1

Exploiting Active RIS in NOMA Networks with

Hardware Impairments
Xinwei Yue, Senior Member, IEEE, Meiqi Song, Chongjun Ouyang, Yuanwei Liu, Fellow, IEEE, Tian
Li, Member, IEEE and Tianwei Hou, Member, IEEE

Abstract—Active reconfigurable intelligent surface (ARIS) is higher data rate, massive connections, low latency, as well
arXiv:2311.14295v1 [cs.IT] 24 Nov 2023

a promising way to compensate for multiplicative fading at- as higher full coverage, and so on. Non-orthogonal multiple
tenuation by amplifying and reflecting event signals to selected access (NOMA) has been referred to as one of the physical
users. This paper investigates the performance of ARIS assisted
non-orthogonal multiple access (NOMA) networks over cascaded layer technologies [5, 6], which solves the problems of limited
Nakagami-m fading channels. The effects of hardware impair- access for multiple users from the views of information-
ments (HIS) and reflection coefficients on ARIS-NOMA networks theoretic. Furthermore, the integration of NOMA with emerg-
with imperfect successive interference cancellation (ipSIC) and ing wireless techniques was explored towards 6G networks [7].
perfect successive interference cancellation (pSIC) are consid- Another noteworthy technique, i.e., reconfigurable intelligent
ered. More specifically, we develop new precise and asymptotic
expressions of outage probability and ergodic data rate with surface (RIS) has sparked the enthusiasm of research from
ipSIC/pSIC for ARIS-NOMA-HIS networks. According to the both academic and industrial communities [8, 9]. It has been
approximated analyses, the diversity orders and multiplexing confirmed to accomplish the control of incident electromag-
gains for couple of non-orthogonal users are attained in detail. netic waves through the programmable device.
Additionally, the energy efficiency of ARIS-NOMA-HIS networks In particular, the RIS is generally arranged by a mount of
is surveyed in delay-limited and delay-tolerant transmission
schemes. The simulation findings are presented to demonstrate passive electromagnetic units designed carefully, which can
that: i) The outage behaviors and ergodic data rates of ARIS- adapt to changes in the environment [10–12]. Up to now,
NOMA-HIS networks precede that of ARIS aided orthogonal numerous treatises have introduced this type of passive RIS
multiple access (OMA) and passive reconfigurable intelligent (PRIS) into wireless communication networks. In [13], a new
surface (PRIS) aided OMA; ii) As the reflection coefficient of design of energy efficiency for PRIS aided multi-user networks
ARIS increases, ARIS-NOMA-HIS networks have the ability to
provide the strengthened outage performance; and iii) ARIS- was highlighted by taking the realistic power consumption
NOMA-HIS networks are more energy efficient than ARIS/PRIS- into account. Furthermore, the effect of phase shifting error
OMA networks and conventional cooperative schemes. on outage behaviors was evaluated for PRIS based wireless
networks [14]. From the practical viewpoint, the authors in
Index terms— Active reconfigurable intelligent surface,
[15] researched the erogdic throughput of PRIS with imperfect
non-orthogonal multiple access, outage behavior, ergodic data
hardware conditions. The deployment issue of PRIS was
rate, energy efficiency.
discussed in [16], where the reflecting probability of PRIS
enabled networks was investigated by exploiting stochastic
I. I NTRODUCTION
geometry. To shed light on the overlay mode, the authors of
As the diversification of business demand increases, wireless [17] applied PRIS to device-to-device communications under
communications are currently confronted with unpredictable Nakagami-m fading channels. With economizing the energy
challenges [2, 3]. It is forecasted that ten thousand mobile users consumption [18], the average error probability of PRIS based
and smart equipment will be linked to the networks by the end wireless power transfer (WPT) was surveyed comprehensively.
of 2022 [4]. The aims of the sixth-generation (6G) networks Obviously, combining PRIS and NOMA can effectively
are predicted to support the key performance indexes, i.e., realize the enhanced spectrum efficiency and dynamic envi-
X. Yue and M. Song are with the Key Laboratory of Information and ronment configuration for 6G networks [19–21]. In [22], the
Communication Systems, Ministry of Information Industry and also with the attractive advantages of applying PRIS to NOMA networks
Key Laboratory of Modern Measurement & Control Technology, Ministry of were highlighted from the viewpoint of user fairness and
Education, Beijing Information Science and Technology University, Beijing
100101, China (email: {xinwei.yue and meiqi.song}@bistu.edu.cn). system scalability. As further progress, the authors in [23]
C. Ouyang is with the School of Electrical and Electronic Engineer- revealed the superiorities of PRIS-NOMA relative to PRIS-
ing, University College Dublin, Dublin, D04 V1W8, Ireland (e-mail: orthogonal multiple access (OMA) from the viewpoint of
[email protected]).
Y. Liu is with the School of Electronic Engineering and Computer Sci- different rate configurations. By using on-off control strategy
ence, Queen Mary University of London, London E1 4NS, U.K. (email: [24], the effect of hardware impairments (HIS) on PRIS-
[email protected]). NOMA networks performance was evaluated carefully. The
T. Li is with the 54th Research Institute of China Electronics Technology
Group Corporation, Shijiazhuang Hebei 050081, China. (email: [email protected]). authors of [25] analyzed the PRIS-NOMA networks’ user
T. Hou is with the School of Electronic and Information Engineering, Bei- prioritization outage characteristics. Furthermore, the authors
jing Jiaotong University, Beijing 100044, China, and also with the Institute for of [26] investigated ergodic data rate and energy efficiency of
Digital Communications, Friedrich-Alexander Universität Erlangen-Nürnberg
(FAU), 91054 Erlangen, Germany (email: [email protected]). Part of this PRIS-NOMA by using suboptimal scheme. Conditioned on
work has been submitted to IEEE VTC 2023 [1]. imperfect detection conditions, the couple error probability
 2

of large PRIS-NOMA networks was outlined in [27]. The is no detailed survey of the outage probability and ergodic data
adoption of PRIS to help two-way NOMA communications rate of network performance. In addition, most of the known
was highlighted in [28], where the message was interchanged studies have been analyzed under ideal hardware conditions,
between user nodes. In addition, the authors of [29] discussed whereas HIS usually have a non-negligible impact in practical
the effect of imperfect HIS on outage performance for PRIS- applications [44, 45]. Hence it is also worth considering how
NOMA. The ergodic capacity of PRIS-NOMA was elucidated HIS affects ARIS-NOMA networks’ performance. In practice,
by taking into consideration the HIS [30]. the successive interference cancellation (SIC) process can lead
Except that the RIS works in the passive mode, it can to decoding errors due to factors such as instrumentation errors
also operate in the active mode by integrating an additional and incorrect propagation. Sparked by the above discussions,
low-power amplifier on each reflecting element [31, 32]. The we concentrate on how ARIS-NOMA networks with HIS
type of active RIS (ARIS) has aroused widespread attention perform over cascade Nakagami-m fading channels where
from both academic and industrial circles [33–35]. ARIS is imperfect successive interference cancellation (ipSIC) results
able to increase system capacity by controlling the phase and in residual interference. More precisely, new exact and approx-
amplitude of reflected signal. More particularly, the authors imative expressions of outage probability and ergodic data rate
in [33] presented a novel amplifier based metasurface to with ipSIC/perfect SIC (pSIC) are deduced of ARIS-NOMA
accomplish the frequency and space domain control of inci- networks. The influence of HIS on ARIS-NOMA networks
dent electromagnetic waves. In [34], the multiplicative fading performance is taken into consideration. The power consump-
influence caused by PRIS was relieved as a sacrifice of power tion of ARIS-NOMA-HIS networks is further highlighted on
dissipation on the active reflection element. Even though ARIS the condition of delay-limited and delay-tolerated transmission
integrates a low-power amplifier at a slightly higher cost schemes. Depending on these specific works, the foremost
than PRIS, it is still less expensive than other repeaters (e.g., contributions of this paper are briefly summed up as follows:
amplify-and-forward (AF) relay and decode-and-forward (DF)
relay). With the emphasis on this issue, the author of [35] 1) We introduce the application of ARIS to NOMA net-
proved that ARIS is capable of turning the double-fading path works in the presence of HIS. We derive the exact
loss of PRIS into the additive form. As a further step, the sub- and asymptotic expressions of outage probability with
connected structure of ARIS was presented to regulate each ipSIC/pSIC for ARIS-NOMA-HIS networks over cas-
element’s phase shift independently [36]. The exploitation of cade Nakagami-m fading channels. Utilizing theoretical
ARIS to wireless communications was outlined in [37], where analysis, we further acquire diversity orders of ARIS-
the ARIS-assisted links were strengthened with the increase NOMA-HIS networks, which are related to both chan-
of active elements. The bit error probability of ARIS based nel ordering and the number of reflective components.
wireless systems was evaluated thoroughly [38]. On closer We derive the outage probability for ARIS-OMA-HIS
inspection, the authors of [39] identified that ARIS should networks as a benchmark for comparison.
be arranged around users with the small amplification power. 2) We contrast ARIS-NOMA-HIS and PRIS-NOMA-HIS
In [40], the achievable rate of ARIS exceeded that of PRIS with that of conventional DF and AF relays in terms
on the condition of same power budget. In contrast to PRIS of outage behaviors. We further demonstrate that ARIS-
[41], the ARIS-WPT systems had the potential to actualize the NOMA-HIS has a higher outage probability than tradi-
enhanced energy-efficiency. On this basis, the authors in [42] tional cooperative communication systems. We regulate
further discussed the system throughput of ARIS aided uplink the distances between BS and ARIS, where deploying
NOMA communication. ARIS on BS side results in better outage behaviors.
By adding ARIS elements number, we find that the
A. Motivation and Contributions outage performance of ARIS first gets improved and
The previous research treatises have supplied a compre- then deteriorates.
hensive viewpoint that applying PRIS in NOMA networks 3) We derive the ergodic data rate of nearby and dis-
can further enhance spectrum efficiency and cell coverage.As tant users with regard to ARIS-NOMA-HIS and PRIS-
previously stated in [34, 35], the multiplicative fading influence NOMA-HIS networks, respectively. We note that users’
is inherent in PRIS-assisted wireless communications, where ergodic data rate converges to the throughput ceiling
the system performance is affected by the double path loss at at high signal-to-noise ratio (SNR). As the reflection
longer distances. To solve this problem, it may be necessary amplification factor increases, ARIS-NOMA-HIS has
to install a large number of passive reflective elements or to the ability to deliver an improved ergodic data rate.
place the PRIS close to the transmitter or receiver. However We derive the ergodic data rate of ARIS-OMA-HIS
this approach is not feasible in practice.The appearance of networks, and verify that ARIS-NOMA-HIS performs
ARIS’s studies focuses on solving the problem of double is superior to ARIS-OMA-HIS in terms of ergodic data
path loss brought by PRIS. Although previous work has rate.
validated the advantages of applying PRIS to NOMA [43], 4) We assess the system throughput of ARIS-NOMA-HIS
treatises on the potential performance gains by integrating and PRIS-NOMA-HIS networks under delay-limited and
ARIS with NOMA are still in their infancy. In [41], the delay-tolerant transmission schemes. In delay-limited
throughput maximization of ARIS-NOMA was discussed un- schemes, we compare the system throughput between
der the energy-limited communication scenarios, while there ARIS-NOMA-HIS and PRIS-NOMA-HIS networks. In
 3

