link budget analysis for reconο¬gurable smart surfaces in
TRANSCRIPT
1
Link Budget Analysis for Reconfigurable Smart Surfaces in AerialPlatforms
Safwan Alfattani, Wael Jaafar, Yassine Hmamouche, Halim Yanikomeroglu, and Abbas YongaçogluNon-terrestrial networks, including Unmanned Aerial Vehicles (UAVs), High Altitude Platform Station (HAPS) nodes and Low
Earth Orbiting (LEO) satellites, are expected to have a pivotal role in sixth-generation wireless networks. With inherent featuressuch as flexible placement, wide footprints, and preferred channel conditions, they can tackle several challenges faced by currentterrestrial networks. However, their successful and widespread adoption relies on energy-efficient on-board communication systems.In this context, the integration of Reconfigurable Smart Surfaces (RSS) into aerial platforms is envisioned as a key enabler ofenergy-efficient and cost-effective aerial platform deployments. RSS consist of low-cost reflectors capable of smartly directing signalsin a nearly passive way. In this paper, we investigate the link budget of RSS-assisted communications for two RSS reflectionparadigms discussed in the literature, namely βspecularβ and βscatteringβ paradigms. Specifically, we analyze the characteristicsof RSS-equipped aerial platforms and compare their communication performance with that of RSS-assisted terrestrial networksusing standardized channel models. In addition, we derive the optimal aerial platform placements for both reflection paradigms.Our results provide important insights for the design of RSS-assisted communications. For instance, given that a HAPS has a largearea for RSS, it provides superior link budget performance in most studied scenarios. In contrast, the limited RSS area on UAVsand the large propagation loss in LEO satellite communications make them unfavorable candidates for supporting terrestrial users.Finally, the optimal location of an RSS-equipped platform may depend on the platformβs altitude, coverage footprint, and type ofenvironment.
Index TermsβReconfigurable Smart Surfaces (RSS), Reconfigurable Intelligent Surfaces (RIS), aerial platform, Unmanned AerialVehicle (UAV), High Altitude Platform Station (HAPS), Low Earth Orbit (LEO) satellite.
I. INTRODUCTION
As the fifth generation (5G) of wireless systems are beingactively deployed, researchers in the wireless community havestarted investigating new technologies and innovative solutionsto tackle the challenges and fulfill the demands of of next-generation (6G) networks. One of the main challenges involvessupporting ubiquitous connectivity with high data rates in anenergy efficient way. With the inherent limitations of terrestrialenvironments, non-terrestrial networks are envisioned as anenabling technology for ubiquitous connectivity in futurewireless communications. Non-terrestrial networks includingsuch platforms as Unmanned Aerial Vehicles (UAVs), HighAltitude Platform Stations (HAPS) nodes, and Low Earth Orbit(LEO) satellites are capable of addressing such challenges ascoverage holes, blind spots, sudden increases in throughput de-mand, and terrestrial network failures. They can address thesechallenges due to their wider coverage footprints, strong line-of-sight (LoS) links, and flexibility of deployment comparedto terrestrial networks [1]β[5]. Moreover, the standardizationefforts of the Third Generation Partnership Project (3GPP)aiming to utilize aerial platforms for 5G and beyond have made
This work is supported in part by a scholarship from King AbdulAzizUniversity, Saudi Arabia, in part by the National Sciences and EngineeringResearch Council of Canada (NSERC), and in part by Huawei Canada.
S. Alfattani is with the Electrical Engineering Department of King Abdu-laziz University, Saudi Arabia, and with the School of Electrical Engineeringand Computer Science, University of Ottawa, Ottawa, ON, Canada, (e-mail:[email protected]).
W. Jaafar and H. Yanikomeroglu are with the Department of Systems andComputer Engineering, Carleton University, Ottawa, ON, Canada, e-mails:(waeljaafar, [email protected])
Y. Hmamouche is with the Mathematical and Electrical Engineering De-partment, IMT-Atlantique, Brest, France, (e-mail: [email protected]).
A. Yongacoglu is with the School of Electrical Engineering and Com-puter Science, University of Ottawa, Ottawa, ON, Canada (e-mail: [email protected]).
significant progress, as demonstrated by the standardizationdocuments TR 38.811 [6], TR 22.829 [7], and TS 22.125 [8].Furthermore, several commercial projects are either in theirinitial phases of deployment or under development, whichaim to design different types of aerial platforms capable ofsupporting wireless communications. Such projects include theStarlink LEO constellation by SpaceX [9], the Stratobus HAPSby Thales [10], and the Nokia Drone Networks [11]. Never-theless, aerial platforms are not yet a cutting-edge technology,and their current size, weight, and power (SWAP) limitationsneed to be further improved.
On the other hand, reconfigurable smart surfaces (RSS)have recently been introduced as an energy-efficient enablingtechnology for next-generation wireless networks [12].1 AnRSS is a thin, lightweight metasurface integrated with passiveelectronic components or switches to provide unique andcontrolled manipulation of the wireless signals. It can alterthe amplitude of the impinging signal, adjust its phase, anddirect it to a target in a nearly passive way [13], [14]. Thedeployment and utilization of RSS in terrestrial networkshas been extensively studied, and several research works,prototypes, and industrial experiments, summarized in [17],[18], demonstrated the potential of this technology.
Given the potential spectral and energy efficiencies of RSS-assisted communications and the stringent energy requirementsof communications through aerial platforms, equipping the lat-ter with RSS presents an attractive solution to the SWAP issue.Indeed, due to the low-cost and negligible energy consumptionof RSS reflectors, their use in aerial platforms is expectedto support low-cost wireless communications for an extendedflight duration. In our previous work [1], we discussed the
1RSS are referred to in the literature by other names, such as software-controlled metasurfaces [13], reconfigurable intelligent surfaces (RIS) [14],intelligent reflecting surfaces (IRS) [15] and smart reflect-arrays [16].
arX
iv:2
008.
1233
4v4
[ee
ss.S
P] 1
6 A
ug 2
021
2
feasibility of integrating RSS in aerial platforms of differenttypes. We proposed a control architecture, detailed potentialuse cases, and examined associated challenges. In the contextof UAVs only, the authors of [19] showed that using RSS inUAVs enables a panoramic view of the environment, which canprovide full-angle 360Β° signal reflections compared to 180Β°reflections in RSS-assisted terrestrial networks. In [20], the au-thors presented potential use cases of RSS mounted on UAVsand discussed related challenges and research opportunities.Similarly, [21] investigated the potential of RSS-equippedUAV swarms, where a use case was studied to demonstrate theachievable data rate performance of such systems. The authorsof [22] studied the problem of wireless sensor data collection,where sensors were assisted by an RSS mounted on a UAVto reach the collecting sink. The objective was to maintaindata freshness through accurate optimization of the UAVβslocation and the RSS phase-shifting configuration. Finally, inthe context of LEO satellites, the authors of [23] investigatedthe utilization of RSS to support inter-satellite links in theterahertz (THz) band. The results demonstrated a significantperformance improvement in terms of bit error rate comparedto non-RSS-assisted communications.
Previous works have not thoroughly investigated RSS-enabled communication links, and so a link budget analysisfor RSS-assisted non-terrestrial networks remains unexamined.In contrast, a number of works have studied path-loss modelsfor RSS-assisted terrestrial communications [14], [24]β[27].While most of these models are based on mathematicalanalysis using different approaches, some of them have beenexperimentally validated [24]. These studies revealed the ex-istence of two regimes that govern the performance of RSS-assisted communication systems. The first is the βspecularβreflection paradigm, where the path-loss model is analyzedusing geometrical optics and imaging theory. The second isthe βscatteringβ reflection paradigm, which obeys plate scat-tering theory and radar cross-section analysis. The factors thatdetermine the governing regime of the RSS-assisted systemsare the geometrical size of the RSS units, the communicationfrequency, and the distances separating an RSS from thetransmitter and receiver. Typically, when an RSS is within arelatively short distance from a transmitter and/or a receiver,or when the RSS units are electrically large (e.g., their dimen-sions are ten times larger than the wavelength denoted by _),the path loss is governed by the specular reflection paradigm[14], [24], [28]. Otherwise, the RSS-assisted communicationfollows the plate scattering reflection paradigm (i.e., when thedistances between the RSS units and transmitter or receiverare large or when the RSS unit dimensions are very small)[29], [30]. It should be noted that the scattering paradigm canbe designated as βfar-fieldβ paradigm, whereas the specularreflection can be referred to as βnear-fieldβ paradigm.
Due to the specific design and environmental characteristicsof aerial platforms compared to terrestrial systems, the former,when equipped with an RSS, are expected to have a differentlink budget analysis. Therefore, it is necessary to investigateand assess the benefits of RSS-enabled aerial platforms. Threemajor factors impact the feasibility of RSS-enabled aerialplatforms, namely the operating frequency or wavelength, the
platformβs surface area reserved for RSS units, and the operat-ing altitude. While higher frequency signals are preferable forlarger capacity links and enabling the deployment of more RSSreflector units, higher frequency signals are more vulnerableto path-loss degradation from the communication distance andatmospheric attenuation. Also, larger platform RSS sizes maylead to a higher reflection gain, which may not be realizable forpractical platform sizes. Finally, although platforms operatingat higher altitudes might be preferable due to their largercoverage footprint, they suffer from excessive propagationlosses that may not be compensated for even large area ofRSS.
