1
A Novel Design of RIS for Enhancing the Physical
Layer Security for RIS-aided NOMA Networks
Zhiqing Tang, Student Member, IEEE, Tianwei Hou, Member, IEEE, Yuanwei Liu, Senior Member, IEEE,
Jiankang Zhang, Senior Member, IEEE, and Caijun Zhong, Senior Member, IEEE
Abstract—This letter proposes a novel design of reconfigurable
intelligent surface (RIS) to enhance the physical layer security
(PLS) in the RIS-aided non-orthogonal multiple access (NOMA)
network. Under the design of the RIS, the problem of increasing
the number of RIS elements damaging the secrecy performance
is solved. Besides, it also ensures that the networks can use
traditional channel coding schemes to achieve secrecy. Our results
show that the novel design of the RIS is ready for enhancing
secrecy performance.
Index Terms—Non-orthogonal multiple access, physical layer
security, reconfigurable intelligent surface.
I. INTRODUCTION
D
UE to the superior spectrum efficiency (SE), non-
orthogonal multiple access (NOMA) will play an impor-
tant role in 5G and beyond. Compared with the conventional
orthogonal multiple access (OMA) network structure, NOMA
has the outstanding ability to strengthen the SE and user
connectivity [1]. Since the signals are broadcast in wireless
communication networks, it is equally important to guarantee
the confidentiality of communication going on between the
base station (BS) and legitimate users (LUs).
In recent times, a new technology which has the ability
to manage the reflection properties of the radio waves, named
reconfigurable intelligent surface (RIS), has been proposed [2].
By properly adapting the amplitude-reflection and phase coef-
ficients, the RIS can enhance or reduce the received signals by
users [3], [4], which provides more possibilities for physical
layer security issues. In [5], the authors studied the secrecy
outage probability (SOP) of an RIS-aided wireless network,
and revealed that the RIS make a motion to boost the secrecy
performance. The PLS of a vehicular network with the assisted
of the RIS has been studied in [6]. In [7], the authors analyzed
the SOP of the RIS-aided NOMA network with multi-user.
This work was supported by the National Natural Science Foundation
of China under Grant 61571401 and 61901416. (Corresponding authors:
Yuanwei Liu; Jiankang Zhang.)
Z. Tang is with the School of Information Engineering, Zhengzhou Uni-
versity, Zhengzhou 450001, China (email:iezqtang@zzu.edu.cn).
T. Hou is with the School of Electronic and Information Engineering, Bei-
jing Jiaotong University, Beijing 100044, China (email:twhou@bjtu.edu.cn).
Y. Liu is with the School of Electronic Engineering and Computer
Science, Queen Mary University of London, London E1 4NS, UK
(email:yuanwei.liu@qmul.ac.uk).
J. Zhang is with the Department of Computing & Informatics, Bournemouth
University, Poole BH12 5BB, U.K. (E-mail: jzhang3@bournemouth.ac.uk).
C. Zhong is with the College of Information Science and
Electronic Engineering, Zhejiang University, Hangzhou 310007, China
(email:caijunzhong@zju.edu.cn).
However, considering the PLS for the RIS-aided NOMA
network, in which both the direct link and reflected links exist,
still have few reported in the literature.
We first put forward a new RIS-aided NOMA network, in
which a BS communicates with 𝑀 LUs and an eavesdrop-
per (Eve) via the assisted of the RIS. It is worth noting that
most of the literature, such as [6], [7], the designs of RIS
for enhancing the performance of LU. However, we propose
a new design of the RIS for eliminating the signals received
by the Eve to enhance the secrecy performance, which gives a
new direction for the PLS design of communication networks.
II. SYSTEM MODEL
We consider a secure downlink (DL) RIS-aided NOMA
network which includes a BS, an RIS, 𝑀 LUs and an Eve.
In the network considered, there are direct links between the
BS and LUs as well as Eve, and the BS also communicates
with LUs and Eve by means of the RIS. We assume that
the RIS has 𝐾 (𝐾 is large enough
1
) elements, positioned at
the appropriate location. Similar to [8], we assume that the
Eve has a powerful detection capability. We also assume that
the BS and all the users, include LUs and Eve, are equipped
with a single antenna. Moreover, we consider that only two
randomly chosen users can share an orthogonal resource block
to communicate with the BS in NOMA.