ARIS closely match practical scenarios. To capture real channel

reflection conditions, we assume that the wireless links of ARIS-NOMA
amplifier phase-
shift patch
networks go through large-scale fading and Nakagami-m
√ − α2  1
circuit
H
hbr fading1. Let hbr = ηdbr hbr , · · · , hlbr , · · · , hL
br and
√ − α   H
hri = ηdri 2 h1ri , · · · , hlri , · · · , hL
ri denote the baseband
hrf
hrn
channel coefficients between BS and ARIS, and between ARIS
and user i (1 ≤ i ≤ K). hlbr and hlri denote the complex chan-
nel factors between BS and ARIS’s l-th reflective element, and
ĂĂ
Obstacle Userg User f between ARIS’s l-th reflective element and user i, respectively.
BS dbr and dri are the distances of BS to ARIS and ARIS to
user i, respectively. α is the path loss exponent, while η is
Fig. 1: System model of ARIS-assisted NOMA-HIS communi- the frequency dependent
√ factor. The reflection matrix
at ARIS
cations, where ARIS is capable of amplifying the superposed is defined as Φ = βdiag ejθ1 , · · · , ejθl , · · · , ejθL , where
signals, and then reflecting to users. θl ∈ [0, 2π) and β indicate the l-th reflection element’s phase
shift and reflection amplification coefficient, respectively. The
reflection elements of ARIS are embedded in active amplifiers
delay-tolerant schemes, we compare the rates of ARIS- that have tunneling diodes or resistive converters. Based on
NOMA-HIS and PRIS-NOMA-HIS and found that both the principle of electromagnetic scattering, each element can
reached an upper bound at high SNRs. We also compare reflect and amplify an incoming radio frequency (RF) signal
the energy efficiency of the two modes and find that with an adjustable amplitude and phase. This allows ARIS to
ARIS-NOMA-HIS networks perform much better than achieve reflection amplification coefficient greater than one,
the other comparative benchmarks, which have a high i.e., β > 1 [34, 37]. In practice, a negative resistive component
level of stability. such as a tunnel diode can be used to amplify the incident
signal by converting the direct current bias power to RF power.
B. Organization and Notations Broadly speaking, the baseband cascade channel gains trans-
The rest of this paper is structured as follows. Section mitted by BS towards ARIS, and thereafter to the considerable
2 2
II introduces the network model of ARIS-NOMA-HIS. The users are sorted as hH rK Φhbr > · · · > hH rg Φhbr > ··· >
2 2
exact and approximate outage probabilities of ARIS-NOMA- hHrf Φhbr > · · · > hHr1 Φhbr . Note that the positions of
HIS networks are highlighted prudentially in Section III. BS and ARIS are usually fixed in real scenarios, the effect
Section IV evaluate the ergodic data rates for ARIS-NOMA- of large-scale fading is essentially constant. Furthermore,
HIS networks, and then present the numerical analyses in the signal undergoes the same attenuation and phase shift
Section VI. Finally, Section VII summarizes this manuscript. modulation and is then reflected back to users by ARIS. This
The proofs of the maths are gathered in the appendix. means that the difference in the signal-to-noise ratio (SINR)
The key notations in this paper are presented as follows. received by users is mainly caused by the large-scale fading in
H
The symbol (·) represents conjugate transpose operation. hri , which are independent of each other [47]. In addition, the
The cumulative distribution function (CDF) and probability hardware characteristics of receivers in a real situation can also
density function (PDF) of random variable X are denoted have an impact on SINR, and these hardware characteristics
by FX (·) and fX (·), respectively. E{·} and D{·} denote can also be considered as independent variables. Therefore, in
the expectation and variance operations, respectively. diag (·) the following steps, the marginal distribution can be utilized
denotes the diagonal matrix with element one. to further calculate user’s outage probability and ergodic data
rata. In particular, a couple of users, i.e., the near-end user
II. N ETWORK M ODEL g and far-end user f are picked out to carry out the NOMA
We consider an ARIS-aided NOMA communication sce- mechanism. We suppose that the instantaneous channel state
nario with the presence of HIS, as shown in Fig. 1, in which information (CSI) is accessible on BS and ARIS with help
the single antenna BS servers K single antenna users by means of channel evaluation or compressive sensing algorithms [48,
of L active reflecting elements. Specifically, existing ARIS 49]. To release this ideal assumption, the impact of imperfect
architectures not only have the same circuitry to control phase CSI on system performance for ARIS-NOMA networks will
shift as PRIS, but also integrate power amplifiers to amplify be discussed in simulation results part.
the radiated signal. The incident superposed signals at ARIS
are amplified with a visible gain and then reflected toward A. Signal Model
the desired users. The inherent multiplicative attenuation in
PRIS-assisted systems will be effectively tackled by active pThe BS p transfers superimposed signals, i.e., x =
elements for the price of low power consumption. In actuality, ag Pb xg + af Pb xf to user φ via ARIS, where xg and
we focus attention on that BS-users’ direct links obstructed by xf are normalized the unity power signals for the user g
high-rise buildings [10, 46], where the deployment of ARIS 1 It is noteworthy that the derivations of performance analyses for ARIS-
can provide the line-of-sight across BS and non-orthogonal aided NOMA networks can be easily applied to multiple fading channels, i.e.,
users. BS and users’ side distortions are considered to more the Gaussian, Reyleigh and Rician fading channels.
 4

√ −α  H
and the user f , respectively. To warrant non-orthogonal users’ where hro = ηdro2 h1ro , · · · , hlro , · · · , hL
ro denotes the
fairness, the power allocation factors ag and af for user g and complex channel coefficient between ARIS and user o. dro
is
user f satisfy the relations ag < af and ag + af = 1. Pb the distance between ARIS and user o and χo = κ2b + κ2o .
denotes the BS transmission power. Unlike PRIS, thanks to
the use of active elements like tunnel or Gunn diodes, ARIS III. O UTAGE PROBABILITY
will amplify the superimposed signals and the noise received
at itself. Within these contexts, the received signals at user φ In this section, we evaluate ARIS-NOMA/OMA-HIS net-
(φ ∈ {g, f }) can be expressed as works performance based on outage behaviour, where the
p p H approximate outage probability formulas for user g and user f
yφ = βhH rφ Θhbr (x + εb ) + βhrφ Θnr + εφ + nφ , (1) are derived over cascade Nakagami-m fading channels. Addi-

tionally, we approximate the thermal noise caused by ARIS
where Θ = diag ejθ1 , · · · , ejθl , · · · , ejθL denotes the phase
as a constant for the purpose of facilitating computational
shift matrix at ARIS. εb and εφ indicate the distortion
analysis [50, 51].
from HISat BS and user φ with  εb ∼ CN 0, κ2b Pb and
2
εφ ∼ CN 0, κ2φ Pb hH br Θhrφ [44]. The factors κb and κφ
A. The Outage Probability of User g
indicate the level of HIS, which can be quantified by the size
of the error vector. nr ∼ CN (0, Ntn IL ) denotes the thermal By the sequence of the user decoding order in SIC progress,
the outage incidents of user g are defined as that: 1) User g
noise at ARIS with the noise power Ntn , and IL ∈ CL×1 cannot detect user f ’s signal xf ; 2) User g can decode xf , but
is the identity matrix. nφ indicates the Gaussian
white noise cannot decode its own signal xg . Its supplementary event can
generated at the user φ with nφ ∼ CN 0, σ 2 . be easily represented that user g is able to detect the signal
Based on NOMA principle, user f ’s information will be first xf , and then decode the information xg [52, 53]. Under these
detected by user g with better channel conditions, then subtract circumstances, the outage probability of user g with ipSIC for
ARIS-NOMA-HIS networks is given by
it before decoding its own signal. Thus the corresponding
HIS


received SINR can be written as PARIS,g = Pr γg→f < γthf + Pr γg→f > γthf , γg < γthg ,
2 (6)
βPb hH
rg Θhbr af
γg→f = , (2)
βPb hH
2 H 2
+ σ2 where γthg = 2Rg − 1 and γthf = 2Rf − 1 denote the target
rg Θhbr χ + ξβNtn hrg Θ
SNRs of signals xg and xf detected by user g and user f ,
where χ = ag + κ2b + κ2g and ξ is the convert parameter. In respectively. Rg and Rf are the target rates of user g and user
ARIS networks ξ is set to 1. More precisely, when ξ is equal to f , respectively. In this case, the approximate outage probability
zero, ARIS will be divided into PRIS. After weeding out the for user g is able to be given in the following theorem.
information of user f , the receiving SINR of user g decoding
its own signal can be given by Theorem 1. In terms of cascade Nakagami-m fading chan-
nels, the outage probability of user g with ipSIC for ARIS-
2
βPb hH rg Θhbr ag NOMA-HIS networks can be approximated as
γg = 2 2 2 ,
H
βPb hrg Θhbr χg + ̟Pb |hI | + ξβNtn hrg Θ + σ H 2 K−g
HIS,ipSIC
X K − g  (−1)k X U
(3) PARIS,g ≈Ψg Hu xm u
g −1
k (g + k) ψg u=1
k=0
where χg = κ2b + κ2g and ̟ = 1 indicates the ipSIC case. " p !#g+k
hI ∼ CN (0, ΩI ) donotes the residual interference from ipSIC. ϕg xu + ϑg
× γ bg + 1, p , (7)
Since the channel conditions of user f are poor, it treats xg cg βmg
as interference and thus directly detects its own signal xf . For
K! g+k
the time being, the SINR for user f detects xf can be written where Ψg = (K−g)!(g−1)! , ψg = Γ (mg ) [Γ (bg + 1)]
,
2
as ϕg = dα α α α
br drg ςg ̟Pb , ϑg = mg dbr drg ςg ξβNtn Lωrg + σ ,
γthg
2 ςg = P a −γ χ2 with the condition of ag > γthg χ2g ,
βPb hH Θh br a f b( g thg g )
rf
γf = , (4) Lµ2g Ω Γ(mr + 12 )Γ(mg + 12 )
2 2 bg = Ωg − 1, cg = µgg , µg = Γ(mr )Γ(m √
g ) mr mg
,
βPb hH rf Θhbr χf + ξβNtn hrf Θ
H + σ2  2
Γ(mr + 12 )Γ(mg + 12 )
  Ωg = 1 − mr1mg Γ(mr )Γ(mg ) , mr and mg denote
where χf = ag + κ2b + κ2f .
the Nakagami-m fading parameters from the BS to ARIS, and
from ARIS to non-orthogonal user g, respectively. Γ (·) is the
B. ARIS-OMA with Hardware Impairments gamma function [54, Eq. (8.310.1)]. Hu and xu represent
With the aim of display the performance advantages of the Gauss-Laguerre integration’s weight and abscissas, sep-
ARIS-NOMA, we utilize the ARIS-OMA with HIS as a arately. Particularly, xu is the u-th zero
point of Laguerre
ex d U −x
benchmark for comparison. Based on the previous subsection polynomial L U (x) = U! dx U x e and the u-th weight
(U!)2 xu
the SINR of user o is given by can be denoted by Hu = [L (x )]2 . U is a tradeoff
U +1 u