In this paper, we aim to provide the link budget analysisfor RSS-enabled aerial platform communication systems forspecular and scattering reflection paradigms. The receivedpower of the system, for different RSS-enabled platforms,is calculated while taking into account the signal strengthlosses due to the specific characteristics of each platform.To minimize signal loss, we derive the optimal platformlocation and maximum feasible number of RSS reflectors on-board each platform. Link budget expressions are then derivedfor realistic communication conditions, as defined by 3GPPstandards. Finally, numerical results are provided to supportthe proposed link budget analysis. The contributions of thepaper are highlighted as follows:
1) To the best of our knowledge, the performance parame-ters of RSS-enabled communications have been derivedonly for terrestrial networks and not for non-terrestrialsystems, where different signal losses are experienceddue to specific characteristics of aerial platforms andatmospheric phenomena. The impact of these factors istaken into account in this work.
2) We investigate the link budget analysis for RSS-enabledaerial platform communications for different platformtypes, namely UAVs, HAPS nodes, and LEO satellites.To improve performance, we optimize platform locationsand the maximum feasible number of mounted RSSreflectors.
3) We extend the link budget analysis to more realisticchannel conditions, as defined in the 3GPP standards.We also provide numerical results to support the relatedparameters evaluation and the link budget analysis.
The rest of the paper is organized as follows. SectionII discusses the conditions of the reflection and scatteringparadigms, and then analyzes the link budget for RSS-assistedterrestrial networks. Section III then exposes the characteristicsof aerial platforms and derives the optimal platform placementfor RSS-assisted aerial communications. This section also in-vestigates the related link budget for both reflection paradigms.Numerical results for the terrestrial and non-terrestrial RSS-assisted systems are presented and elaborated in Section IV.Finally, Section V concludes the paper.
II. LINK BUDGET ANALYSIS FOR RSS-ASSISTEDTERRESTRIAL NETWORKS
Here, we present the link budget analysis for RSS-enabledterrestrial communications (e.g., RSS units mounted on a
3
Figure 1. System model of RSS in a terrestrial environment.
building). Given the identified specular and scattering reflec-tion paradigms, we derive the related received power expres-sions.
Typically, terrestrial environments are characterized byblockages that result in high path loss, especially in dense-urban and urban environments. Accordingly, terrestrial net-work planning depends on cellular densification, where multi-ple base stations (BSs) are deployed in a relatively small areato ensure coverage of all users within the area. But this comesat the expense of additional costs and inter-cell interference.To alleviate such inconveniences, the RSS can be deployed onthe facades of buildings and used to either extend the cellularcoverage or improve the signal quality in poorly served areas.As shown in Fig. 1, the signal forwarded by the RSS from theBS to the user equipment (UE) can either substitute the directlink when the latter is absent, or it can be added constructivelyto the weak direct link in order to strengthen the receivedsignal.
A. The Specular Reflection Paradigm
The relation that governs the paradigm of the RSS-assistedcommunications has been defined in [24], [31] as follows:
πlim =2π΄π‘_, (1)
where πlim denotes the maximum distance between the RSSand either the transmitter (Tx) or the receiver (Rx) in thespecular reflection paradigm2, and π΄π‘ is the total RSS area.
When the length and width dimensions of the RSS units arelarge enough (i.e., above 10_), and the distance separating theRSS from the Tx/Rx is less than πlim, then the RSS can beconsidered in the near-field. In this paradigm, the impingingspherical wave forms a circular and divergent phase gradient
2An example of (Tx,Rx) in the terrestrial environment is (BS,UE).
on the RSS area. Accordingly, the RSS acts as an anomalousmirror and the two-hop link acts as a one-hop path. Hence,the distance path loss is affected by the summation of thetraveled distances (i.e., Tx-RSS and RSS-Rx distances), whichis known as the specular reflection paradigm [14], [24], [28].
For a Tx-RSS (resp. RSS-Rx) distance π· β€ πlim and a large-sized RSS (LRSS), where the reflector unit sizes are 10_Γ10_m2, the minimum required number of reflectors for specularreflection, denoted πmin, can be calculated as
π· =2π΄π‘_
=2πmin (10_)2
_= 200_πmin β πmin =
π·
200_. (2)
The defined πmin will be later used to assess the feasibility ofthe RSS using the specular reflection paradigm.
In order to conduct the link budget analysis in this paradigm,we assume the Tx-Rx communication assisted by a building-mounted LRSS. Let π₯(π‘) be the transmitted signal by the Tx.Then, the received (noise-free) signal at the Rx, denoted byπ¦(π‘), can be written as [14]
π¦(π‘) = π π₯(π‘), (3)
where π is the wireless channel coefficient, expressed for thesake of simplicity with the log-distance path-loss model. Thelatter is given by
π =βοΈππ‘πΊπ‘πΊπ
(_
4ππ0
) ((π0ππ
)πΎ+
πβοΈπ=1
ππΎ
0 πππβ π (\π+ππ)
(ππ‘π + πππ )πΎ
), (4)
where ππ‘ , πΊπ‘ , and πΊπ are the transmit power and the trans-mitter and receiver gains, respectively. Also, π0 denotes thereference distance, ππ is the distance between Tx and Rx,3
2πΎ = πΌ is the path-loss exponent, and π is the total number ofRSS reflector units. Finally, ππ , ππ‘π , πππ represent the reflectionloss of the ππ‘β RSS reflector, the distance between Tx and RSSππ‘β reflector, and distance between the RSS ππ‘β reflector andRx, respectively. Also, \π and ππ are the corresponding incidentand reflection angles. Now, the received power at Rx, denotedby ππ , can be written as
ππ = ππ‘πΊπ‘πΊπ
(_ π
(πΎβ1)04π
)2 (1ππΎ
π
+πβοΈπ=1
πππβ π (\π+ππ)
(ππ‘π + πππ )πΎ
)2
. (5)
For the sake of simplicity, we assume here that the LRSS canperfectly adjust the desired phase shifts, and that the reflectorsare ideal without any reflection loss, that is,4
\π + ππ = 0 and ππ = 1, βπ = 1, . . . , π. (6)
Moreover, assuming that the variation of ππ‘π and πππ isnegligible across the RSS, we have
ππ‘π + πππ β 2π, βπ = 1, . . . , π, (7)
3We assume here that the direct link Tx-Rx is a weak link [32].4In practice, RSS reflection loss depends on the configuration technology
and building materials [17], [33]. Also, since continuous phase shift imple-mentation is difficult, only a finite discrete set of phase shifts is typicallydesigned. It has been shown that near-optimal RSS performance can berealized using a small number of phase-shift levels [34], [35]. In any case,the loss of a few dB due to the material properties or due to sub-optimal RSSconfiguration is insignificant compared to the signal propagation loss [32].
4
where π = ππ/2. Hence, the received power can be rewrittenas
ππ = ππ‘πΊπ‘πΊπ
(_
4π
)2(π(πΌβ2)0(2π)πΌ
)(1 + π)2 . (8)
According to (8), the location of the RSS has no impact onthe received power, while the path loss is the dominant factor(e.g., πΌ β₯ 3 in urban environments [36]). Thus, in typicalterrestrial environments, although the received power enhancesquadratically with the number of RSS reflectors, it degradesat a much higher rate with the propagation distance.
B. The Scattering Reflection Paradigm
Assuming that the Tx-RSS and RSS-Rx distances are large(i.e., higher than πlim), and that tiny RSS reflector units areused (i.e., with dimensions between 0.1_ and 0.2_), then eachreflector capturing the transmitted signal behaves as a newsignal source that re-scatters the signal towards the UE. Inthis paradigm, βthe scattering reflection paradigmβ the totaleffect on the transmitted signal is the resultant of the cascadedindividual channels Tx-RSS and RSS-Rx [29]β[31].
A scattering paradigm for RSS-assisted communication istypically presented as an alternative to the degradation of directlinks caused by strong blockages [25]. Accordingly, the effectof the direct link is ignored, and the effective received signalis only the one scattered by the RSS. Since in this paradigm,the use of tiny RSS reflector units is advocated, we name thesesmall-sized RSS (SRSS).
To accurately assess the SRSS-assisted terrestrial commu-nications, we present next the link budget analysis for twochannel models, namely the log-distance and 3GPP basedmodels5
1) Log-Distance Channel ModelFor the ππ‘β reflector unit of the SRSS, the channel effect of
the received signal, denoted ππ , is resulting from two cascadedchannels, i.e., Tx-ππ‘β reflector and ππ‘β reflector-Rx [29], [30].The channel coefficient is given by
ππ =βοΈππ‘πΊπ‘πΊπ βπ‘ππππ πππ
β π ππ , π = 1, . . . , π, (9)
where ππ is the adjusted phase shift of the reflector, while βπ‘πand πππ are the complex-valued coefficients representing thelinks between the ππ‘β reflector and both Tx and Rx. The latterare defined by
βπ‘π =
(_
4ππ0
) (π0ππ‘π
)πΎπ π \π‘π , and πππ =
(_
4ππ0
) (π0πππ
)πΎπ π \ππ ,
(10)with \π‘π and \ππ denoting the transmit and receive channelphases, respectively. Following the generalization to the π
reflectors, the received power can be expressed by
ππ = ππ‘πΊπ‘πΊπ
(_
4ππ0
)4π(2πΌ)0
(πβοΈπ=1
πππβ π (ππβ\π‘πβ\ππ )
(ππ‘ππππ )πΎ
)2
. (11)
5Several researchers have recently raised practical concerns about thespecular reflection paradigm and the use of LRSS [17], [37]. Subsequently,we adopt in this paper the practical 3GPP channel model under the scatteringreflection paradigm only.