For simplicity, we assume that the selected NOMA users
and the Eve are denoted by 𝑚, 𝑛 (𝑚, 𝑛 ∈
(
1, 2, ··· , 𝑀
)
) and
𝑒, respectively. The distances between user 𝑖 (𝑖 ∈ {𝑚, 𝑛, 𝑒})
and the BS as well as the RIS are denoted by 𝑑
𝑑,𝑖
and 𝑑
𝑟 ,𝑖
,
respectively. Practically, the positions of the BS and the RIS
are settled. Therefore, we consider that the distance between
the BS and the RIS is fixed, denoted by 𝑑
1
. Therefore, the
large-scale fading of the reflected links for user 𝑖 can be
expressed by 𝐿
𝑖
= 𝑑
−𝛼
1
1
𝑑
−𝛼
𝑟 ,𝑖
𝑟 ,𝑖
, where 𝛼
1
and 𝛼
𝑟 ,𝑖
are the path
loss exponents.
The small-scale fading is denoted by h
𝑟
to describe
the channel between the BS and the RIS, where h
𝑟
=
[ℎ
𝑟 ,1
, ℎ
𝑟 ,2
, . . . , ℎ
𝑟 ,𝐾
]
𝑇
is a 𝐾 × 1 vector, whose elements
follow the Nakagami-𝑚 distribution with fading parameter 𝑡
1
.
Moreover, the small-scale fading between user 𝑖 and the RIS
can be expressed by g
𝑖
= [𝑔
𝑖,1
, 𝑔
𝑖,2
, . . . , 𝑔
𝑖,𝐾
], whose elements
follow the Nakagami-𝑚 distribution with fading parameter 𝑡
𝑟 ,𝑖
.
Because of the complicated scattering environment, the direct
1
In order to eliminate the signal received at Eve, similar to Lemma 1 in [4],
the number of RIS elements needs to meet the condition of 𝐾
2
≥ 𝑑
−𝛼
𝑑,𝑒
𝑑,𝑒
/𝐿
𝑒
,
which is beyond our research content of this letter.
2
links between user 𝑖 and the BS follow the Rayleigh fading,
are denoted by ℎ
𝑖
, and we assume that ℎ
𝑚
, ℎ
𝑛
and ℎ
𝑒
are
independent and unrelated.
The BS sends s =
√
𝑎
𝑚
𝑠
𝑚
+
√
𝑎
𝑛
𝑠
𝑛
to the paired NOMA
users, where 𝑠
𝑚
and 𝑠
𝑛
denote the signal aim to user 𝑚
and user 𝑛, respectively. While
√
𝑎
𝑚
and
√
𝑎
𝑛
representing
the power allocation factors, respectively. Therefor, the signal
received from the BS for user 𝑖 can be expressed by
𝑦
𝑖
=
g
𝑖
Φh
𝑟
𝐿
𝑖
+ ℎ
𝑖
𝑑
−𝛼
𝑑,𝑖
𝑑,𝑖
𝑃s + 𝑁
𝑖
, (1)
where 𝑃 is the transmit power of the BS, Φ ,
diag[𝛽
1
𝜙
1
, 𝛽
2
𝜙
2
, . . . , 𝛽
𝐾
𝜙
𝐾
] is the response matrix of the
RIS, which includes the phase shift 𝜙
𝑘
and the amplitude re-
flection coefficient 𝛽
𝑘
, and 𝛼
𝑑,𝑖
denotes the path loss exponent
of the BS to user 𝑖. Finally, 𝑁
𝑖
∼
0, 𝜎
2
𝑖
denotes the additive
white Gaussian noise (AWGN) at user 𝑖.
III. ELIMINATING THE THREAT OF EVE WITH RIS
In this section, we force our attention on the design of RIS
to eliminate the risk of information be eavesdropped, then
we analyze its secrecy performance. Specifically, in order to
simultaneously control the RIS, we consider the global CSI,
as in [9], can be perfectly available.