H 2
Rparametric to certify the complexity-accuracy. γ (a, x) =
βPb hro Θhbr x a−1 −t
γo = 2 , (5) 0
t e dt denotes the lower incomplete gamma function
βPb |hH Θh | χ + ξβN kh H Θk2 + σ 2
ro br o tn ro [54, Eq. (8.350.1)].
 5

 2
Proof. See Appendix A. 1 Γ(mr + 21 )Γ(mf + 12 )
mr mf Γ(mr )Γ(mf ) , mf denote the multipath fading
Corollary 1. When the parameter ̟ is set to be zero, the parameter from ARIS to non-orthogonal user f .
closed-form outage probability of user g with pSIC for ARIS-
NOMA-HIS can be given by Proof. See Appendix B.
K−g
X K − g (−1)k 
HIS,pSIC 1 Corollary 4. When the parameter ξ is set to be zero, the
PARIS,g =Ψg closed-form outage probability of user f for PRIS-NOMA-HIS
k g + k Γ (bg + 1)
k=0 can be given by
p !#g+k
ϑg,pSIC K−f  k 
× γ bg + 1, √ , (8) X K − f (−1) 1
cg β PPHIS
RIS,f =Ψf
k f + k Γ (bf + 1)
2

k=0
where ϑg,pSIC = dα α
br drg ςg ξβNtn Lωrg + σ .  q f +k
dα dα ς σ2
br rf f
Corollary 2. When the parameter ξ is set to be zero, the × γ bf + 1,  . (13)
outage probability of user g with ipSIC for PRIS-NOMA-HIS cf
is approximated as
K−g
X K − g  (−1)k X U
C. The Outage Probability of User o
PPHIS,ipSIC
RIS,g ≈Ψ g Hu xm
u
g −1
k (g + k) ψg u=1 Analogous to the above analytical progresses, the outage
k=0
" p !#g+k probability of orthogonal user o for ARIS-OMA-HIS networks
ϕg xu + ϑP RIS,g
× γ bg + 1, √ , is given by
cg mg
HIS
(9) PARIS,o = Pr (γo < γtho ) , (14)
2
where ϑP RIS,g = mg dα α
br drg ςg σ . where γtho denotes the target SNRs of user o.
Corollary 3. When the parameter ̟ and ξ are both set to be Theorem 3. Conditioned on cascade Nakagami-m fading
zero, the closed-form outage probability of user g with pSIC channels, the outage probability of user o for ARIS-OMA-HIS
for PRIS-NOMA-HIS can be written as networks can be expressed as
K−g
X K − g (−1)k   √ 
HIS,pSIC 1 HIS 1 ϕo + ϑo
PP RIS,g =Ψg PARIS,o = γ bo + 1, √ , (15)
k g + k Γ (bg + 1) Γ (bo + 1) co β
k=0
 q g+k 2
dα α 2 where ϕo = ξβNtn dα α α α
br dro ςo Lωro , ϑo = dbr dro ςo σ , ςo =
br drg ςg σ γtho
× γ bg + 1, . (10) Pb (1−γtho χo ) with the condition of γtho χo < 1, and mo denote

cg the multipath fading parameter from ARIS to orthogonal user
o.
B. The Outage Probability of User f Corollary 5. When the parameter ξ is set to be zero, the
With respect to user f with worse channel conditions, the closed-form outage probability of user o for PRIS-OMA-HIS
interrupt is generated if it cannot detect or decode its own can be expressed as
message xf . In this case, the outage probability of user f for p α !
dbr dα 2
ARIS-NOMA-HIS networks is given by HIS 1 ro ςo σ
PP RIS,o = γ bo + 1, . (16)
HIS

Γ (bo + 1) co
PARIS,f = Pr γf < γthf . (11)
Theorem 2. Conditioned on cascade Nakagami-m fading
D. Diversity Analysis
channels, the closed-form outage probability of user f for
ARIS-NOMA-HIS networks can be expressed as In this section, we pick the diversity order to analyze the
K−f outage behaviors for ARIS-NOMA-HIS networks. To put it in
X K − f  (−1)k  1
HIS
PARIS,f =Ψf another way, the diversity order can describe the rate at which
k f + k Γ (bf + 1) the probability of disruption decreases as SNR increases [55].
k=0
p !#f +k As the grading order becomes boosted, the outage probability
ϕf + ϑf decays faster. To be more specific, the diversity order is defined
× γ bf + 1, √ , (12)
cf β as

where Ψf = K! α α log (Pr (ρ))

(K−f)!(f −1)! , ϕf = ξβNtn dbr drf ςf Lωrf , d = − lim , (17)
α α 2 γthf ρ→∞ log (ρ)
ϑf = dbr drf ςf σ , ςf = P a −γ χ with the condition
b( f thf f )

of af > γthf χf , ψf = Γ (mf ) [Γ (bf + 1)]

f +k
, bf = where ρ = σP2b means the transmit SNR. Pr (ρ) represents the
Lµ2f Ωf Γ( r 2) ( f 2)
m + 1
Γ m + 1 asymptotic outage probability in case of high SNRs.
Ωf − 1, cf = µf , µf = Γ(mr )Γ(mf ) mr mf , Ωf = 1 −
√
 6

Corollary 6. When Pb approaches infinity, the asymptotic Corollary 8. When Pb approaches infinity, the asymptotic
outage probability of user g with ipSIC/pSIC for ARIS-NOMA- outage probability of user f for ARIS/PRIS-NOMA-HIS can
HIS can be respectively expressed as be respectively given by
K−g
X 
(−1)k K−f
  k
ipSIC,∞ K −g HIS,∞ K − f (−1)
X
PARIS,g ≈Ψg = Ψf
PARIS,f
k (g + k) ψg k f +k
k=0 k=0
U
" √ !#g+k " # f +k
X
mg −1
ϕg xu  Γ (2m ) Γ (2m ) Υ L (ϕf + ϑf )Lmr 
× Hu xu γ bg + 1, p , ×
r f f

,
u=1
cg βmg  Λf Γ mr + mf + 12 2Lmr β Lmr (2Lmr − 1)! 
(18)
(22)
and
and
K−g
X 
pSIC,∞ K − g (−1)k  K−f 
PARIS,g = Ψg K − f (−1)
X k
k g+k PPHIS,∞
k=0 RIS,f = Ψf
" g+k k f +k
# k=0
 Γ (2m ) Γ (2m ) Υ L (ϑ Lmr −Lmr 
g,pSIC ) β Lmr f +k

×
r g g
, " # 
 Λg Γ mr + mg + 21

 Γ (2m ) Γ (2m ) Υ L dα
 dα
ς σ 2 

2Lmr (2Lmr − 1)!  r f f br rf f
× 1

,
(19)  Λ f Γ mr + mf + 2
 2Lmr (2Lmr − 1)  

(23)
where Υg = Υ1√ F 2mr , mr − mg + 12 ; mr + mg + 21 ; 1 ,
m √
Υ1 = 4mr −mf +1 π(mr mf ) r and F (·, ·; ·; ·) is the or- mr
where Λf = Γ (mr ) Γ (mf ), Υ2 = 4mr −mf +1 π(m

r mf ) ,
dinary hypergeometric function [54, Eq. (9.100)]. δ = 1 1
Υf = Υ2 F 2mr , mr − mf + 2 ; mr + mf + 2 ; 1 .
(2Lmr − 1)!, Λg = Γ (mr ) Γ (mg ).
Remark 2. Upon substituting (22) and (23) into (17), the
Corollary 7. When Pb approaches infinity, the asymptotic
diversity orders of user f for ARIS-NOMA-HIS and PRIS-
outage probability of user g with ipSIC/pSIC for PRIS-NOMA-
NOMA-HIS networks are equal to Lmr (f + k).
HIS can be respectively expressed as
Corollary 9. When Pb approaches infinity, the asymptotic
K−g
X  k
K −g (−1) outage probability of user o for ARIS-OMA-HIS and PRIS-
PPipSIC,∞
RIS,g ≈Ψg
k (g + k) ψg OMA-HIS can be separately expressed as
k=0
U   √ 
ϕg xu g+k
X " #L
Lm
× mg −1
Hu xu γ bg + 1, √ , HIS,∞ Γ (2mr ) Γ (2mo ) Υo (ϕo + ϑo ) r
cg mg PARIS,o =
, (24)
u=1 Λo Γ mr + mo + 12 2Lmr δβ Lmr
(20)
and
and
p α !
dbr dα 2
K−g
X  k 1 ro ςo σ
K − g (−1) PPHIS,∞ = γ bo + 1, , (25)
PPpSIC,∞
RIS,g = Ψg
RIS,o
Γ (bo + 1) co
k g+k
k=0
" g+k 1 1


where Υo = Υ 3 F 2mr , mr − mo + 2 ; mr + mo + 2 ; 1 ,
#
 Γ (2m ) Γ (2m ) Υ L dα dα ς σ 2
Lmr  √
g m
×
r g g

br rg
. Υ3 = 4mr −mo +1 π(mr mo ) . r

 Λg Γ mr + mg + 12 2Lmr (2Lmr − 1)! 