We assume lossless reflectors, i.e.,
ππ = 1,βπ = 1, . . . , π, (12)
and that
ππ‘ β ππ‘π and ππ β πππ ,βπ = 1, . . . , π, (13)
where ππ‘ and ππ are reference distances measured between thecenter of the SRSS and the Tx and Rx, respectively. Subse-quently, the received power can be maximized by coherentlycombining the received signals through the π reflectors, i.e.,ππ = \π‘π + \ππ . Hence, the received power can be written as
ππ = ππ‘πΊπ‘πΊπ
(_
4π
)4(π(2πΌβ4)0
(ππ‘ππ )πΌ
)π2. (14)
According to (14), ππ degrades faster than in (8), due to thedistances multiplication. Also, ππ is maximized when the RSSis the closest to either Tx or Rx.
2) 3GPP Channel ModelRSS in terrestrial environments can be placed on facades
of buildings to smartly reflect signals toward users. Sincesuch smart buildings are expected to be available in modernurbanized environments, we assume in the following the urbanscenario of the 3GPP standard model [38]6. The total path lossfor the Tx-SRSS and SRSS-Rx links can be written as
ππΏ = PLoSππΏLoS + PNLoSππΏNLoS + ππΏπ, (15)
where PLoS and PNLoS are the LoS and NLoS probabilities,ππΏLoS and ππΏNLoS are the associated losses in the LoS andNLoS conditions, and ππΏπ accounts for the extra loss of indoorusers. The latter vary greatly in terms of building type, locationwithin the building, and movement in the building. For anaccurate calculation of ππΏπ, we refer the reader to [39]. Butsince the focus in our system is on outdoor users, ππΏπ isignored.
The LoS probability in a terrestrial environment between theSRSS and Tx or Rx, assuming that the heights of the RSS-equipped building and Rx are below 13 m, can be given by[38, Table 7.4.2-1]
PLoS =
1 if d2D β€ 18m18π2π·
+ exp(βπ2π·
63
) (1 β 18
π2π·
)if d2D > 18m,
(16)where π2π· is the 2D separating distance (projected on theground) between the SRSS and the Tx or Rx, whereas thepath loss for LoS and NLoS links is given as follows [38,Table 7.4.1-1]:
ππΏLoS = 28 + 22 log(π3π·) + 20 log( π ) + π (17)
andππΏNLoS = max(ππΏLoS, ππΏ
NLoS) (18)
6Practically speaking, we envision that RSS to be deployed on the facadesof high-rise buildings, which are available in urban environments. Also, weexpect RSS to be implemented where they would bring profit to serviceproviders. Due to the high density of customers in urban areas, it is morelikely that RSS will be deployed massively in urban environments and perhapsrarely or never in rural environments.
5
Figure 2. System model of an aerial platform equipped with RSS.
where
ππΏNLoS
= 13.54 + 39.08 log(π3π·) + 20 log( π )β 0.6 (π»π₯ β 1.5) + π, π₯ β {RSS,Rx}, (19)
π3π· is the 3D Tx-SRSS or SRSS-Rx separation distance inmeters, π is the carrier frequency in GHz, π is a log-normalrandom variable denoting the shadow fading, with standarddeviation π = 4 dB and π = 7.8 dB for LoS and NLoS links,respectively, and π»π₯ denotes the SRSS/Rx height. Specifically,for the Tx-SRSS link, π»π₯ = π»π ππ , where π»π ππ is the heightof the building coated with the RSS, while for the SRSS-Rxlink, π»π₯ = π»π π₯ , with π»π π₯ is the Rx height.
Accordingly, the received power, in dBm, can be written as
ππ = ππ‘ + πΊπ‘ + πΊπ β ππΏπ π₯βππ ππ β ππΏππ ππβπ π₯ + 20 log (π),(20)
where ππΏπ π₯βππ ππ and ππΏππ ππβπ π₯ represent the path loss forthe Tx-SRSS and SRSS-Rx links respectively, calculated using(15).
III. LINK BUDGET ANALYSIS FOR RSS-ASSISTEDNON-TERRESTRIAL NETWORKS
In this section, we start by presenting the aerial platformscharacteristics. Then, we derive the link budget analysis ofRSS-enabled non-terrestrial communications (i.e., when anRSS is mounted on an aerial platform, such as a UAV, HAPS,or LEO satellite). Given the identified specular and scatteringreflection paradigms, we derive the related received powerexpressions.
A. Aerial Platforms Characteristics
Three types of aerial platforms can be used to host RSStechnology, namely UAVs, HAPS nodes, and LEO satellites.These platforms are usually deployed at different altitudesand target different coverage areas. Specifically, UAVs aredeployed around 100 m altitude (and can operate up to300 m [40]), to serve an area below 5 km radius [41],
[42]. By contrast, HAPS nodes are quasi-stationary platformspositioned in the stratosphere at an altitude between 17 and50 km. However, most HAPS projects target an altitude of20 km due to its preferred atmospheric characteristics for theplatform stability and communication quality [2]. Thus, HAPSsystems have a much wider footprint than UAVs that spansfrom 40 to 100 km for high throughput [2], [43] and can go upto 500 km according to the International TelecommunicationsUnion (ITU) [44]. Finally, LEO satellites orbit the earth atan altitude between 400 and 2,000 km, with an orbital periodbetween 88 and 127 minutes [6]. Accordingly, LEO satelliteshave the largest coverage footprint. Nevertheless, the coveragearea significantly depends on the satelliteβs altitude, elevationangle, and coverage scheme (i.e., whether it uses a spot orwide communication beam). Consequently, the LEO satelliteβsfootprint has a radius between hundreds and thousands ofkilometers [45]β[47]. To communicate with UAVs and HAPSnodes, UE can use the same device as it would to communicatewith a terrestrial BS. However, different equipment is neededto communicate with LEO satellites, since they require LEOtracking, either mechanically or electronically, in order tocompensate for the satellite motion and achieve a reliablecommunication [6].
In addition to the aforementioned characteristics, the phys-ical size of the aerial platform is crucial to enable hosting theRSS. UAVs have the smallest size, and hence can dedicate onlya small area for RSS. On the other hand, two types of HAPSare identified, namely the aerostatic HAPS and the aerody-namic HAPS. Aerostatic HAPS nodes, especially airships, aregiant platforms whose lengths are typically between 100 and200 m, whereas aerodynamic HAPS nodes have wingspansbetween 35 and 80 m [2]. Finally, the size of current LEOsatellites is less than 10 m [48], [49]. Since HAPS nodes andLEO satellites are typically equipped with solar panel arrays,a part of the platform surface can be used to mount RSSequipment.
B. The Specular Reflection Paradigm
When the Tx and Rx7 are separated by a relatively longdistance, an aerial platform equipped with an LRSS canbe used to assist the communication. Specifically, the Txtransmits its signal to the LRSS-equipped aerial platform.Then, the LRSS smartly reflects the incident signal towardsthe Rx, as illustrated in Fig. 2. Hence, the received noise-freesignal in the specular paradigm is identical to the one in (3),whereas the channel effect is expressed using the free-spacepath-loss formula as
π =βοΈππ‘πΊπ‘πΊπ
(_
4π
) πβοΈπ=1
πππβ π (\π+ππ)
ππ‘π + πππ, (21)
where the LoS wireless link component is predominant, i.e.,the path-loss exponent πΌ = 2πΎ = 2. By following the same
7Notice that Tx and Rx in the non-terrestrial communication context canbe different according to the aerial platform used. For instance, (Tx, Rx) canbe (BS, UE) for a UAV, or it can be (Gateway, UE) for a HAPS platform ora LEO satellite. A HAPS UE might be a mobile, vehicular, or fixed cellularuser device, whereas a LEO has a fixed UE such as a household receiver [9].
6
assumptions as in (12)β(13) and adopting a similar phase-shiftconfiguration as in (6), the received power can be given by
ππ = ππ‘πΊπ‘πΊπ
(_
4π
)2π2
(ππ‘ + ππ )2 . (22)
Unlike (8), the parameters _, π , and (ππ‘ + ππ ) have thesame scaling law for the received power improvement ordegradation.
One of the distinct features of aerial platforms comparedto terrestrial networks is their flexibility of placement, whichallows to further enhance the system performance.
Proposition 1. Given that Tx and Rx are collinearly separatedby distance ππ = 2π, and that the aerial platform is located ataltitude π»π ππ and horizontally separated from the Tx by a dis-tance π , i.e., ππ‘ =
βοΈπ»2
π ππ+ π2 and ππ =
βοΈπ»2
π ππ+ (2π β π)2,
then the optimal placement of the aerial platform underspecular reflection is given by
πβ = π. (23)
That is, the platform is placed over the perpendicular bisectorof the segment Tx-Rx at altitude π»π ππ .