A. RIS Design
In this subsection, we force our attention to design the
response matrix of the RIS. The channel gain of the 𝑖-th user
can be expressed by
|
˜
ℎ
𝑖
|
2
=
g
𝑖
Φh
𝑟
𝐿
𝑖
+ ℎ
𝑖
𝑑
−𝛼
𝑑,𝑖
𝑑,𝑖
2
. (2)
To enhance the network’s secrecy rate, we can reach it
from the next two aspects: 1) by enhancing the capacity of
LUs; 2) by reducing the capacity of Eve. In this letter, we
consider the second way to enhance the PLS of RIS-aided
NOMA network. The design of the RIS as follows:
g
𝑚
Φh
𝑟
𝐿
𝑚
= 0, (3a)
g
𝑛
Φh
𝑟
𝐿
𝑛
= 0, (3b)
g
𝑒
Φh
𝑟
𝐿
𝑒
= −ℎ
𝑒
𝑑
−𝛼
𝑑,𝑒
𝑑,𝑒
. (3c)
It is worth mentioning that since the design of (3c) is for
eliminating the signals received by the Eve. Therefore, it is
difficult to evaluate the channel gains of LUs. In order to
prevent the direct signals and reflected signals received at LUs
mutual elimination, we propose a feasible proposal as (3a)
and (3b).
Therefore, we can rewrite (3) as
˜
H
D
˜
Φ = b, (4)
where b =
h
0, 0, −ℎ
𝐸
𝑑
−𝛼
𝐸
𝐸
i
T
,
˜
Φ = Φ[1, ··· , 1]
T
and
˜
H
D
=
𝑔
𝑚,1
ℎ
𝑟 ,1
√
𝐿
𝑚
··· 𝑔
𝑚,𝐾
ℎ
𝑟 ,𝐾
√
𝐿
𝑚
𝑔
𝑛,1
ℎ
𝑟 ,1
√
𝐿
𝑛
··· 𝑔
𝑛,𝐾
ℎ
𝑟 ,𝐾
√
𝐿
𝑛
𝑔
𝑒,1
ℎ
𝑟 ,1
√
𝐿
𝑒
··· 𝑔
𝑒,𝐾
ℎ
𝑟 ,𝐾
√
𝐿
𝑒
. (5)
To reach the targets design purpose, the global solution
of (4) can be expressed as
˜
Φ = pinv(
˜
H
D
)b, (6)
where pinv(
˜
H
D
) represents the pseudo-inverse of the matrix
˜
H
D
. Thus, we can obtain Φ = diag(
˜
Φ).
B. New Channel Statistics
Let us denote
ˆ
𝜆
𝑚
and
ˆ
𝜆
𝑛
are the channel gains of user 𝑚
and 𝑛, respectively. Generality, we assume that
ˆ
𝜆
1
≤ ··· ≤
ˆ
𝜆
𝑚
≤ ··· ≤
ˆ
𝜆
𝑛
··· ≤
ˆ
𝜆
𝑀
. Substituting (3a) and (3b) into (1),
the instantaneous signal-to-noise ratio (SNR) of user 𝑚 and 𝑛
can be written as
𝛾
𝑚
=
𝑎
𝑚
ˆ
𝜆
𝑚
𝑎
𝑛
ˆ
𝜆
𝑚
+
1
𝜌
𝑏
, (7)
𝛾
𝑛
= 𝜌
𝑏
𝑎
𝑛
ˆ
𝜆
𝑛
, (8)
respectively, where 𝜌
𝑏
=
𝑃
𝜎
2
𝑖
is the transmit SNR.
Lemma 1. Based on the design of the RIS, the cumulative
distribution function (CDF) of 𝛾
𝑛
can be written as
𝐹
𝛾
𝑛
(𝑥) = 𝜑
𝑛
𝑀 −𝑛
𝑞=0
𝑀 − 𝑛
𝑞
(−1)
𝑞
𝑛 + 𝑞
1 − 𝑒
−
𝜖
𝑛
𝑥
𝑎
𝑛
𝜌
𝑏
𝑛+𝑞
, (9)
where 𝜑
𝑛
=
𝑀 !
(𝑀−𝑛)!(𝑛−1)!
and 𝜖
𝑛
= 𝑑
𝛼
𝑑,𝑛
𝑑,𝑛
.
