listed above, we pro-
orbits, they have entirely different orbital characteristics vide some potential further direction, as follows:
in terms of altitude, inclination, the number of orbits, and ■■ Synchronize the frequency, phase, and time: Each satel-
so on. To finally reach distributed beamforming for lite has its local oscillator (LO) working independently.
satellite -to -phone communication, an appropriate These LOs, generated by crystal oscillators, typically
constellation design or AI-based resource allocation is exhibit variations that may cause a phase drift. A mas-
indispensable. Take the current Starlink constellation, for ter–slave architecture may work on synchronization.
example, which gives the satellite a fixed elevation angle. In brief, one satellite first transmits a reference signal
When the satellite is about to travel out of the coverage to another; afterward, the receiving satellite uses
area of the UE, the UE is already under the coverage of phase-locked loops to lock the signal, thus achieving
other satellites. Consequently, the UE evaluates each sat- frequency and phase synchronization. Time alignment
ellite’s coverage and chooses the optimal one. But in the could also be attempted by the satellites sending each
case of multiple satellites working together in this article, other trigger reference signals. Laser communication
the situation will undoubtedly become more complicat- within ISLs may provide more help in sharing informa-
ed. The system needs to select the specific satellite com- tion and reference signals.
position according to the received power increment ■■ Suppress the Doppler effect: The high running speed of
required by the UE, the position of each satellite, and LEO satellites yields a Doppler shift, which may
their state information to decide beam switching. severely affect beamforming performance. Therefore,
the influence of the extended frequency offset on the
System Integration With Other Communications distributed beamforming performance should be ana-
Spectrum sharing among different communication sys- lyzed. Besides, finding solutions, i.e., the Doppler shift
tems becomes common as the demands on frequency compensation techniques, to suppress the effect could
resources increase. In lower-frequency bands, the LEO be important future work.
satellite network needs to share the spectrum with the ■■ Combine with a reconfigurable intelligent surface: A
terrestrial cellular network, while in higher-frequency reconfigurable intelligent surface (RIS) integrates a
bands, it may need to share the resources with medium- large number of low-cost passive reflective elements on
orbit or geostationary satellites. Spectrum sharing may the plane. It can significantly improve the performance
cause more interference among different communication of wireless communication networks by intelligently
systems, so managing the interference among multiple reconfiguring the wireless propagation environment.
systems and applying resource allocation schemes is Numerous advantages, such as low power consump-
necessary. Relevant work was presented in [15], where a tion, flexible deployment, and almost no extra delay
general spectrum-sharing framework in satellite and ter- and thermal noise, make it competitive in wireless com-
restrial networks was introduced, both in the downlink munication. Thus, the combination of distributed
and uplink. beamforming and a RIS would have significant poten-
tial. From the “Simulation Results and Discussion” sec-
Conclusion and Future Work tion, it is known that satellite parameters can change
In this article, we first introduced the dependence of the the shape of the pattern. Taking advantage of this fea-
existing LEO satellite network on the ground terminal. ture, future work should not only design the coverage
Without the terminal, UE, such as standard smart- pattern but also combine it with a RIS to intelligently
phones, cannot access satellite networks directly. The transmit or reflect the beams according to the needs of
relevant link budget calculation was then given. A new user coverage. In the meantime, the requirements of
method named distributed beamforming was proposed distributed beamforming performance will also guide
to compensate for the insufficient received power. With constellation design and beam switching management.
this technology, LEO satellites can offer the Internet to ■■ Apply to the 5G Frequency Range 2 band: The frequency
UE directly, making it possible to access the LEO satellite we use in this article belongs to the 5G Frequency Range

92 ||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | DECEMBER 2023

 (FR) 1 band. Future phones may use the 5G FR2 band,    Ve d a p r a b h u B a s a v a r a j a p p a
with a typical frequency of 28 GHz, which can significant- ([email protected]) is with the
ly improve bandwidth. But conversely, signals at higher 5G/6G Innovation Center, Institute for
frequencies experience more attenuation. Adopting dis- Communication Systems, University of
tributed beamforming technology is a competitive solu- Surrey, GU2 7XH Guildford, U.K. His
tion because it can provide the closest to the theoretical research interests include 5G antennas,
maximum received power increment. And accordingly, millimeter-wave technology, satellite antennas, and
fewer satellites would be needed to cooperate, which phased-array antennas. He is a Senior Member of IEEE.
saves satellite scheduling resources. Furthermore, work- Rahim Tafazolli (r.tafazolli@surrey.
ing in such a frequency band may interfere with other ac.uk) is currently a professor of mobile
satellite systems, and finding a smooth way to deal with and personal communications at and the
this problem would be necessary. director of the 5G/6G Innovation Center,
Institute for Communication Systems, Uni-
Acknowledgment versity of Surrey, GU2 7XH Guildford, U.K.
This work was supported by the National Natural Science He is a Senior Member of IEEE and a fellow of the Royal
Foundation of China, under Grant 62341105. Yue Gao is Academy of Engineering, Institution of Engineering and
the corresponding author. Technology, and Wireless World Research Forum.