Proof. The specular equivalent path-loss distance, denotedππ π , can be written as
ππ π = ππ‘ + ππ =
βοΈ(π»2
π ππ+ π2) +
βοΈπ»2
π ππ+ (2π β π)2. (24)
To maximize ππ , we need to minimize ππ π through nulling itsfirst derivative, i.e,
πππ π
ππ=
πβοΈπ2 + π»2
π ππ
β 2π β πβοΈ(2π β π)2 + π»2
π ππ
= 0, (25)
from which we obtain πβ = π. When substituting this solutionin the second derivative of ππ π , π2ππ π
ππ2 , we get
π2ππ π
ππ2 =
π»2π ππ
( (π2 + π»2
π ππ
) 32 +
((2π β π)2 + π»2
π ππ
) 32)
((2π β π)2 + π»2
π ππ
) 32(π2 + π»2
π ππ
) 32
(π=π)=
2π»2π ππ
(π2 + π»2π ππ
) 32. (26)
Since π2ππ πππ2 > 0, then πβ = π is the optimal value that
minimizes ππ π . οΏ½
Accordingly, the maximal received power can be calculatedusing (22) for πβ = π as follows:
πβπ = ππ‘πΊπ‘πΊπ
(_
4π
)2π2
4(π»2π ππ
+ π2). (27)
C. The Scattering Reflection Paradigm
Similar to Section II, the scattering reflection paradigm isinvestigated for two channel models, namely the log-distanceand the 3GPP models.
1) Log-Distance Channel ModelSimilarly to the SRSS-assisted terrestrial network, the re-
ceived power at the Rx in the scattering paradigm can beexpressed using (14) for πΌ = 2, thus:
ππ = ππ‘πΊπ‘πΊπ
(_
4π
)4 (π
ππ‘ππ
)2. (28)
Unlike the specular paradigm, where the optimum aerialplatform location is over the perpendicular bisector of seg-ment Tx-Rx, the best platform location under the scatteringparadigm is expected to be different due to the cascadedchannel effect.
Proposition 2. Given that Tx and Rx are collineraly separatedby distance ππ = 2π, and that the aerial platform is locatedat altitude π»π ππ and horizontally separated from Tx by adistance π , then the optimal placement of the aerial platformunder scattering reflection is given by
πβ =
{π Β±
βοΈπ2 β π»2
π ππif π β₯ π»π ππ
π otherwise.(29)
Proof. The scattering equivalent path-loss distance, denotedππ π , can be expressed through
π2π π = (ππ‘ππ )2 =
(π»2
π ππ + π2) (π»2
π ππ + (2π β π)2). (30)
To maximize ππ , we must minimize π2π π by nulling its first
derivative as follows:
ππ2π π
ππ= 4π3 β 12ππ2 +
(8π2 + 4π»2
π ππ
)π β 4ππ»2
π ππ = 0. (31)
By solving (31), we obtain the follwing roots:
πβ =
π +
βοΈπ2 β π»2
π ππ
π ββοΈπ2 β π»2
π ππ
π.
(32)
The second derivative is given by
π2π2π π
ππ2 = 12π2 β 24ππ + 8π2 + 4π»2π ππ , (33)
in which we substitute the roots of (32). Consequently, wefind that (33) is positive only when {π = π and π β€ π»π ππ}or {π = π Β±
βοΈπ2 β π»2
π ππand π β₯ π»π ππ}. These (π, π) values
dictate the best aerial platform locations where received poweris maximal. οΏ½
Physically speaking, Proposition (2) implies that under thescattering reflection paradigm, when the height of a platformis larger than its designed coverage radius, the platform shouldbe placed at the mid-point between the Tx and Rx. However,when the targeted coverage radius is larger than the platformβsheight, two optimal locations can be used, which are beingclose to either the Tx or the Rx.
By substituting the result of (32) in (30) and then (28), thereceived power at Rx can be written as
πβπ = ππ‘πΊπ‘πΊπ
(_
4π
)4π2
(πβπ π)2 . (34)
7
where πβπ π is the equivalent path-loss distance for the optimalplatform location, given by
πβπ π =
{π»2
π ππ+ π2 if π β€ π»π ππ
2π»π πππ otherwise.(35)
Unlike (22), the impact of π», π, and _, is more importantthan π in (34) due to the scattering effect.
2) 3GPP Channel ModelFor a fair comparison with RSS-assisted terrestrial com-
munications, we investigate here the link budget analysis forRSS-enabled aerial platforms with realistic 3GPP channelmodels. In what follows, given that platforms operate atdifferent altitudes and thus may experience different attenu-ation phenomena, we study the link budget for RSS-assistedUAVs separately from RSS-equipped HAPS nodes and LEOsatellites.
β’ The UAV-Based Model:RSS-enabled UAVs are envisioned to cooperate with terres-
trial BSs to support terrestrial users. They can tackle coveragegaps or increase the capacity of terrestrial users by reflectingthe BSsβ signals towards users [1]. In such context, it isacceptable to consider the 3GPP channel model betweenterrestrial BSs and UAVs [40]. Due to the relatively lowaltitude of UAVs, the latter are generally assumed to havefull LoS conditions. Consequently, the path loss in rural andurban environments8 can be given by [40, Table B-2]
ππΏrural = max[23.9 β 1.8 log10 (π»π₯), 20
]log10 (π3π·)
+ 20 log10 (40π π
3) + π, π₯ β {UAV,UE} (36)
and
ππΏurban = 28+22 log10 (π3π·)+20 log10 ( π )+π, π₯ β {UAV,Rx},(37)
where π»π₯ , π3π· , π , and π are defined as in (17)β(19), exceptthat π has a log-normal distribution with standard deviationgiven by9
π =
{4.2 π (β0.0046 π»π₯ ) in rural environment,4.64 π (β0.0066 π»π₯ ) in urban environment.
(38)
Subsequently, the received power in dBm can be expressed by
ππ = ππ‘+πΊπ‘+πΊπβππΏπ π₯βππ΄π βππΏππ΄πβπ π₯+20 log (π), (39)
where ππΏπ π₯βππ΄π and ππΏππ΄πβπ π₯ are the path losses for theTx-UAV and UAV-Rx links, calculated using either (36) or(37), depending on the considered environment.
β’ The HAPS/LEO based Model:The support of future wireless networks through HAPS
and LEO satellites is envisioned for both rural and urbanareas [2], [4]. To assess the performance in these different
8In contrast to RSS-assisted terrestrial communications, where RSS de-ployment is expected to be concentrated in urban areas, RSS-equipped aerialplatforms have the flexibility needed for both urban and rural areas.
9Note that variable π may be modified to include the effect of atmosphericturbulence, and thus the value of π can be adjusted accordingly.
Table IPARAMETERS OF HAPS/LEO LOS PROBABILITY MODEL (40)
Parameter Dense Urban Urban Ruralπ1 0.04235 9.668 -99.95π2 1.644 0.547 -0.5895π3 27.32 -10.58 104.1
Figure 3. LoS probability of HAPS/LEO relative to elevation angle (differentenvironments).
environments, LoS probabilities are required. Based on theelevation angle of the aerial platform relative to the terrestrialTx or Rx, denoted by π, LoS probabilities can be estimated fordifferent environments using [6, Table 6.6.1-1]. For the sakeof simplicity, we propose to substitute [6, Table 6.6.1-1] by aLoS probability function, defined as
PπΏππ = π1ππ2 + π3, (40)
where ππ , π = {1, 2, 3} are the parameters that depend on theenvironment, determined in Table I. The accuracy of (40) isvalidated in Fig. 3, where it is shown to agree with the resultsof [6, Table 6.6.1-1]. Now, the path loss for LoS and NLoSconditions, denoted by ππΏ, can be written as
ππΏ = PLoSππΏLOS + PNLoSππΏNLOS, (41)
where Pπ¦ and ππΏπ¦ , π¦ β {LoS,NLoS}, are defined as in (15).The signal path between a HAPS/LEO and a terrestrial Tx orRx undergoes several stages of propagation and attenuation.Specifically, the path losses ππΏπ¦ , π¦ β {LoS,NLoS} arecomposed as follows [6]:
ππΏπ¦ = ππΏπ¦
π+ ππΏπ + ππΏπ + ππΏπ, π¦ β {LoS,NLoS}, (42)
where ππΏπ¦
πis the basic path loss, ππΏπ is the attenuation
due to atmospheric gasses, ππΏπ is the attenuation due toeither ionospheric or tropospheric scintillation, and ππΏπ is thebuilding entry loss, expressed in dB.ππΏ
π¦
πaccounts for the signalβs free-space propagation
(πΉπππΏ), clutter loss (πΆπΏπ¦), and shadow fading (π π¦), i.e.,
ππΏπ¦
π= πΉπππΏ + πΆπΏπ¦ + π π¦ , π¦ β {LoS,NLoS}, (43)
8
Table IIAVERAGE CLUTTER LOSS AND SHADOW FADING STANDARD DEVIATION IN
THE KA-BAND
Parameter Dense Urban Urban RuralπΆπΏππΏππ 38.6 38.6 23.15ππΏππ 1.75 4 1.15πππΏππ 14.7 6 10.75
where
πΉπππΏ = 32.45 + 20 log10 ( π ) + 20 log10 (π3π·), (44)
with π3π· being the 3D distance between the HAPS/LEOand the terrestrial Tx or Rx, expressed as a function of theHAPS/LEO altitude π»π§ , π§ β {HAPS,LEO} and the platformβselevation angle π as follows:
π3π· =
βοΈπ 2πΈ
sin2 (π) + π»2π§ + 2π»π§π πΈ β π πΈsin(π), (45)
where π πΈ denotes the earthβs radius. The clutter loss, πΆπΏπ¦ ,represents the attenuation caused by buildings and environ-mental objects. Its value depends on π, π , and the environmenttype. In LoS conditions, πΆπΏLoS = 0, while for NLoS, theπΆπΏNLoS values of [6, Tables 6.6.2-1 to 6.6.2-3] can be usedfor the typical Ka spectrum band (i.e., between 26.5 and 40GHz). Finally, π π¦ is a zero-mean normal distribution withstandard deviation ππ¦ , π¦ β {LoS,NLoS}, whose values aredetermined in [6, Tables 6.6.2-1 to 6.6.2-3]. For the sake ofsimplicity, we present the average values of the parameters inTable II10.ππΏπ is the attenuation caused by absorption due to atmo-
spheric gases. Its value depends mainly on π , π, and π3π· .According to [6], the effect of atmospheric gases is negligiblefor π β€ 10 GHz. However, for higher frequency bands suitablefor RSS operations with a large number of reflectors, theselection of frequency windows with minimum atmosphericeffect is important. In addition to the aforementioned factors,ππΏπ depends on the dry air pressure π, water-vapor densityb, and temperature π [50]. An illustrative example is shownin Fig. 4 for different Tx-Rx link lengths, selected for typicaldistances between a HAPS and Tx or Rx (20 and 100 km)and between a LEO satellite and Tx or Rx (1,000 km). Therelated parameters (π, b and π) are selected on the basis ofthe mean annual global reference atmosphere, i.e., π = 101300Pa, b = 7.5π/π3, and π = 15Β°C [51]. The ππΏπ calculation iscarried out using the steps detailed in [50].