Remark 3. Upon substituting (24) and (25) into (17), the
(21)
diversity orders of user o for ARIS-OMA-HIS and PRIS-OMA-
Remark 1. As Pb approaches infinity, the outage probability HIS networks are equal to Lmr .
can be expressed as a constant plus the power exponent of
Pb . We note that diversity order is the power exponent, which
determines the rate of Pb converging to the error floor. Upon E. Delay-Limited Transmission
substituting (18) and (20) into (17), the diversity order of user In delay-limited schemes, BS sends information at a steady
g with ipSIC is equivalent to zero. It is due to the effect of rate and breaks based on the random fading of the wireless
residual interference from ipSIC. Upon substituting (19) and channel. The corresponding system throughput is given by
(21) into (17), the diversity order of user g with pSIC for
 
is equal to Lmr (g + k). We can find that diversity order is HIS,Λ
RARIS,dl HIS,Λ
= 1 − PARIS,g HIS
Rg + 1 − PARIS,f


Rf , (26)
associated with the number of reflected elements and the order
of channels and that the introduction of thermal noise from HIS,ipSIC HIS,pSIC
ARIS does not have an effect on the diversity order. where Λ ∈ {ipSIC, pSIC}, PARIS,g , PARIS,g and
HIS
PARIS,f can be obtained from (7), (8) and (12), respectively.
 7

IV. E RGODIC DATA RATE Theorem 5. Conditioned on cascade Nakagami-m fading

channels, the ergodic data rate of user f for ARIS-NOMA-
Ergodic data rate is also an important index for evaluating HIS networks is approximated as
the performance of a communication system. When channel
N √
conditions determine the user’s target rate, the ergodic data rate πaf X 2χf 1 − xn 2
HIS
can be expressed as the maximum rate at which the system can RARIS,f≈
2 ln 2N χf n=1 2χf + af (xn +1)
transmit correctly over cascade Nakagami-m fading channels ( K−f
X K − f  (−1)k 
and it can be defined as 1
× 1 − Ψf
k f + k Γ (bf + 1)
k=0
R = E [log (1 + γφ )] , (27) p !#f +k 
(xn +1) (̟f 1 + ̟f 2 ) 
× γ bf + 1, p , (30)
where γφ denotes the receiving SINR of user φ. cf τf (1 − xn ) 

α α 2
where ̟f 1 = ξβNtn dα α
br drf Lωrf , ̟f 2 = dbr drf σ and τf =
A. The Ergodic Data Rate of ARIS networks βPb χf .
The average user g achievable rate can be derived from (27), Proof. See Appendix D.
erg
which is defined as RARIS,g = E [log (1 + γg )]. In this case,
the expression of user g with ipSIC/pSIC for ARIS-NOMA-
Theorem 6. Conditioned on cascade Nakagami-m fading
HIS networks can be assigned in the following theorem.
channels, the ergodic data rate of the orthogonal user o for
Theorem 4. Conditioned on cascade Nakagami-m fading ARIS-OMA-HIS networks is approximated as
channels, the ergodic data rate of user g with ipSIC for ARIS-
N √
NOMA-HIS networks can be approximated as π X 2χo 1 − xn 2
HIS
RARIS,o ≈
√ ( 2N χo ln 2 n=1 2χo + (xn +1)
N U
HIS,ipSIC πag X 2χg 1 − xn 2 X " p !#
RARIS,g ≈ 1− 1 (xn +1) (̟o1 + ̟o2 )
2N χg ln 2 n=1 2χg + ag (xn +1) u=1
× 1− γ bo + 1, p ,
Γ (bo + 1) co τo (1 − xn )
K−g
X K − g k 
(−1) 1 (31)
× Hu xm
u
g −1
Ψg
k (g + k) Γ (mg ) [Γ (bg + 1)]
k=0
p !#g+k  where τo = βPb χo , ̟o1 = ξβNtn dα α
br dro Lωro and ̟o2 =
(xn +1) (̟g1 xu + ̟g2 )  α α 2
dbr dro σ .
× γ bg + 1, p , (28)
cg mg τg (1 − xn ) 

where τg = βPb χg , ̟g1
= dα α

̟g2 =
br drg ̟Pb and B. The Ergodic Data Rate of PRIS networks
dbr drg mg ξβNtn Lωrg + σ . xn = cos 2n−1
α α 2
2N π is the ab-
scissas in the Gauss-Chebyshev integral formula [56, Eq. The ergodic data rate of PRIS-NOMA-HIS networks can be
(25.4.38)]. obtained with the special case, i.e., β = 1, since the reflective
element for the PRIS does not carry amplification.
Proof. See Appendix C.
Corollary 11. In the unique scenario of β = 1, the ergodic
data rate of user g with ipSIC for PRIS-NOMA-HIS over
Corollary 10. When the parameter ̟ is set to be zero, the er- cascade Nakagami-m fading channels is approximated as
godic data rate of user g with pSIC for ARIS-NOMA-HIS over
cascade Nakagami-m fading channels can be approximated as N √ ( U
πag X 2χg 1 − xn 2 X
RPHIS,ipSIC
RIS,g ≈ 1−
N √ 2N χg ln 2 n=1 2χg + ag (xn +1) u=1
HIS,pSIC πag X 2χg 1 − xn 2
RARIS,g ≈ K−g
X K − g k 
2N χg ln 2 n=1 2χg + ag (xn +1) (−1) 1
× Hu xm
u
g −1
Ψg
( K−g k (g + k) Γ (mg ) [Γ (bg + 1)]
X K − g  (−1)k  1 k=0
× 1 − Ψg p !#g+k 
k g + k Γ (bg + 1) (xn +1) (̟g1 xu + ̟′′ g2 ) 
k=0 × γ bg + 1, p , (32)
p !#g+k  cg mg τg (1 − xn ) 
̟g2,pSIC (xn +1) 
× γ bg + 1, p , (29)
cg τg (1 − xn ) 
2
where ̟′′ g2 = dα α
br drg mg σ .


where ̟′ g2 = dα α 2 Corollary 12. On the condition of ̟ = 0 and β = 1, the
br drg ξβNtn Lωrg + σ .
ergodic data rate with pSIC of user g for PRIS-NOMA-HIS
 8

over cascade Nakagami-m fading channels is approximated and

as N √
√ pSIC,∞ πag X 2χg 1 − xn 2
πag X N
2χg 1 − xn 2 RARIS,g ≈ . (38)
HIS,pSIC
RP RIS,g ≈ 2N χg ln 2 n=1 2χg + ag (xn +1)
2N χg ln 2 n=1 2χg + ag (xn +1)
( K−g
Remark 4. Substituting (37) and (38) into (36), the multi-
X K − g  (−1)k  1 plexing gains for user g for ARIS-NOMA-HIS are equal to
× 1 − Ψg
k g + k Γ (bg + 1) zero, which is inconsistent with traditional NOMA.
k=0
 q g+k 
dα d α σ 2 (x +1) 
 According to (30), the throughput ceiling on the ergodic
br rg n
× γ bf + 1,
 p   . (33) data rate of user f when Pb tends to infinity can be written
cg τg (1 − xn ) 
 by
N √
Corollary 13. On the condition of β = 1, the ergodic data HIS,∞ πaf X 2χf 1 − xn 2
RARIS,f = . (39)
rate of user f for PRIS-NOMA-HIS over cascade Nakagami-m 2N χf ln 2 n=1 2χf + af (xn +1)
fading channels is approximated as
N √ Remark 5. Substituting (39) into (36), user f ’s multiplexing
πaf X 2χf 1 − xn 2 gains for ARIS-NOMA-HIS is also equivalent to zero. This is
RPHIS
RIS,f ≈
2 ln 2N χf n=1 2χf + af (xn +1) because the HIS has an effect on the system.
( K−f
X K − f  (−1)k  1 According to (31), the throughput ceiling on the ergodic
× 1 − Ψf data rate of user o when Pb approaches infinity is
k f + k Γ (bf + 1)
k=0
!#f +k  N √
p  HIS,∞ π X 2χo 1 − xn 2
(xn +1) ̟f 2 RARIS,o = . (40)
× γ bf + 1, p . (34) 2N χo ln 2 n=1 2χo + (xn +1)
cf τf (1 − xn ) 
Remark 6. Substituting (40) into (36), ARIS-OMA-HIS net-
Corollary 14. The ergodic data rate of the orthogonal user
works yield a zero multiplexing gain of user o, which is caused
o for PRIS-OMA-HIS networks can be expressed as
by the the effect of HIS.
N √
HIS π X 2χo 1 − xn 2
RP RIS,o ≈
2 ln 2N χo n=1 2χo + (xn +1) D. Delay-Tolerate Transmission
" p α !#
dbr dα 2 In delay-tolerant schemes, BS transmits data at a steady rate
1 ro σ (xn +1)
× 1− γ bo + 1, p . depending on the user’s channel condition constraints, using
Γ (bo + 1) co τo (1 − xn )
(35) the ergodic capacity as an upper limit. Accordingly, the system
throughput for ARIS-NOMA-HIS networks can be expressed
as
C. Multiplexing Gains Analysis
HIS,Λ HIS,Λ HIS
The multiplexing gains at high SNRs is used as an indicators RARIS,dt = RARIS,g + RARIS,f , (41)
evaluation metric aimed at capturing the diversity of traversal HIS,ipSIC HIS,pSIC
rates using the transmit SNRs [57, 58], and can be given by where Λ ∈ {ipSIC, pSIC}, RARIS,g , RARIS,g and
HIS
∞
RARIS,f can be obtained from (28), (29) and (30), respec-
Rerg (ρ) tively.
S = − lim , (36)
ρ→∞ log (ρ)

∞
where Rerg (ρ) is the asymptotic ergodic data rate at high V. E NERGY E FFICIENCY
SNRs.
For ARIS-NOMA-HIS networks, the system’s total power
Based on (28), we find the ergodic data rate of user g
consumption primarily comprises the power transmitted at BS,
reaches a throughput ceiling when Pb tends to infinity, which
the power consumed at BS, the power consumed at ARIS, the
can be respectively approximated as
output power of ARIS and the hardware power consumption
N √
ipSIC,∞ πag X 2χg 1 − xn 2 at users. This can be denoted specifically by
RARIS,g ≈
2N χg ln 2 n=1 2χg + ag (xn +1) Ptotal = κPb + PBS + LPRE + Pout + PU,φ , (42)
( K−g
XU X K − g  (−1)k ∆
× 1− Hu xmu
g −1
Ψ g where κ = ν −1 with ν is the transmitter power amplifier
u=1
k (g + k) Γ (mg ) efficiency, Pb indicates the emitted power at BS. PBS indicates
k=0
  q g+k  the total static power loss at BS, LPRE is the total hardware
dα dα xmg −1 

1 br rg u static power loss of L reflector elements at ARIS. Pout
× γ bg + 1, p  ,
[Γ (bg + 1)] cg βmg χg (1 − xn ) 
 indicates the output power of ARIS and PU,φ indicates the
hardware static power consumption of the φ-th user. It is
(37) possible to derive that the system’s energy efficiency can be
 9

TABLE I: The simulation parameters for ARIS-NOMA.