Author Information References

[1] I. del Portillo, B. G. Cameron, and E. F. Crawley, “A technical com-
Zhuoao Xu ([email protected]) is current- parison of three low earth orbit satellite constellation systems to
ly pursuing his Ph.D. degree with the 5G/6G provide global broadband,” Acta Astronaut., vol. 159, pp. 123–135,
Jun. 2019, doi: 10.1016/j.actaastro.2019.03.040.
Innovation Center, Institute for Communi- [2] S. Chen, Y.-C. Liang, S. Sun, S. Kang, W. Cheng, and M. Peng, “Vision,
cation Systems, University of Surrey, GU2 requirements, and technology trend of 6G: How to tackle the chal-
lenges of system coverage, capacity, user data-rate and movement
7XH Guildford, U.K. His research interests speed,” IEEE Wireless Commun., vol. 27, no. 2, pp. 218–228, Apr. 2020,
include distributed beamforming, low- doi: 10.1109/MWC.001.1900333.
[3] O. Popescu, “Power budgets for CubeSat radios to support ground
Earth orbit constellations, and satellite-to-phone technolo- communications and inter-satellite links,” IEEE Access, vol. 5, pp.
gies. He is a Student Member of IEEE. 12,618–12,625, 2017, doi: 10.1109/ACCESS.2017.2721948.
[4] Lynk Global, “Messaging from an orbital base station to cellular
Yue Gao ([email protected]) is a chair user equipment applications with message processing via a card
professor in the School of Computer Sci- operating system,” U.S. Patent US20210360587A1, Nov. 18, 2021.
[5] M. Y. Abdelsadek, H. Yanikomeroglu, and G. K. Kurt, “Future ultra-
ence and the director of the Intelligent dense LEO satellite networks: A cell-free massive MIMO approach,”
Networking and Computing Research Cen- in Proc. IEEE Int. Conf. Commun. Workshops (ICC), 2021, pp. 1–6, doi:
10.1109/ICCWorkshops50388.2021.9473753.
ter, Fudan University, Shanghai 200483, [6] B. Ramamurthy, W. G. Cowley, and G. Bolding, “MIMO applicability
China, and a visiting professor at the to UHF SATCOM,” in Proc. IEEE Global Commun. Conf. (GLOBECOM),
2016, pp. 1–7, doi: 10.1109/GLOCOM.2016.7841983.
­University of Surrey, GU2 7XH Guildford, U.K. His research [7] H. Peng, C. Qi, Y. Chen, Q. Zhang, D. Chen, and R. Ding, “Channel
interests include smart antennas, sparse signal process- modeling and signal transmission for land mobile satellite MIMO,”
in Proc. IEEE Global Commun. Conf. (GLOBECOM), 2021, pp. 1–6, doi:
ing, and cognitive networks for mobile and satellite sys- 10.1109/GLOBECOM46510.2021.9685823.
tems. He is a Senior Member of IEEE. [8] L. Zheng and D. N. C. Tse, “Diversity and multiplexing: A funda-
mental tradeoff in multiple-antenna channels,” IEEE Trans. Inf. The-
Gaojie Chen ([email protected]. ory, vol. 49, no. 5, pp. 1073–1096, May 2003, doi: 10.1109/TIT.2003.
uk) is currently an assistant professor at 810646.
[9] P. Ioannides and C. A. Balanis, “Uniform circular and rectangu-
the 5G/6G Innovation Center, Institute for lar arrays for adaptive beamforming applications,” IEEE Anten-
Communication Systems, University of Sur- nas Wireless Propag. Lett., vol. 4, pp. 351–354, 2005, doi: 10.1109/
LAWP.2005.857039.
rey, GU2 7XH Guildford, U.K. His research [10] R. Mudumbai, D. R. B. Iii, U. Madhow, and H. V. Poor, “Distributed trans-
interests include information theory, mit beamforming: Challenges and recent progress,” IEEE Commun.
Mag., vol. 47, no. 2, pp. 102–110, Feb. 2009, doi: 10.1109/MCOM.2009.
wireless communications, cooperative communica- 4785387.
tions, cognitive radio, the Internet of Things, secrecy [11] 5G; NR; User Equipment (UE) Radio Transmission and Reception; Part 1:
Range 1 Standalone (3GPP TS 38.101-1 Version 17.6.0 Release 17), ETSI
communications, and random geometric networks. He is TS 138 101-1 V17.6.0, 3GPP, Sophia Antipolis, France, Jun. 2022.
a Senior Member of IEEE. [12] H. G. Schantz, “On the superposition and elastic recoil of electro-
magnetic waves,” 2014, arXiv:1407.1800.
Ryan Fernandez (r.fernandez@surrey. [13] H. Kaushal and G. Kaddoum, “Optical communication in space:
ac.uk) is currently pursuing his Ph.D. Challenges and mitigation techniques,” IEEE Commun. Surveys Tuts.,
vol. 19, no. 1, pp. 57–96, First Quart. 2017, doi: 10.1109/COMST.2016.
degree with the 5G/6G Innovation Center, 2603518.
Institute for Communication Systems, Uni- [14] I. Ali, N. Al-Dhahir, and J. E. Hershey, “Doppler characterization for
LEO satellites,” IEEE Trans. Commun., vol. 46, no. 3, pp. 309–313,
versity of Surrey, GU2 7XH Guildford, U.K. Mar. 1998, doi: 10.1109/26.662636.
His research interests include low-Earth [15] C. Zhang, C. Jiang, L. Kuang, J. Jin, Y. He, and Z. Han, “Spatial spec-
trum sharing for satellite and terrestrial communication networks,”
orbit megaconstellations and direct satellite-to-mobile IEEE Trans. Aerosp. Electron. Syst., vol. 55, no. 3, pp. 1075–1089, Jun.
technologies. 2019, doi: 10.1109/TAES.2018.2889585.

DECEMBER 2023 | IEEE VEHICULAR TECHNOLOGY MAGAZINE ||| 93