Finally, the scintillation loss ππΏπ is caused by rapid fluc-tuations of the received signal amplitude and phase. Thereare two types of scintillation losses, namely the ionosphericscintillation and the tropospheric scintillation. The former onlysignificantly disrupts signals at frequencies below 6 GHz,whereas the latter affects only signals in frequencies above6 GHz. The impact of the ionospheric scintillation is onlysignificant for latitudes in the range [β20Β°, 20Β°] [6]. Theionospheric scintillation loss can be calculated as detailed in
10To be noted that variable π π¦ can be modified to include the effect ofexternal factors, such as the atmospheric turbulence, and platformβs drift.Hence, the value of ππ¦ can be modified accordingly.
0 100 200 300 400 500 600 700 800 900 1000
Frequency (GHz)
10-1
100
101
102
103
104
105
106
107
108
Atm
osp
her
icga
sesatt
enuat
ion
PL
g(d
B)
dry air (20 km)water vapor (20 km)dry air (100 km)water vapor (100 km)dry air (1000 km)water vapor (1000 km)
Figure 4. Attenuation due to atmospheric gases ππΏπ vs. frequency (differentpath lengths).
[52]. Specifically, it is derived from the measured peak-to-peakfluctuation as follows:
ππΏπ =ππΉβ
2, (46)
where ππΉ is the peak-to-peak fluctuation, equal to 1.1 dB forπ = 4 GHz, and calculated by the following equation for π β€ 6GHz:
ππΉ ( π β€6 GHz) = ππΉ ( π =4 GHz) Β· ( π /4)β1.5. (47)
Since high frequency bands are advocated for the RSS-enabled HAPS/LEO platforms, tropospheric scintillation istaken into account in the path-loss model. Specifically, thewireless signal fluctuations are due to sudden changes in therefractive index caused by temperature, water vapor content,and barometric pressure variations. Also, low elevation angles(especially below 5Β°) are significantly affected by the scintil-lation loss, due to the longer path of the signal and the widerbeam width. The value of the tropospheric scintillation loss isseason and region dependent. We refer the reader to [53] forthe detailed calculation steps. To give an idea of the typicalpower attenuation level, the tropospheric attenuation with 99%probability at 20 GHz in Toulouse, France is tabulated in [6,Table 6.6.6.2.1-1] for different elevation angles. For the sakeof simplicity, we model the related tropospheric scintillationloss by
ππΏπ = 14.7 π (β1.136) , π β [0β¦, 90β¦], (48)
which is a valid approximation of [6, Table 6.6.6.2.1-1] asshown in Fig. 5.
Subsequently, the received power in dBm at Rx can beexpressed by
ππ = ππ‘ + πΊπ‘ + πΊπ β ππΏπ π₯βπ»πΏβ ππΏπ»πΏβπ π₯
+ 20 log (π), π»πΏ β {π»π΄ππ, πΏπΈπ}, (49)
where ππΏπ π₯βπ»πΏand ππΏπ»πΏβπ π₯ are calculated using (41).
9
10 20 30 40 50 60 70 80 90
Elevation angle #/ (degrees)
0
0.2
0.4
0.6
0.8
1
1.2Troposphericscintillationloss(dB)
3GPPEq.(48)
Figure 5. Scintillation loss ππΏπ vs. elevation angle.
The link budget analysis of Section III is summarized inTable III.
IV. RESULTS AND DISCUSSION
In this section, we evaluate the received power for differentRSS-mounted platforms, and the impact of several parametersis investigated. Based on the unique features of each platform,as discussed in Sections II and III, we assume typical valuesof altitude, coverage radius, and RSS area for each platform,as shown in Table IV. Moreover, we consider the height ofthe terrestrial Tx π»Tx = 25 m, while the receiverβs height isπ»Rx = 1.5 m.
A. Impact of the Number of Reflectors and Environment Type
Given the selected platform, the type of mounted RSSreflectors is of great concern, since different communicationparadigms can be followed and these have different costs.According to [30], the cost is expected to be approximatelyproportional to the RSS size and number of RSS reflectorunits. Consequently, larger reflectors, with dimensions above10_ Γ 10_ and operating in the specular reflection paradigm,are more expensive than small reflectors (i.e., with dimensionslower than 0.2_ Γ 0.2_), which operate in the scatteringreflection paradigm. Hence, the maximal number of reflectorsto install on a platform, denoted πmax, is limited by thereserved area on the platform for the RSS π΄π‘ and the reflectorβssize π΄π = π1_Γπ2_, where π1_ > 0 and π2_ > 0 are the lengthand width of a reflector unit, respectively. Their relation isdefined by
πmax =π΄π‘
π΄π=
π΄π‘
π1π2_2 . (50)
In order to highlight the potential gains of using RSS-equipped platforms, we consider that a communication be-tween a terrestrial Tx and a terrestrial Rx is assisted by anRSS-equipped platform, namely a building facade, a UAV,a HAPS, or a LEO satellite, characterized as in Table IV.For the sake of simplicity, we assume that Tx and Rx are
100 102 104
Number of re.ectors N
-160
-140
-120
-100
-80
-60
-40
-20
0
Receivedpower(dBm)
TerrestrialUAVHAPSLEONmax (Eq.(50))Nmin (Eq.(2))
Figure 6. Received power vs. number of reflectors (different platforms;specular reflection paradigm).
located at the edges of the platformβs coverage footprint (i.e.,within a distance 2π). Moreover, due to the flexibility of aerialplatforms, we assume that they are placed at the optimallocation according to the considered reflection paradigm, whilethe terrestrial RSS is assumed to be midway between Tx andRx11. The operating frequency π = _/π = 30 GHz (Ka-band),where π is the lightβs velocity in m/s, the receive antenna gainπΊπ = 1 (i.e., 0 dBi), and the path-loss πΌ = 4 for the ter-restrial communication. Finally, the following transmit/receiveparameters are set as follows for the log-distance channelmodel: transmit power ππ‘ = 40 dBm and transmit antennagain πΊπ‘ = 1 (i.e., 0 dBi). For the 3GPP model, ππ‘ and πΊπ‘ arefixed according to the related standards. Specifically, for theterrestrial and UAV systems, we assume that ππ‘ = 35 dBmand πΊπ‘ = 8 dBi [38], while for the HAPS and LEO systems,we set ππ‘ = 33 dBm and πΊπ‘ = 43.2 dBi [6].
Given the specular reflection paradigm, Fig. 6 presentsthe received power as a function of the number of reflec-tors for different platforms. We notice that for UAVs andLEO satellites, the minimum required number of reflectorsfor specular reflection is greater than the maximum numberof reflectors that can be placed on the platformβs surface.Therefore, specular reflection cannot be achieved for RSS-equipped UAVs and LEO satellites for the specified coverageareas. This is due to the limited area available for RSS onUAVs and to the relatively high operating altitude of LEOsatellites. However, specular reflection can be realized usingRSS-equipped HAPS systems or terrestrial environments. Thisis because of the clear LoS links with HAPS nodes andthe relatively short communication distances in terrestrialenvironments. As we can also see in Fig. 6, πmin = 27, 000reflectors for HAPS provides coverage in a 100 km circulararea at ππ = β64 dBm. For a smaller coverage footprint, itis expected that a lower number of reflectors would be used.
11This assumption for the terrestrial RSS is justified by the fact that thelatter cannot be moved later to another location given that Tx and Rx locationsmay change.
10
Table IIILINK BUDGET SUMMARY
Scenario Size of RSS reflector Channel Model Link Budget Dominant parameter(s)LRSS-assisted terrestrial communication
(Specular paradigm) 10_ Γ 10_ Log-distance Eq. (8) π, _ and π (for πΌ = 2)π (for πΌ > 2)
SRSS-assisted terrestrial communication(Scattering paradigm) [0.1_ Γ 0.1_, 0.2_ Γ 0.2_]
Log-distance
3GPP
Eq. (14)
Eq. (20)
_ (for πΌ < 4)_, ππ‘ , and ππ (for πΌ = 4)
π
LRSS-assisted non-terrestrial communication(Specular paradigm) 10_ Γ 10_ Log-distance Eq. (27) _, π , π»π ππ and π
SRSS-assisted non-terrestrial communication(Scattering paradigm) [0.1_ Γ 0.1_, 0.2_ Γ 0.2_]
Log-distance3GPP (UAV)
3GPP (HAPS/LEO)
Eq. (34)Eq. (39)Eq. (49)
_
π
π
Table IVTYPICAL PARAMETERS OF DIFFERENT TYPES OF PLATFORMS.