0
10
Monte Carlo simulations repeated 106 iterations
ag = 0.25 10
-2
The power allocation factors of user g and user f
af = 0.75

Outage Probability
Simulation
Rg = 1.5 BPCU -4 Asymptotic
The targeted rates of user g and user f 10
DF-HD-OMA-HIS
Rf = 1.5 BPCU DF-FD-OMA-HIS
AF-HD-OMA-HIS
The distance between BS and ARIS dbr = 10 m
10-6
AF-FD-OMA-HIS
User o PRIS-OMA-HIS
The distance between ARIS and user g drg = 10 m User o ARIS-OMA-HIS
User g PRIS-NOMA-HIS ipSIC
The distance between ARIS and user f drf = 20 m -8 User g PRIS-NOMA-HIS pSIC
10 User f PRIS-NOMA-HIS
The distance between ARIS and user o dro = 30 m User g ARIS-NOMA-HIS ipSIC
User g ARIS-NOMA-HIS pSIC
The reflection coefficient of ARIS β=5 -10 User f ARIS-NOMA-HIS
10
The thermal noise power at ARIS Ntn = −30 dBm 0 10 20 30 40 50 60
Pb (dBm)
The AWGN power at users σ2 = −20 dBm
Pass loss expression α = 2.2 Fig. 2: Outage probability versus the transmit Pb , with K = 3,
g = 3, f = 2, L = 5, β = 5, and m = 0.5, Rg = Rf = 1.5
BPCU.
written as the ratio of sum rate to total power consumption of
system, which is represented as
RΦ 0
ηEE = , (43) 10
Ptotal
 
HIS,Λ HIS,Λ HIS,Λ HIS,Λ
where RΦ ∈ RARIS,dl , RARIS,dt . RARIS,dl and RARIS,dt 10
-2

Outage Probability
can be obtained from (26) and (41), respectively.
-4
10 m=0.5, 0.7, 1.0
VI. S IMULATION R ESULTS
Simulation
In this section, the numerical results are provided to ver- 10-6
Error floor
Asymptotic
ify the effectiveness of analytical results for ARIS-NOMA User o ARIS-OMA-HIS
User g ARIS-NOMA-HIS ipSIC
networks. We also consider the impact of HIS on outage User g ARIS-NOMA-HIS pSIC
-8 User f ARIS-NOMA-HIS
probability, ergodic data rate and energy efficiency of ARIS- 10
0 10 20 30 40 50
NOMA networks. The noise power and bandwidth are set to Pb (dBm)
be σ 2 = −174 + 10 log (BW ) and BW = 1000 MHz, respec-
Fig. 3: Outage probability versus the transmit Pb , with L = 5,
tively. For illustration purposes, the simulation results used are
K = 3, g = 3, f = 2, β = 5, and Rg = Rf = 1.5 BPCU.
concluded in Table I [25, 39], where the abridgement of BPCU
denotes bit per channel use. To facilitate comparison, PRIS-
NOMA-HIS, PRIS-OMA-HIS are regarded as benchmarks for
ARIS-NOMA-HIS. In addition, conventional cooperative com- fairness than OMA when serving multiple customers at the
munication schemes, i,e., multi-antenna AF relaying is also se- same time. Another critical finding is that ARIS-OMA-HIS
lected to compare the behaviour of ARIS-NOMA/OMA-HIS. outperforms DF and AF relay switching between full-duplex
Something deserving of clarification is that to demonstrate the (FD) and half-duplex (HD) modes for outages at high SNRs.
accuracy of the results, we have kept the total power consumed The reason for it is that ARIS enhances the signals reflected
by PRIS-NOMA-HIS and ARIS-NOMA-HIS the same. More back to users. This indicates that even though ARIS generates
specifically, PbARIS = PBS ARIS ARIS
+ PRIS + L (PSW + PDC ), amplified noise, its outage performance is still superior to that
P RIS P RIS
Pb = PBS + LPSW and PbARIS = PbP RIS = Pb , of traditional cooperative communications, i.e., HD/FD AF
ARIS P RIS
where PBS and PBS are the transmit power of BS in relaying.
ARIS
ARIS and PRIS networks, PRIS is the signal output power Fig. 3 plots the outage probability for ARIS-NOMA/OMA-
ARIS ARIS 2 2
in ARIS and PRIS = PBS βkΘhbr k + βNtn kΘk , PSW HIS networks versus Pb for different values of the fading
indicates the amount of power used by each RIS element’s parameter m with K = 3, L = 5, g = 3, f = 2, β = 5
control circuit and phase shift switches, PDC means the direct and Rg = Rf = 1.5 BPCU. It can be observed that the
current biasing power [40]. outage probability of ARIS-NOMA-HIS increases with the
Fig. 2 plots the outage probability of ARIS-NOMA-HIS fading factor m. As m becomes higher, i.e., m = 0.5, 0.7 and
networks versus Pb with K = 3, g = 3, f = 2, L = 5, β = 5, 1, ARIS-NOMA-HIS networks are enabling improved outage
m = 0.5 and Rg = Rf = 1.5 BPCU. It shows that the Monte performance. The main reason for this phenomenon is that
Carlo simulation values of outage probability agree perfectly larger values of m correspond to smaller channel fading, and
with the above theoretical analyses. We can find that the outage when m → ∞ indicates no fading. It is worth noting that
probability of user g with ipSIC congregates to an error floor at when m = 1, the cascaded Nakagami-m fading channel is
higher transmit power Pb , which is identical to the conclusion converted to a Rayleigh fading channel. As a further advance,
in Remark 1. ARIS-NOMA-HIS outage performance outper- Fig. 4 shows the impact of HIS on system outage performance.
forms that of ARIS-OMA-HIS because NOMA has higher It is visible that the outage behaviors of ARIS-NOMA-HIS are
 10

10
0
100

-2 -10
10 10
Outage Probability

Outage Probability
-4 -20
10 =0.01, 0.08, 0.15 10

Simulation
Error floor
User g PRIS-NOMA-HIS ipSIC
10-6 Asymptotic
10
-30 User g PRIS-NOMA-HIS pSIC
User o ARIS-OMA-HIS
User f PRIS-NOMA-HIS
User g ARIS-NOMA-HIS ipSIC
User g ARIS-NOMA-HIS ipSIC
User g ARIS-NOMA-HIS pSIC
User g ARIS-NOMA-HIS pSIC
User f ARIS-NOMA-HIS
User f ARIS-NOMA-HIS
0 10 20 30 40 50 10-40
20 40 60 80 100 120 140
Pb (dBm) RIS element number L

Fig. 4: Outage probability versus the transmit Pb , with L = 5, Fig. 6: Outage probability versus the reflection elements
K = 3, g = 3, f = 2, m = 0.5, β = 5, and Rg = Rf = 1.5 number L, with K = 3, g = 3, f = 2, m = 0.5, β = 10, and
BPCU. Rg = Rf = 1.5 BPCU.

0 0
10 10

-1 -2
10 10

P b= 25 dBm
Outage Probability

Outage Probability
-2 Ntn = -30 dBm -4
10 10

-3 -6
10 10

-4
10 10-8
Ntn = -60 dBm P b= 30 dBm

-5 User o ARIS-OMA-HIS -10 User o ARIS-OMA-HIS

10 User g ARIS-NOMA-HIS ipSIC 10 User g ARIS-NOMA-HIS ipSIC
User g ARIS-NOMA-HIS pSIC User g ARIS-NOMA-HIS pSIC
-6 User f ARIS-NOMA-HIS -12 User f ARIS-NOMA-HIS
10 10
0 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 45 50
The distance between BS and ARIS (m) Reflection amplification coefficient

Fig. 5: Outage probability versus the different distance be- Fig. 7: Outage probability versus the reflection amplitude
tween BS and ARIS, with L = 5, K = 3, g = 3, f = 2, factors, with L = 5, K = 3, g = 3, f = 2, m = 0.7,
m = 0.5, β = 5, and Rg = Rf = 1.5 BPCU. and Rg = Rf = 1.5 BPCU.

in connection with the HIS. As κ increases, the gap between networks versus the different reflection elements number L
outage probabilities becomes larger at high SNRs. It takes with Pb = 20 dBm, K = 3, g = 3, f = 2, m = 0.5,
Pb = 10 to improve outage performance even if κ = 0.01. β = 10 and Rg = Rf = 1.5 BPCU. One can observe that ini-
The slope of the curve stays identical regardless of the value tially both ARIS-NOMA-HIS and PRIS-NOMA-HIS achieve
of κ changes. This phenomenon indicates that we can improve better performance when L rises. However, as L increase
the outage behaviours of ARIS-NOMA-HIS by reducing the to a certain level, the outage performance of ARIS-NOMA-
HIS. HIS starts to deteriorate. The reason for this phenomenon
Fig. 5 plots the outage probability of ARIS-NOMA/OMA- is that as L increases, thermal noise from ARIS becomes
HIS networks versus the different distance from BS to ARIS. larger. This means that ARIS-NOMA-HIS networks performs
Assuming that the BS is in the same plane as user g and about similar as PRIS-NOMA-HIS networks when there are
user f , and they are located at (0, 0), (30, 0) and (40, −10), fewer reflective elements. It is worth considering the trade-
respectively. Besides, ARIS are located at (xARIS , 10), where off between reflection elements number of ARIS and outage
xARIS indicates the distance between BS and ARIS. It can performance.
be noted that the outage performance of user f is gradually Fig. 7 plots the outage probability of ARIS-NOMA/OMA-
deteriorating. The causes of this phenomenon is that as drf HIS networks versus the different reflection amplitude factors
decreases, the increasing trend of denominator in (4) is much β with L = 5, K = 3, g = 3, f = 2, m = 0.7 and
greater than numerator. In summary, ARIS deployed on BS Rg = Rf = 1.5 BPCU. It can be seen that as β increases,
side offers higher performance gains. In addition, reducing the the outage performance gain gradually reaches saturation. This
thermal noise values Ntn does not improve the performance is because that as β tends to infinity, the outage probability
of ARIS-OMA-HIS networks. approaches a particular value. Moreover, the gain in outage
Fig. 6 plots the outage probability of ARIS-NOMA-HIS probability for the ARIS-NOMA-HIS networks with pSIC is
 11