Platform Altitude(π»π ππ)
Coverage radius(π)
RSS area(π΄π‘ )
Terrestrial 5 m 0.5 km 5 Γ 10 m2
UAV 200 m 2 km 0.25 Γ 0.25 m2
HAPS 20 km 50 km 40 Γ 20 m2
LEO 500 km 500 km 5 Γ 10 m2
100 102 104 106 108
Number of re.ectors N
-350
-300
-250
-200
-150
-100
-50
0
Receivedpower(dBm)
Nmax (Eq.(50))Dense urban env.Urban env.Rural env.
Terrestrial
UAV
HAPS
LEO
Figure 7. Received power vs. number of reflectors (different platforms andenvironment types; scattering reflection paradigm).
Although a small number of reflectors is required in terrestrialenvironment for specular reflection, the received power isworse than in the HAPS scenario. This is due to degradedterrestrial communication channels, compared to LoS wirelesslinks for the RSS-equipped HAPS. Accordingly, a HAPS is thepreferred RSS mounting platform in the specular reflectionparadigm.
For the scattering reflection paradigm, we present in Fig. 7the received power as a function of the number of reflectors fordifferent platforms. For the sake of simplicity, we show onlythe results corresponding to the 3GPP channel models wheredifferent types of environments are studied, namely denseurban, urban, and rural. Given the same number of reflectors,the RSS-equipped UAV system achieves the best power per-
formance due to short distances and clear LoS wireless links.Moreover, the difference in performance between the rural andurban environments is minimal, at about 2 dB. For a number ofreflectors close or equal to πmax, the LEO system has the worstreceived power performance, which degrades significantly inaccordance with the density of the urban environment. Incontrast, the HAPS system realizes the best ππ values fora number of reflectors near or equal to πmax. This remainsthe case even in dense urban environments, where a HAPSsystem may compensate for performance degradation by usinga higher number of RSS reflectors or by reducing its coveragefootprint (i.e., where the Tx and Rx are closer). Consequently,it is worth noting that RSS-equipped LEO systems may requirefurther redesigning in order to be feasible, while the HAPScoverage footprint needs to be adjusted according to theenvironment type of the served area.
B. Impact of the Carrier Frequency
As shown in Table III, the carrier frequency is a dominantparameter in the link budget analysis. Although using higherfrequencies enables high capacity links and addresses thespectrum scarcity issues, such frequencies as millimeter waveand Terahertz may suffer from significant signal attenuation.Nevertheless, high frequencies enable the use of small-sizedreflectors. And since a large number of these can be used insmall areas, this may counterbalance the signal attenuation.
In Fig. 8, we evaluate the received power as a function of thecarrier frequency, for the same communication model intro-duced previously. Here, we assume that each platform hosts themaximal number of reflectors πmax, and that the link budgetanalysis for the scattering reflection paradigm is realized forthe log-distance channel model for the sake of comparison.The results show that in both specular and scattering reflectionparadigms, the HAPS system provides the best performancedue to its large surface area that accommodates the highestnumber of RSS reflectors. On the other hand, the RSS-equipped UAV performs worse than the terrestrial RSS underthe specular reflection paradigm because of its small surfacearea, which accommodates a very small number of reflectors.In particular, when π β€ 12 GHz, a UAV with dimensionsof 0.25 Γ 0.25 m2, cannot host a single reflector. Given thescattering reflection paradigm, all platforms demonstrate stableperformance for any carrier frequency. Indeed, by combining
11
Specular reflection
Scattering reflection
Figure 8. Received power vs. carrier frequency (different platforms).
(50) into (14) and (34) respectively, we obtain the maximalreceived power, expressed by
πmaxπ = ππ‘πΊπ‘πΊπ
(_
4π
)4π΄2π‘(
π1π2_2)2 (ππ‘ππ )πΌ
=ππ‘πΊπ‘πΊπ
(4π)4
(π΄π‘
π1π2
)2 1(ππ‘ππ )πΌ
, (51)
for the terrestrial environment. However, for the non-terrestrialone, (if π β€ π»π ππ) πmax
π is expressed as
πmaxπ = ππ‘πΊπ‘πΊπ
(_
4π
)4π΄2π‘(
π1π2_2)2(π»2
π ππ+ π2
)2 (52)
=ππ‘πΊπ‘πΊπ
(4π)4
(π΄π‘
π1π2
)2 1(π»2
π ππ+ π2
)2 ,
otherwise (i.e., π > π»π ππ),
πmaxπ =
ππ‘πΊπ‘πΊπ
(4π)4
(π΄π‘
π1π2
)2 1(2π»π πππ)2 . (53)
According to (51)β(53), the received power is no longerdependent on the frequency (or the wavelength), but ratheron π΄π‘ , π1 and π2.
Fig. 9 shows the received power as a function of thecarrier frequency, given different RSS-equipped platforms andenvironment types. These results were obtained using the3GPP channel models for the scattering reflection paradigm,i.e., (20), (39) and (49). Also, we assume the use of themaximal number of reflectors πmax, dry air atmospheric at-tenuation, and an average tropospheric scintillation of 0.5 dBfor the HAPS and LEO systems. We notice that the receivedpower of both terrestrial and UAV systems are insensitive tofrequency. Indeed, the frequency attenuation is successfullyaddressed through the deployment of a higher number ofreflectors, since πmax increases with π . In contrast, HAPSand LEO systems performance is affected by frequency, anddeeply by the atmospheric attenuation at specific frequency
LEO
Terrestrial
UAV
HAPS
Figure 9. Received power vs. carrier frequency (different platforms andenvironment types).
ranges. Nevertheless, some spectrum regions present a stablereceived power behaviour, such as below 40 GHz, and inthe 150-350 GHz and 500-700 GHz bands. The latter canbe fully exploited for high capacity communications. Finally,depending on the operating frequency band and environment(dense urban, urban, or rural), one system may be moresuitable than another (e.g., the RSS-equipped HAPS is themost interesting one in rural areas for most of the frequencybands).
C. Data Rate Evaluation and Impact of Platform Location
For the following simulations, we consider the same as-sumptions as in Fig. 7. Subsequently, (20), (39) and (49) canbe used to evaluate the received power. The related data rateis calculated as
R = π΅π€ log2
(1 + ππ
ππ
), (54)
where π΅π€ denotes the bandwidth, πΉ stands for the noise figure,and ππ refers to the noise power, given by
ππ = πΎππ΅π€πΉ, (55)
where πΎ = 1.38 Γ 10β23J.β¦Kβ1 is the Boltzmann constant andπ is the temperature in β¦K. According to [6], when aerialnetworks operate in frequency bands π β₯ 6 GHz, π΅π€ can beup to 800 MHz in both uplink and downlink, while πΉ = 7 dB.
Assuming that each platform uses πmax reflectors and thatπ΅π€ = 100 MHz, we depict in Fig. 10 the resulting data rates asa function of the receiver gain πΊπ . As typical UE has πΊπ β€ 5dBi, both terrestrial and HAPS systems equipped with RSScan directly support the downlink communication to users,and this is mainly due to their large RSS areas. However, RSS-equipped UAVs and LEO satellites are unable to match thisdue to the small RSS areas of UAVs and the high propagationloss of LEO satellites. Alternatively, mounting RSS over UAVsand LEO satellites may support inter-UAV and inter-satellitecommunications [23]. Also, a swarm of RSS-equipped UAVs
12
Terrestrial
HAPS
LEO
UAV
Figure 10. Data rate vs. receiver antenna gain.
can be utilized to assist communications cooperatively [1],[21].
The platform placement has an important impact on the datarate performance. Unlike LEO satellites, UAVs and HAPSnodes can be placed at fixed positions above an intendedcoverage area. However, due to wind and turbulence, theseplatforms may drift from their initial position, thus degradingcommunication performance. To assess such an effect, wepresent in Fig. 11 the data rate performance as a functionof a = 2 β π
π, the normalized horizontal Rx-RSS distance.
Moreover, we identify the optimal aerial platform location ascalculated by (32) and the one provided by the simulations.Here, we assume that πΊπ = 0 dBi, while the remainingparameters are as for Fig. 10.
As discussed above, the optimal placement of an RSS-equipped aerial platform depends on the latterβs altitude π»π ππ
and coverage radius π. For the RSS-equiped UAV, the obtainedoptimal UAV location πβ that achieves the highest data rate (reddot) agrees with that of (32) (blue circle), for any environmenttype. For the RSS-equipped HAPS, we distinguish betweentwo cases, namely for π = 50 km and π = 10 km. In bothcases, the optimal simulated HAPS locations and those of (32)agree in the rural environment, but the latter drift away asthe environment becomes urbanized. This is mainly due tothe effect of the additional shadowing and NLoS links in the3GPP model of urban environments, which were ignored inthe calculation of (32). Moreover, this location gap is largerfor π = 10 km due to a higher shadowing impact. For theterrestrial networks, the best location is either being the closestto Tx or Rx, with a preference for Tx. Indeed, since the BS(at altitude 25 m) has a strong LoS link towards the RSS, thereceived signal is slightly better than being the closest to Rx(at altitude 1.5 m).
When an RSS-equipped HAPS operates in a rural environ-ment, drifting from its initial location would have a smallimpact on communication performance. However, the datarate significantly degrades in an urban environment when theHAPS moves towards the middle of the Tx-Rx segment. A
HAPS (d=10 km)
HAPS(d=50 km)
Terrestrial
UAV
TxRx
Figure 11. Data rate vs. normalized Rx-RSS distance a (different platformsand environment types).
terrestrial RSS achieves the highest performance at optimallocations. However, due to the inflexibility of its deploymentand varying Tx and Rx locations, it may not perform at itsbest. In any case, when designed to assist cellular networks,RSS deployed closest to the BSs would eventually guaranteeoperating at near-optimal performance.
D. Outage Probability Analysis
Links between aerial platforms and terrestrial terminalsmay be subject to random variations due to shadowing andblockages. The instantaneous received power at a terrestrialterminal can be generally written as
ππ = ππ + π, (56)
where ππ is the average received power, defined as
ππ = ππ‘+πΊπ‘+πΊπβ(ππΏπ π₯βπ ππ+ππΏπ ππβπ π₯)+20 log(π), (57)
where ππΏπ π₯βπ ππ and ππΏπ ππβπ π₯ denote the average pathloss of links between the RSS-equipped aerial platform andthe transmitter and the receiver, respectively. Also, π rep-resents the resulting shadow fading of both links, modeledas a zero-mean normal distribution with a standard deviationππ =
βοΈπ2π π₯βπ ππ + π2
π ππβπ π₯. The probability density function
(pdf) of the received power can be written as
π (ππ ) =1
ππ
β2π
exp
(β (ππ β ππ )2
2π2π
), ππ β₯ 0. (58)
Accordingly, the outage probability can be obtained from thecumulative distribution function (cdf) as
Pout = P(ππ β€ π₯) =β« π₯
0π (ππ ) πππ = 1 β 1
2erfc
(π₯ β ππππ
β2
),
(59)where π₯ reflects the receiver sensitivity and erfc(π₯) =
2βπ
β« βπ₯
exp(βπ‘2) ππ‘ is the complementary error function.
13
-130 -120 -110 -100 -90 -80 -70 -60x [dBm]
0
10
20
30
40
50
60
70
80
90
100Pro
b(P
r5
x)[%
]
HAPS dense urbanHAPS urbanHAPS ruralUAV urbanUAV ruralTerrestrial
Figure 12. cdf of the received power at ground terminals assisted by RSS-equipped aerial platforms (different environment types).
Using the 3GPP models for the scattering reflectionparadigm (Table III) with the characteristics of aerial platforms(Table IV) and same assumptions as in Fig. 7, we depict inFig. 12 the cdf of the received power at user terminals with0 dB gain, and assisted by RSS mounted on different aerialplatforms in various environments. For any aerial platform, asthe receiverβs sensitivity degrades (i.e., π₯ becomes larger), theoutage probability increases. This performance degradation ismore significant in denser urban environments due to highershadowing loss. Nevertheless, the outage performance gapbetween rural and urban environments is more noticeable fora HAPS-assisted communication than for a UAV-assisted one.
The outage probability performance depends on the cover-age area of the aerial platform. To analyze the impact of thecoverage area, we illustrate in Fig. 13 the outage performanceas a function of the coverage radius. For these simulations,we assume a receiver power sensitivity of πth
π = β115 dBm.From Fig. 13, an outage probability lower than 10% requiresa coverage radius below 0.5 km for a terrestrial RSS andbelow 1 km for a UAV-assisted communication, respectively,whereas a HAPS-assisted communication can support an areaof between 40 km and 80 km in radius (β 80 times larger thanfor a terrestrial RSS). This demonstrates the high potential ofdeploying RSS on HAPS compared to the alternatives.
V. CONCLUSION
In this paper, we conducted a link budget analysis forthe envisioned RSS-equipped aerial platforms, namely UAVs,HAPS nodes, and LEO satellites. In a review of the literature,we identified two reflection paradigms: specular and scattering.For each reflection paradigm, we discussed its realizationconditions and then derived the associated optimal RSS-equipped platform location that maximizes the received power.Numerical results provided a number of insights into thedesign conditions of RSS-assisted communications:
10-1 100 101 102
Coverage Radius [km]
0
10
20
30
40
50
60
70
80
90
100
Outa
gepro
bab
ility
[%]
HAPS dense urbanHAPS urbanHAPS ruralUAV urbanUAV ruralTerrestrial
Figure 13. Outage probability vs. coverage radius for RSS-assisted commu-nications (different environment types).
β’ The scattering paradigm has gained more interest in theresearch community than the specular one due to thepractical accuracy of the former.
β’ RSS-equipped HAPS presents superior performance indifferent types of environments, compared to RSS-basedterrestrial and other aerial platforms communications.
β’ The received power performance is limited by the sizesof the RSS area and number of reflectors.
β’ When using the maximal number of reflectors for anRSS, the link budget of the scattering reflection paradigmbecomes independent from the carrier frequency.
β’ Supporting ground users with RSS-equipped UAVs andLEO satellites might not be feasible. Nevertheless, RSScan be used to assist inter-UAV or inter-LEO communica-tions. Also, a swarm of RSS-equipped UAVs can supportterrestrial users.
β’ The best RSS-equipped platform location depends greatlyon the operating altitude, coverage footprint, and environ-ment type.
β’ Finally, unlike other RSS-assisted communications, theones assisted by an RSS-equipped HAPS sustain the bestoutage probability/coverage performance in any type ofenvironment.
REFERENCES
[1] S. Alfattani, W. Jaafar, Y. Hmamouche, H. Yanikomeroglu, A. Yon-gaΓ§oglu, N. ΓΓ o, and P. Zhu, βAerial platforms with reconfigurable smartsurfaces for 5G and beyond,β IEEE Commun. Mag., vol. 59, no. 1, pp.96β102, Jan. 2021.
[2] G. Kurt, M. G. Khoshkholgh, S. Alfattani, A. Ibrahim, T. S. Darwish,M. S. Alam, H. Yanikomeroglu, and A. Yongacoglu, βA vision andframework for the high altitude platform station (HAPS) networks ofthe future,β IEEE Commun. Surveys. Tuts., vol. 23, no. 2, pp. 729β779,Secondquarter 2021.
[3] F. Rinaldi, H.-L. Maattanen, J. Torsner, S. Pizzi, S. Andreev, A. Iera,Y. Koucheryavy, and G. Araniti, βNon-terrestrial networks in 5G &beyond: A survey,β IEEE Access, vol. 8, pp. 165 178β165 200, 2020.
[4] O. Kodheli, E. Lagunas, N. Maturo, S. K. Sharma, B. Shankar, J. Mon-toya, J. Duncan, D. Spano, S. Chatzinotas, S. Kisseleff et al., βSatellitecommunications in the new space era: A survey and future challenges,β
14
IEEE Commun. Surveys Tuts, vol. 23, no. 1, pp. 70β109, Firstquarter2021.
[5] Y. Zeng, Q. Wu, and R. Zhang, βAccessing from the sky: A tutorial onUAV communications for 5G and beyond,β Proc. IEEE, vol. 107, no. 12,pp. 2327β2375, Dec. 2019.
[6] Technical Specification Group Radio Access Network; Study on NewRadio (NR) to Support Non-Terrestrial Networks, 3GPP TR 38.811V15.4.0, Sep. 2020.
[7] Technical Specification Group Services and System Aspects; Enhance-ment for Unmanned Aerial Vehicles, 3GPP TR 22.829 V17.1.0, Sep.2019.
[8] Technical Specification Group Services and System Aspects; UnmannedAerial System (UAS) Support in 3GPP, 3GPP TS 22.125 V17.2.0, Sep.2020.
[9] βStarlink.β [Online]. Available: https://www.starlink.com/[10] βStratobus: Why this stratospheric airship is already
being called a βswiss knifeβ in the sky.β [Online].Available: https://www.thalesgroup.com/en/worldwide/space/magazine/stratobus-why-stratospheric-airship-already-being-called-swiss-knife-sky
[11] βNokia drone networks.β [Online]. Available: https://dac.nokia.com/applications/nokia-drone-networks/
[12] M. Di Renzo, M. Debbah, D.-T. Phan-Huy, A. Zappone, M.-S.Alouini, C. Yuen, V. Sciancalepore, G. C. Alexandropoulos, J. Hoydis,H. Gacanin et al., βSmart radio environments empowered by reconfig-urable AI meta-surfaces: An idea whose time has come,β EURASIP J.Wireless Commun. Netw., vol. 129, no. 1, pp. 1β20, May 2019.
[13] C. Liaskos, S. Nie, A. Tsioliaridou, A. Pitsillides, S. Ioannidis, andI. Akyildiz, βA new wireless communication paradigm through software-controlled metasurfaces,β IEEE Commun. Mag., vol. 56, no. 9, pp. 162β169, Sep. 2018.
[14] E. Basar, M. Di Renzo, J. De Rosny, M. Debbah, M.-S. Alouini, andR. Zhang, βWireless communications through reconfigurable intelligentsurfaces,β IEEE Access, vol. 7, pp. 116 753β116 773, 2019.
[15] Q. Wu and R. Zhang, βIntelligent reflecting surface enhanced wirelessnetwork via joint active and passive beamforming,β IEEE Trans. WirelessCommun., vol. 18, no. 11, pp. 5394β5409, Nov. 2019.