3 7
PRIS-OMA-HIS Thtoughput ceiling = 0.08
ARIS-OMA-HIS PRIS-OMA-HIS
2.5 PRIS-NOMA-HIS ipSIC 6 ARIS-OMA-HIS
System Throughout (BPCU)
PRIS-NOMA-HIS pSIC PRIS-NOMA-HIS ipSIC
ARIS-NOMA-HIS ipSIC PRIS-NOMA-HIS pSIC
5
ARIS-NOMA-HIS pSIC ARIS-NOMA-HIS ipSIC
2

Rate (BPCU)
ARIS-NOMA-HIS pSIC
4
= 0.15
1.5
3
1
2

0.5 1

0 0
0 10 20 30 40 50 0 10 20 30 40 50 60
Pb (dBm) Pb (dBm)

Fig. 8: System throughput versus the transmit Pb , with L = 5, Fig. 10: Rate versus the transmit Pb , with L = 2, K = 3,
K = 3, g = 3, f = 2, m = 0.7, and Rg = Rf = 1.5 BPCU. g = 3, f = 2, m = 0.7, and Rg = Rf = 1.5 BPCU.

5 rates versus the transmit Pb in delay-tolerate schemes with

Simulation
Throughput ceiling
User o PRIS-OMA-HIS
L = 2, K = 3, g = 3, f = 2, m = 0.7 and Rg = Rf = 1.5
4
User o ARIS-OMA-HIS
User g PRIS-NOMA-HIS ipSIC
BPCU. The influence of changing the HIS factor κ on ergodic
User g PRIS-NOMA-HIS pSIC
User f PRIS-NOMA-HIS
data rate of system was plotted, which basing on (41). We can
observe that as κ increases, i.e., from 0.08 to 0.15, the ergodic
Rate (BPCU)

User g ARIS-NOMA-HIS ipSIC

3 User g ARIS-NOMA-HIS pSIC
User f ARIS-NOMA-HIS
performance degradation in ergodic data rate increases, and is
2 more pronounced in ARIS/PRIS-NOMA-HIS with pSIC. At
the same time, the ergodic data rata reached an upper ceiling
1 due to the effects of HIS. This demonstrates the importance
of accurately modeling HIS values when evaluating ARIS-
0 NOMA networks. Moreover, the residual interference from
0 10 20 30 40 50 60
Pb (dBm)
ipSIC has no significant effect on ARIS-NOMA-HIS networks
when Pb is lower.
Fig. 9: Rate versus the transmit Pb , with L = 2, K = 3,
g = 3, f = 2, m = 0.7, β = 5 and Rg = Rf = 1.5 BPCU. 30 9
PRIS-OMA-HIS PRIS-OMA-HIS
ARIS-OMA-HIS ARIS-OMA-HIS
8
PRIS-NOMA-HIS ipSIC PRIS-NOMA-HIS ipSIC
25
PRIS-NOMA-HIS pSIC PRIS-NOMA-HIS pSIC
7
Energy Effrciency (bit/J)

Energy Effrciency (bit/J)

ARIS-NOMA-HIS ipSIC ARIS-NOMA-HIS ipSIC
ARIS-NOMA-HIS pSIC ARIS-NOMA-HIS pSIC
20 6
=5, 10, 15
=5, 10, 15

more pronounced when Pb is increased from 25 dBm to 30 15

5

4
dBm. Additionally, Fig. 8 plots the system throughput versus 10 3

the transmit Pb in delay-limited case with L = 5, K = 3, 5

2

g = 3, f = 2, m = 0.7 and Rg = Rf = 1.5 BPCU. The 0

1

0
system throughput of ARIS-NOMA-HIS is plotted on the basis 0 5 10 15 20
Pb (dBm)
25 30 35 40 5 10 15 20 25
Pb (dBm)
30 35 40

of (26). As can be observed that ARIS-NOMA-HIS is able to

(a) Delay-tolerant mode (b) Delay-limited mode
reach throughput ceiling earlier than PRIS-NOMA-HIS. The
throughput ceiling of ARIS/PRIS-NOMA-HIS with ipSIC is Fig. 11: Energy efficiency in delay-tolerant and delay-limited
lower than that with pSIC because of the residual interference transmission mode, with L = 2, K = 3, g = 3, f = 2,
from ipSIC. The system throughput of ARIS-NOMA-HIS m = 0.5, and Rg = Rf = 1.5 BPCU.
is far more favourable than that of PRIS-NOMA-HIS and
ARIS-OMA-HIS. The major contributing factors is that ARIS- Fig. 11(a) and Fig. 11(b) plot the energy efficiency curves
NOMA-HIS networks have better spectral efficiency and the in delay-tolerant and delay-limited transmission modes with
capability to offer service for multiple users. L = 2, K = 3, g = 3, f = 2, m = 0.5 and Rg = Rf = 1.5
Fig. 9 plots the ergodic data rates versus the transmit power BPCU, respectively. Fig. 11(a) shows the delay-tolerant trans-
Pb with L = 2, K = 3, g = 3, f = 2, m = 0.7, β = 5 mission model of ARIS/PRIS-NOMA-HIS networks. It can
and Rg = Rf = 1.5 BPCU. We can observe that the ergodic be observed that ARIS-OMA-HIS and PRIS-OMA-HIS have
data rate for the ARIS-NOMA-HIS networks is still higher lower energy efficiency and that of the ARIS-NOMA-HIS
than that for PRIS-NOMA-HIS. Future more, ARIS-NOMA- networks is significantly improved compared to PRIS-NOMA-
HIS with ipSIC achieves a weaker rate than that with pSIC. HIS. As can be seen the system’s energy efficiency begins to
As Pb increases, the rate for user f gradually reaches the deteriorate at Pb > 15 dBm, This can be inferred by the fact
upper bound and this is in alignment with the findings in that the total power consumption of the system becomes larger
Remark 4 and Remark 5. Fig. 10 plots the ergodic data as Pb increases. Furthermore, the energy efficiency of the
 12

ARIS-NOMA-HIS dramatically increases as the growth of the having the parameter of (mg , mg ) and the corresponding PDF
reflection amplification β. This conclusions can also be applied is given by
in delay-limited schemes. The difference is that the energy
efficiency of ARIS-NOMA-HIS with ipSIC is significantly mg mg xmg −1 −mg x
fX (x) = e . (A.2)
worse than that with pSIC. This demonstrates how delay- Γ (mg )
limited systems are substantially more sensitive to residual
interference from ipSIC than delay-tolerant ones. As the weak It may be seen that the CDF of Y cannot be derived in
trend of ARIS-NOMA-HIS networks indicates its stability in a straightforward way. However, we can make use of the
transmitting information. Laguerre polynomials to provide the approximated CDF. Let
Yl = hlbr hlrg , the expectation and variance of Yl can be
Γ(mr + 12 )Γ(mg + 21 )
respectively given by µg = E (Yl ) = Γ(mr )Γ(m ) m m and
√
VII. C ONCLUSION  g2 r g
Γ(mr + 2 )Γ(mg + 2 )
1 1

This paper has investigated the performance of ARIS- Ωg = D (Yl ) = 1 − mr1mg Γ(mr )Γ(mg ) , where mr
NOMA-HIS networks over the cascade Nakagami-m fading denote the Nakagami-m fading parameter from BS to ARIS.
channels. More specifically, we derived the outage probability Conditioned that there is no channel sorting,y the PDF of
and ergodic data rate of ARIS-NOMA-HIS networks with Y can be approximated as fY (y) ≈ y bg e
−
cg
[59, Eq.
b +1
ipSIC/pSIC. On the basis of the asymptotic results, the di- cgg Γ(bg +1)
L[E(Yl )]2 D(Yl )
versity orders and multiplexing gains for user g and user f (2.76)], where bg = − 1, cg =
D(Yl ) E(Yl ) .
The integration
are acquired. The results indicated that the outage performance operations of fY (y) are carried out, andthen its CDF
 can
of ARIS-NOMA-HIS outperformed PRIS-NOMA-HIS, ARIS- be further derived as FY (y) ≈ Γ(bg1+1) γ bg + 1, cyg . As a
OMA-HIS, and other traditional collaborative relays. Based on further advance, the CDF of Y with channel sorting can be
the analysis, the diversity order is related to reflection elements expressed as
number and the order of the channel. Numerical results re-
vealed that the outage probability and ergodic data rate of user K−g
X K − g
K!
g in ARIS-NOMA-HIS networks was significantly better than FY (y) ≈
that of orthogonal users at high SNRs. We studied the system (K − g)! (g − 1)! k
k=0
throughput and energy efficiency of ARIS-NOMA-HIS in both k   g+k
(−1) 1 y
delay-limited and delay-tolerant schemes. The single antenna × γ bg + 1, . (A.3)
g + k Γ (bg + 1) cg
and perfect CSI setup in this paper leads to an overestimation
in performance of ARIS-NOMA-HIS networks, so our future By substituting (A.2) and (A.3) into (A.1), the outage proba-
work will consider the influence of imperfect CIS and take bility of user g for ARIS-NOMA-HIS is approximated as
into account more complex assumptions. !
K−g
ipSIC K! X K − g (−1)k mg mg
PARIS,g =
(K − g)! (g − 1)! k=0 k (g + k) ψg
A PPENDIX A: P ROOF OF T HEOREM 1   q g+k
Z ∞ mg −1 ϕg x + ϑ̃g
Based on the definitions of outage probability for user g, x
× γ bg + 1, √  dx.
the proof processes start by substituting (2) and (3) into (6), 0 emg x cg β
and then using simple arithmetic operations, the corresponding
(A.4)
outage probability for ARIS-NOMA-HIS networks can be
calculated as
  Applying the Gauss-Laguerre quadrature [60, Eq. (8.6.5)]
ipSIC
2
into the above equation, we will acquire (7). The proof is
PARIS,g =Pr 0 < hH
rg Θhbr < ςg β −1 2
̟Pb |hI | + ng
completed.
 