[16] X. Tan, Z. Sun, D. Koutsonikolas, and J. M. Jornet, βEnabling IndoorMobile Millimeter-wave Networks Based on Smart Reflect-arrays,β inProc. IEEE Conf. Comput. Commun. (INFOCOM), Honolulu, HI, USA,2018.
[17] Q. Wu, S. Zhang, B. Zheng, C. You, and R. Zhang, βIntelligentreflecting surface aided wireless communications: A tutorial,β IEEETrans. Commun., vol. 69, no. 5, pp. 3313β3351, May 2021.
[18] S. Gong, X. Lu, D. T. Hoang, D. Niyato, L. Shu, D. I. Kim, and Y.-C.Liang, βToward smart wireless communications via intelligent reflectingsurfaces: A contemporary survey,β IEEE Commun. Surveys Tuts, vol. 22,no. 4, pp. 2283β2314, Fourthquarter 2020.
[19] H. Lu, Y. Zeng, S. Jin, and R. Zhang, βEnabling panoramic full-anglereflection via aerial intelligent reflecting surface,β in proc. 2020 IEEEInt. Conf. Commun. Wksh. (ICC Wksh.), Dublin, Ireland, Jun. 2020, pp.1β6.
[20] A. S. Abdalla, T. F. Rahman, and V. Marojevic, βUAVs with reconfig-urable intelligent surfaces: Applications, challenges, and opportunities,βarXiv preprint, [Online]. Available: https://arxiv.org/abs/2012.04775,Dec. 2020.
[21] B. Shang, R. Shafin, and L. Liu, βUAV swarm-enabled aerial re-configurable intelligent surface,β arXiv preprint, [online]. Available:https://arxiv.org/abs/2103.06361, 2021.
[22] M. Samir, M. Elhattab, C. Assi, S. Sharafeddine, and A. Ghrayeb,βOptimizing age of information through aerial reconfigurable intelligentsurfaces: A deep reinforcement learning approach,β IEEE Trans. Veh.Technol., vol. 70, no. 4, pp. 3978β3983, Apr. 2021.
[23] K. Tekbiyik, G. K. Kurt, A. Ekti, A. GΓΆrcin, and H. Yanikomeroglu,βReconfigurable intelligent surface empowered terahertz communica-tion for LEO satellite networks,β arXiv preprint, [Online]. Available:https://arxiv.org/abs/2007.04281, Dec. 2020.
[24] W. Tang, M. Z. Chen, X. Chen, J. Y. Dai, Y. Han, M. Di Renzo,Y. Zeng, S. Jin, Q. Cheng, and T. J. Cui, βWireless communications withreconfigurable intelligent surface: path loss modeling and experimentalmeasurement,β IEEE Trans. Wireless Commun., vol. 20, no. 1, pp. 421β439, Jan. 2021.
[25] I. Yildirim, A. Uyrus, and E. Basar, βModeling and analysis of reconfig-urable intelligent surfaces for indoor and outdoor applications in futurewireless networks,β IEEE Trans. Commun., vol. 69, no. 2, pp. 1290β1301, Feb. 2021.
[26] O. Γzdogan, E. BjΓΆrnson, and E. G. Larsson, βIntelligent reflectingsurfaces: physics, propagation, and pathloss modeling,β IEEE WirelessCommun. Lett., vol. 9, no. 5, pp. 581β585, May 2020.
[27] Q.-U.-A. Nadeem, A. Kammoun, A. Chaaban, M. Debbah, and M.-S.Alouini, βIntelligent reflecting surface assisted wireless communication:Modeling and channel estimation,β arXiv preprint, [Online]. Available:https://arxiv.org/abs/1906.02360, Dec. 2019.
[28] M. Di Renzo, K. Ntontin, J. Song, F. H. Danufane, X. Qian, F. Lazarakis,J. de Rosny, D. T. Phan-Huy, O. Simeone, R. Zhang, M. Debbah,G. Lerosey, M. Fink, S. Tretyakov, and S. Shamai, βReconfigurable in-telligent surfaces vs. relaying: Differences, similarities, and performancecomparison,β IEEE Open J. Commun. Soc., vol. 1, p. 798β807, 2020.
[29] E. Basar and I. Yildirim, βSimris channel simulator for reconfigurableintelligent surface-empowered communication systems,β in proc. 2020IEEE Latin-American Conf. Commun. (LATINCOM), Santo Domingo,Dominican Republic, Nov. 2020, pp. 1β6.
[30] S. W. Ellingson, βPath loss in reconfigurable intelligentsurface-enabled channels,β arXiv preprint, [Online]. Available:https://arxiv.org/abs/1912.06759, Dec. 2019.
[31] J. C. B. Garcia, A. Sibille, and M. Kamoun, βReconfigurable intelligentsurfaces: Bridging the gap between scattering and reflection,β IEEE J.Sel. Areas Commun., vol. 38, no. 11, pp. 2538β2547, Nov. 2020.
[32] E. BjΓΆrnson, H. Wymeersch, B. Matthiesen, P. Popovski, L. Sanguinetti,and E. de Carvalho, βReconfigurable intelligent surfaces: A signal pro-cessing perspective with wireless applications,β arXiv preprint, [online].Available: https://arxiv.org/abs/2102.00742, 2021.
[33] S. V. Hum and J. Perruisseau-Carrier, βReconfigurable reflectarrays andarray lenses for dynamic antenna beam Control: A review,β IEEE Trans.Antennas Propag., vol. 62, no. 1, pp. 183β198, Jan. 2014.
[34] Q. Wu and R. Zhang, βBeamforming optimization for wireless networkaided by intelligent reflecting surface with discrete phase shifts,β IEEETrans. Commun., vol. 68, no. 3, pp. 1838β1851, Mar. 2020.
[35] C. Huang, G. C. Alexandropoulos, A. Zappone, M. Debbah, andC. Yuen, βEnergy efficient multi-user miso communication using lowresolution large intelligent surfaces,β in IEEE Globecom Workshops (GCWkshps). IEEE, 2018, pp. 1β6.
[36] T. S. Rappaport, Wireless Communications: Principles and Practice.Prentice Hall PTR, 1996.
[37] E. BjΓΆrnson, O. Γzdogan, and E. G. Larsson, βReconfigurable intelligentsurfaces: Three myths and two critical questions,β IEEE Commun. Mag.,vol. 58, no. 12, pp. 90β96, Dec. 2020.
[38] Study on Channel Model for Frequencies from 0.5 to 100 GHz, 3GPPTR 38.901 V14.3.0, Jan. 2018.
[39] Prediction of Building Entry Loss, ITU-R P.2109-1 P Series, Aug. 2019.[40] Study on Enhanced LTE Support for Aerial Vehicles, 3GPP TR 36.777
V1.0.0, Dec. 2017.[41] M. Alzenad, A. El-Keyi, F. Lagum, and H. Yanikomeroglu, β3-D place-
ment of an unmanned aerial vehicle base station (UAV-BS) for energy-efficient maximal coverage,β IEEE Wireless Commun. Lett., vol. 6, no. 4,pp. 434β437, Aug. 2017.
[42] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, and M. Debbah, βAtutorial on UAVs for wireless networks: Applications, challenges, andopen problems,β IEEE Commun. Surveys. Tuts., vol. 21, no. 3, pp. 2334β2360, thirdquarter 2019.
[43] M. S. Alam, G. K. Kurt, H. Yanikomeroglu, P. Zhu, and N. D.ΓΓ o, βHigh altitude platform station based super macro base stationconstellations,β IEEE Commun. Mag., vol. 59, no. 1, pp. 103β109, Jan.2021.
[44] Preferred Characteristics of Systems in the Fixed Service Using HighAltitude Platforms Operating in the Bands 47.2-47.5 GHz and 47.9-48.2GHz, ITU-R F.1500, May 2000.
[45] C. E. Fossa, R. A. Raines, G. H. Gunsch, and M. A. Temple, βAnoverview of the Iridium (R) low earth orbit (LEO) satellite system,β inProc. IEEE Nat. Aerospace Electron. Conf. (NAECON), Dayton, OH,USA, 1998, pp. 152β159.
[46] S. Cakaj, βThe coverage belt for low earth orbiting satellites,β in Proc.39th Int. Convent. Info. Commun. Technol., Electron. Microelectron.(ICICTEM), Opatija, Croatia, May 2016, pp. 554β557.
[47] Y. Su, Y. Liu, Y. Zhou, J. Yuan, H. Cao, and J. Shi, βBroadband LEOsatellite communications: Architectures and key technologies,β IEEEWireless Commun., vol. 26, no. 2, pp. 55β61, Apr. 2019.
[48] βHosting payloads on a communication constellation.β [Online].Available: https://directory.eoportal.org/web/eoportal/satellite-missions/i/iridium-next
[49] βStarlink satellite dimension estimates.β [Online]. Available: https://lilibots.blogspot.com/2020/04/starlink-satellite-dimension-estimates.html
15
[50] Attenuation by Atmospheric Gases, ITU-R P.676-10, Sep. 2013.[51] Reference Standard Atmospheres, ITU-R P.835-6 P Series, Dec. 2017.[52] Ionospheric Propagation Data and Prediction Methods Required for the
Design of Satellite Networks and Systems, ITU-R P.531-14 P Series,Aug. 2019.
[53] Propagation Data and Prediction Methods Required for the Design ofEarth-space Telecommunication Systems, ITU-R P.618-13 P Series, Dec.2017.