v  
u
 L u 
 X u 
=Pr 
0 < hlbr hlrg <u
t ζ g β −1 ̟P |h |2 +n 
 b I g  A PPENDIX B: P ROOF OF T HEOREM 2
 l=1
|{z} 
| {z } X
Y Substituting (4) into (11), the outage probability for user f
Z ∞ p  is evaluated by
= fX (x) FY ζg β −1 (̟Pb x + ng ) dx, (A.1)
0 h 2
i
HIS
PARIS,f = Pr hH rf Θhbr < ςf β −1 ξβNtn Lωrf + σ 2
where ̟ = ξ = 1, ng = ξβNtn Lωrg + σ 2 , ζg = dα α
br drg ςg .  
Note that the equation (A.1) is derived with the assistance
of coherent phase shifting, which can maximize the received X 
 L l l q 
= Pr 
 α α
hbr hrf < dbr drf ςf β nf 
−1
performance of the desired users. To compute the equation 
(A.1), we should first solvePthe PDF and CDF of random vari-  l=1
| {z }

ables X = |hI |2 and Y = L l l
l=1 hbr hrg , respectively. Since q W 
hI is assumed to obey the Nakagami-m distribution with the
2 = FW dα α
br drf ςf β
−1 n
f , (B.1)
parameter of mg , |hI | is subjected to the Gamma distribution
 13

where ξ = 1, nf = ξβNtn Lωrf + σ 2 . For the same reason as A PPENDIX D: P ROOF OF T HEOREM 5
(A.2) and (A.3), we can derive the CDF of W with respect to Substituting (4) into (27), the ergodic data rate for user f
the sorted as of ARIS-NOMA-HIS networks can be calculated by
K−f
X K − f  erg
K! RARIS,f = E [log (1 + X2 )]
FW (w) ≈
(K − f )! (f − 1)! k βPb af
Z 1
βPb (ag +κ2 +κ2 ) 1 − FX2 (y)
k=0
  f +k = b f
dy, (D.1)
(−1)k 1 w ln 2 0 1 + βPb af y
× γ bf + 1, .
f + k Γ (bf + 1) cf and
(B.2) 2
L
P
βPb d−α −α
br drf hlbr hlrf af
By substituting (B.2) into (B.1), we will acquire (12). The l=1
proof is completed. X2 = 2 .
L
P
βPb d−α −α
br drf hlbr hlrf χf + ξβNtn Lωrf + σ2
l=1
A PPENDIX C: P ROOF OF T HEOREM 4 This can be written specifically as
 
Substituting (3) into (27) and further exploiting the coherent L
v
u ydα dα (ξβN Lω + σ 2 )
X
phase shifting scheme, the ergodic data rate for user g of 
FX2 (y) = Pr 
u br rf
hlbr hlrf < t 
tn rf
 

ARIS-NOMA-HIS is calculated by l=1 1 − yβPb a g + κ2
b + κ2
f
 
erg v
RARIS,g = E [log (1 + X1 )] u ydα dα (ξβN Lω + σ 2 )
u br rf  tn rf  
Z 1 = FY t . (D.2)
βPb ag βPb (κ2b +κ2g ) 1 − FX1 (y) 2
1 − yβPb ag + κb + κf 2
= dy, (C.1)
ln 2 0 1 + βPb ag y
Substituting (A.2) and (A.3) into (D.2) and approximating
and with the Gauss-Laguerre integral is able to obtain
2 K−f
!
L
P K! X K − f (−1)k
βPb ag d−α −α
br drg hlbr hlrg FX2 (y) =
(K − f )! (f − 1)! k=0 k f +k
l=1
X1 = 2 .   √ 
L
X 1 y̟f 1 + y̟f 2 f +k
2 × γ bf + 1, √ . (D.3)
βPb d−α −α
br drg hlbr hlrg χg + ̟Pb |hI | +ng Γ (bf + 1) cf 1 − yτf
|{z}
l=1
| {z } X Substituting (D.3) into (D.1), we will acquire (30). The
Y proof is completed.
This can be written specifically as
 R EFERENCES
L
s
2


X ydα d α
br rg ̟P b |hI | + ng [1] M. Song, X. Yue, C. Ouyang, Y. Liu, T. Li, and T. Hou, “Outage perfor-
FX1 (y) = Pr  hlbr hlrg < 
mance of active RIS in NOMA networks with hardware impairments,”
1 − yβPb χg
l=1 in IEEE Trans. of Veh. Technol. Conf. (VTC), Hong Kong, CHN, Oct.
s !
Z 2023, pp. 1–6.
∞
ydα α
br drg (̟Pb x + ng ) [2] S. Chen, Y.-C. Liang, S. K. S. Sun, W. Cheng, and M. Peng, “Vision,
= fX (x)FY
dx. (C.2)
0 1 − yβPb κ2b + κ2g requirements, and technology trend of 6G: how to tackle the challenges
of system coverage, capacity, user data-rate and movement speed,” IEEE
Substituting (A.2) and (A.3) into (C.2) and approximating Wireless Commun., vol. 27, no. 2, pp. 218–228, Apr. 2020.
with the Gauss-Laguerre integral is able to obtain [3] D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, D. Niyato,
O. Dobre, and H. V. Poor, “6G internet of things: A comprehensive
U
! K−g
survey,” IEEE Internet Things J., vol. 9, no. 1, pp. 359–383, Jan. 2022.
X m −1K−g X(−1)k [4] Cisco, “Cisco visual networking index: Global mobile data traffic
FX1 (y) = H u x u g Ψg forecast update 2017-2022 white paper,” 2019.
u=1 k=0
k (g + k) Γ (mg )
[5] Y. Liu, S. Zhang, X. Mu, Z. Ding, R. Schober, N. Al-Dhahir, E. Hossain,
" √ !#g+k
and X. Shen, “Evolution of NOMA toward next generation multiple
1 y̟g1 xu + y̟g2 access (NGMA) for 6G,” IEEE J. Sel. Areas Commun., vol. 40, no. 4,
× γ bg + 1, p ,
[Γ (bg + 1)] cg mg (1 − yτg ) pp. 1037–1071, Jan. 2022.
(C.3) [6] X. Pei, Y. Chen, M. Wen, H. Yu, E. Panayirci, and H. V. Poor,
“Next-generation multiple access based on NOMA with power level
In order to adequately simplify the results obtained, we use modulation,” IEEE J. Sel. Areas Commun., vol. 40, no. 4, pp. 1072–
1083, Apr. 2022.
the following Gauss-Chebyshev integral formula [7] Y. Liu, W. Yi, Z. Ding, X. Liu, O. A. Dobre, and N. Al-Dhahir,
“Developing noma to next generation multiple access: Future vision and
Z b N   research opportunities,” IEEE Wireless Communications, vol. 29, no. 6,
πb X (xn +1) b p
f (x)dx ≈ f 1 − xn 2 . (C.4) pp. 120–127, Dec. 2022.
0 2N n=1 2 [8] T. Cui, M. Qi, X. Wan, J. Zhao, and Q. Cheng, “Coding metamaterials,
digital metamaterials and programmable metamaterials,” Light: Science
Substituting (C.3) into (C.1) and simplifying the result ob- & Applications,, vol. 3, no. 10, Oct. 2014.
[9] S. V. Hum and J. Perruisseau-Carrier, “Reconfigurable reflectarrays and
tained by substituting it into (C.4), we will acquire (28). The array lenses for dynamic antenna beam control: A review,” IEEE Trans.
proof is completed. Antennas Prop., vol. 62, no. 1, pp. 183–198, Jan. 2014.
 14

[10] Q. Wu and R. Zhang, “Towards smart and reconfigurable environment: [32] J. Lonǎar and Z. Šipuš, “Challenges in design of power-amplifying active
Intelligent reflecting surface aided wireless network,” IEEE Commun. metasurfaces,” in in Proc. IEEE Int. Symp., ELMAR, Zadar, Croatia,
Mag., vol. 58, no. 1, pp. 106–112, Jan. 2020. Oct. 2020, pp. 9–12.
[11] S. Basharat, S. A. Hassan, H. Pervaiz, A. Mahmood, Z. Ding, and [33] X. Wang, J. Han, S. Tian, D. Xia, L. Li, and T. Cui, “Amplification
M. Gidlund, “Reconfigurable intelligent surfaces: Potentials, applica- and manipulation of nonlinear electromagnetic waves and enhanced
tions, and challenges for 6G wireless networks,” vol. 28, no. 6, pp. nonreciprocity using transmissive space-time-coding metasurface,” Adv.
184–191, Dec. 2021. Sci., vol. 9, no. 11, p. 2105960, Apr. 2022.
[12] Y. Liu, X. Liu, X. Mu, T. Hou, J. Xu, M. D. Renzo, and N. Al-Dhahir, [34] Z. Zhang, L. Dai, X. Chen, C. Liu, F. Yang, R. Schober, and H. V. Poor,
“Reconfigurable intelligent surfaces: Principles and opportunities,” IEEE “Active RIS vs. passive RIS: Which will prevail in 6G?” IEEE Trans.
Commun. Surveys Tuts., vol. 23, no. 3, pp. 1546–1577, May 2021. Commun., vol. 71, no. 3, pp. 1707–1725, Mar. 2023.
[13] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and [35] E. Basar and H. V. Poor, “Present and future of reconfigurable intelli-
C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in gent surface-empowered communications,” IEEE Signal Process. Mag.,
wireless communication,” vol. 18, no. 8, pp. 4157–4170, Aug. 2019. vol. 38, no. 6, pp. 146–152, Nov. 2022.
[14] T. Wang, M.-A. Badiu, G. Chen, and J. P. Coon, “Outage probability [36] K. Liu, Z. Zhang, L. Dai, S. Xu, and F. Yang, “Active reconfigurable
analysis of RIS-assisted wireless networks with von mises phase errors,” intelligent surface: Fully-connected or sub-connected?” IEEE Commun.
IEEE Wireless Commun. Lett., vol. 10, no. 12, pp. 2737–2741, Dec. Lett., vol. 26, no. 1, pp. 167–171, Jan. 2022.
2021. [37] R. Long, Y. C. Liang, Y. Pei, and E. G. Larsson, “Active reconfigurable
[15] A.-A. A. Boulogeorgos and A. Alexiou, “How much do hardware im- intelligent surface-aided wireless communications,” IEEE Trans. Com-
perfections affect the performance of reconfigurable intelligent surface- mun., vol. 20, no. 8, pp. 4962–4975, Aug. 2021.
assisted systems?” IEEE Open J. Commun. Soc., vol. 8, no. 7, pp. 1185– [38] R. A. Tasci, F. Kilinc, E. Basar, and G. C. Alexandropoulos, “A new
1195, Aug. 2021. RIS architecture with a single power amplifier: Energy efficiency and
[16] Y. Chen, B. Zhang, M. Ding, D. López-Pérez, M. Hassan, M. Debbah, error performance analysis,” IEEE Access, vol. 10, pp. 44 804–44 815,
and Z. D. Chen, “Downlink performance analysis of intelligent reflecting Apr. 2022.
surface-enabled networks,” IEEE Trans. Veh. Technol., vol. 72, no. 2, pp. [39] C. You and R. Zhang, “Wireless communication aided by intelligent
2082–2097, Feb. 2023. reflecting surface: Active or passive?” IEEE Wireless Commun. Lett.,
[17] Y. Ni, Y. Liu, J. Wang, Q. Wang, H. Zhao, and H. Zhu, “Performance vol. 10, no. 12, pp. 2659–2663, Dec. 2021.
analysis for RIS-assisted D2D communication under Nakagami-m fad- [40] K. Zhi, C. Pan, H. Ren, K. K. Chai, and M. Elkashlan, “Active RIS
ing,” IEEE Trans. Veh. Technol., vol. 70, no. 6, pp. 5865–5879, Jun. versus passive RIS: Which is superior with the same power budget?”
2021. IEEE Commun. Lett., vol. 26, no. 5, pp. 1150–1154, May 2022.
[18] F. E. Bouanani, S. Muhaidat, P. C. Sofotasios, O. A. Dobre, and O. S. [41] P. Zeng, D. Qiao, Q. Wu, and Y. Wu, “Active IRS aided WPCNs: A
Badarneh, “Performance analysis of intelligent reflecting surface aided new paradigm towards higher efficiency and wider coverage,” 2021.
wireless networks with wireless power transfer,” IEEE Commun. Lett., [Online]. Available: https://arxiv.org/abs/2111.11600v1.
vol. 25, no. 3, pp. 793–797, Mar. 2021.
[42] G. Chen, Q. Wu, C. He, W. Chen, J. Tang, and S. Jin, “Active IRS
[19] Z. Ding, L. Lv, F. Fang, O. A. Dobre, G. K. Karagiannidis, N. Al-Dhahir,
aided multiple access for energy-constrained IoT systems,” IEEE Trans.
R. Schober, and H. V. Poor, “A state-of-the-art survey on reconfigurable
Wireless Commun., vol. 22, no. 3, pp. 1677–1694, Mar. 2023.
intelligent surface-assisted non-orthogonal multiple access networks,”
[43] H. Wang, C. Liu, Z. Shi, Y. Fu, and R. Song, “Power minimization
Proc. IEEE, vol. 110, no. 9, pp. 1358–1379, Sep. 2022.
for uplink ris-assisted comp-noma networks with gsic,” IEEE Trans.
[20] A. Khaleel and E. Basar, “A novel NOMA solution with RIS partition-
Commun., vol. 70, no. 7, pp. 4559–4573, Jul. 2022.
ing,” IEEE J. Sel. Signal Process., vol. 16, no. 1, pp. 70–81, Jun. 2022.
[44] E. Björnson, M. Matthaiou, and M. Debbah, “A new look at dual-hop
[21] Y. Liu, X. Mu, X. Liu, M. Di Renzo, Z. Ding, and R. Schober,
relaying: Performance limits with hardware impairments,” IEEE Trans.
“Reconfigurable intelligent surface-aided multi-user networks: Interplay
between NOMA and RIS,” IEEE Wireless Commun., vol. 29, no. 2, pp. Commun., vol. 61, no. 11, pp. 4512–4525, Nov. 2013.
169–176, Apr. 2022. [45] M. Jian, G. C. Alexandropoulos, E. Basar, C. Huang, R. Liu, Y. Liu,
[22] A. S. d. Sena, D. Carrillo, F. Fang, P. H. J. Nardelli, D. B. d. and C. Yuen, “Reconfigurable intelligent surfaces for wireless communi-
Costa, U. S. Dias, Z. Ding, C. B. Papadias, and W. Saad, “What role cations: Overview of hardware designs, channel models, and estimation
do intelligent reflecting surfaces play in multi-antenna non-orthogonal techniques,” Intelligent and Converged Networks, vol. 3, no. 1, pp. 1–32,
multiple access?” IEEE Wireless Commun., vol. 27, no. 5, pp. 24–31, Mar. 2022.
Oct. 2020. [46] E. Björnson, O. Özdogan, and E. G. Larsson, “Intelligent reflecting
[23] B. Zheng, Q. Wu, and R. Zhang, “Intelligent reflecting surface-assisted surface versus decode-and-forward: How large surfaces are needed to
multiple access with user pairing: NOMA or OMA?” IEEE Commun. beat relaying?” IEEE Wireless Commun. Lett., vol. 9, no. 2, pp. 244–
Lett., vol. 24, no. 4, pp. 753–757, Apr. 2020. 248, Feb. 2020.
[24] Z. Ding and H. V. Poor, “A simple design of IRS-NOMA transmission,” [47] X. Yue, Y. Liu, Y. Yao, X. Li, R. Liu, and A. Nallanathan, “Secure com-
IEEE Commun. Lett., vol. 24, no. 5, pp. 1119–1123, May. 2020. munications in a unified non-orthogonal multiple access framework,”
[25] T. Hou, Y. Liu, Z. Song, X. Sun, Y. Chen, and L. Hanzo, “Reconfig- IEEE Trans. Wireless Commun., vol. 19, no. 3, pp. 2163–2178, Jun.
urable intelligent surface aided NOMA networks,” IEEE J. Sel. Areas 2020.
Commun., vol. 38, no. 11, pp. 2575–2588, Nov. 2020. [48] B. Zheng, C. You, W. Mei, and R. Zhang, “A survey on channel estima-
[26] X. Yue and Y. Liu, “Performance analysis of intelligent reflecting surface tion and practical passive beamforming design for intelligent reflecting
assisted NOMA networks,” IEEE Trans. Wireless Commun., vol. 21, surface aided wireless communications,” IEEE Commun. Surveys Tuts.,
no. 4, pp. 2623–2636, Apr. 2022. vol. 24, no. 2, pp. 1035–1071, Feb. 2022.
[27] L. Bariah, S. Muhaidat, P. C. Sofotasios, F. E. Bouanani, O. A. Dobre, [49] A. L. Swindlehurst, G. Zhou, R. Liu, C. Pan, and M. Li, “Channel esti-
and W. Hamouda, “Large intelligent surface-assisted non-orthogonal mation with reconfigurable intelligent surfaces - A general framework,”
multiple access for 6G networks: Performance analysis,” IEEE Internet Proc. IEEE, vol. 110, no. 9, pp. 1312–1338, Sep. 2022.
Things J., vol. 8, no. 7, pp. 5129–5140, Apr. 2021. [50] B. Hassibi and B. Hochwald, “How much training is needed in multiple-
[28] Z. Liu, X. Yue, C. Zhang, Y. Liu, Y. Yao, Y. Wang, and Z. Ding, antenna wireless links?” IEEE Trans. Inf. Theory, vol. 49, no. 4, pp.
“Performance analysis of reconfigurable intelligent surface assisted two- 951–963, Apr. 2003.
way NOMA networks,” IEEE Trans. Veh. Technol., vol. 71, no. 12, pp. [51] J. Jose, A. Ashikhmin, T. L. Marzetta, and S. Vishwanath, “Pilot
13 091–13 104, Dec. 2022. contamination problem in Multi-Cell TDD systems,” in 2009 IEEE
[29] A. Hemanth, K. Umamaheswari, A. C. Pogaku, D.-T. Do, and B. M. Lee, International Symposium on Information Theory, 2009, pp. 2184–2188.
“Outage performance analysis of reconfigurable intelligent surfaces- [52] Z. Ding, Z. Yang, P. Fan, and H. V. Poor, “On the performance of
aided NOMA under presence of hardware impairment,” IEEE Access, non-orthogonal multiple access in 5G systems with randomly deployed
vol. 8, pp. 212 156–212 165, Nov. 2020. users,” IEEE Signal Process. Lett., vol. 21, no. 12, pp. 1501–1505, Dec.
[30] M. H. N. Shaikh, V. A. Bohara, A. Srivastava, and G. Ghatak, “A down- 2014.
link ris-aided NOMA system with hardware impairments: Performance [53] X. Yue, Y. Liu, S. Kang, A. Nallanathan, and Z. Ding, “Exploiting
characterization and analysis,” vol. 3, pp. 288–305, Jul. 2022. full/half-duplex user relaying in NOMA systems,” IEEE Trans. Com-
[31] F. Amato, C. W. Peterson, B. P. Degnan, and G. D. Durgin, “Tunneling mun., vol. 66, no. 2, pp. 560–575, Feb. 2018.
RFID tags for long-range and low-power microwave applications,” IEEE [54] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series and
J. Radio Freq. Identificat., vol. 2, no. 2, pp. 93–103, Jun. 2018. Products, 6th ed. New York, NY, USA: Academic Press, 2000.
 15

[55] Y. Pei, X. Yue, W. Yi, Y. Liu, X. Li, and Z. Ding, “Secrecy outage
probability analysis for downlink RIS-NOMA networks with on-off
control,” IEEE Trans. Veh. Technol., pp. 1–15, Apr. 2023.
[56] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions
With Formulas, Graphs, and Mathematical Tables, 1964.
[57] L. Zheng and D. Tse, “Diversity and multiplexing: a fundamental
tradeoff in multiple-antenna channels,” IEEE Trans. Inf. Theory, vol. 49,
no. 5, pp. 1073–1096, May. 2003.
[58] X. Yue, J. Xie, Y. Liu, Z. Han, R. Liu, and Z. Ding, “Simultaneously
transmitting and reflecting reconfigurable intelligent surface assisted
NOMA networks,” IEEE Trans. Wireless Commun., vol. 22, no. 1, pp.
189–204, Jan. 2023.
[59] V. K. S. Primak and V. Lyandres, Stochastic Methods and their Applica-
tions to Communications: Stochastic Differential Equations Approach,
West Sussex, U.K.: Wiley, 2004.
[60] E. Hildebrand, Introduction to numerical analysis, New York, NY, USA:
Dover, 1987.