Proof. Denote 𝜆
𝑛
as the disordered channel gain of the BS
to user 𝑛, and it is exponentially distributed random vari-
ables (RVs) with parameter 𝜖
𝑛
= 𝑑
𝛼
𝑑,𝑛
𝑑,𝑛
. The CDF of 𝐹
𝜆
𝑛
(𝑥)
is
𝐹
𝜆
𝑛
(𝑥) = 1 − 𝑒
−
𝜖
𝑛
𝑥
𝑎
𝑛
𝜌
𝑏
. (10)
Let
ˆ
𝜆
𝑛
as the ordered channel gain of the BS to the 𝑛-th
user links. Based on [8], we have
𝐹
ˆ
𝜆
𝑛
(𝑥) = 𝜑
𝑛
𝑀 −𝑛
𝑞=0
𝑀 − 𝑛
𝑞
(−1)
𝑞
𝑛 + 𝑞
𝐹
𝜆
𝑛
(𝑥)
𝑛+𝑞
. (11)
By substituting (10) into (11), (9) can be obtained.
Lemma 2. Based on the design of the RIS, the CDF of 𝛾
𝑚
is
𝐹
𝛾
𝑚
(𝑥) = 𝐻 (𝑥 −
𝑎
𝑚
𝑎
𝑛
) + 𝐻(
𝑎
𝑚
𝑎
𝑛
− 𝑥)𝜑
𝑚
×
𝑀 −𝑚
𝑞=0
𝑀 − 𝑚
𝑞
(−1)
𝑞
𝑚 + 𝑞
1 − 𝑒
−
𝜖
𝑚
𝑥
(
𝑎
𝑚
−𝑎
𝑛
𝑥
)
𝜌
𝑏
𝑚+𝑞
,
(12)
where 𝜑
𝑚
=
𝑀 !
(𝑀−𝑚)!(𝑚−1)!
, 𝜖
𝑚
= 𝑑
𝛼
𝑑,𝑚
𝑑,𝑚
and
𝐻(𝑥) =
(
1, 𝑥 > 0
0, 𝑥 ≤ 0
is the unit step function.
Proof. The CDF of 𝐹
𝛾
𝑚
(𝑥) can be expressed as
𝐹
𝛾
𝑚
(𝑥) =
𝑃𝑟
ˆ
𝜆
𝑚
<
𝑥
𝜌
𝑏
(𝑎
𝑚
− 𝑎
𝑛
𝑥)
| {z }
,
Φ
𝐵
𝑥 <
𝑎
𝑚
𝑎
𝑛
1, 𝑥 ≥
𝑎
𝑚
𝑎
𝑛
.
(13)
3
Let us denote 𝜆
𝑚
as the disordered channel gain of the BS
to the 𝑚-th user links, and it is exponentially distributed RVs
with parameter 𝜖
𝑚
= 𝑑
𝛼
𝑑,𝑚
𝑑,𝑚
. In the case of 𝑥 <
𝑎
𝑚
𝑎
𝑛
, the CDF
of 𝐹
𝛾
𝑚
(𝑥) is
𝐹
𝜆
𝑚
(𝑥) = 1 − 𝑒
−
𝜖
𝑚
𝑥
(
𝑎
𝑚
−𝑎
𝑛
𝑥
)
𝜌
𝑏
. (14)
Based on [8], Φ
𝐵
can be expressed as
Φ
𝐵
= 𝜑
𝑚
𝑀 −𝑚
𝑞=0
𝑀 − 𝑚
𝑞
(−1)
𝑞
𝑚 + 𝑞
𝐹
𝜆
𝑚
(𝑥)
𝑚+𝑞
.
(15)
By substituting (15) and (14) into (13), then through the
medium of the unit step function, (12) can be obtained.
Remark 1. Based on the design of the RIS, the capacity of
the Eve is zero.
C. Performance Analysis
According to the network described above, the capacity of
user ℓ can be expressed as 𝐶
𝑈
ℓ
= log
2
(1 + 𝛾
ℓ
). While the
capacity of the Eve, based on the design of the RIS, is given
by 𝐶
𝑒,ℓ
= log
2
(1 +𝛾
𝑒,ℓ
). The secrecy rate of user ℓ is defined
by 𝐶
ℓ
= [𝐶
𝑈
ℓ
− 𝐶
𝐸
ℓ
]
+
, where [𝑥]
+
= max[𝑥, 0]. The SOP is
defined by the probability that the secrecy rate less than 𝑅
ℓ
,
who is a given secrecy rate.
Under the proposed design of the RIS, the capacity of
eavesdropping channel is forced to zero, there is no secrecy
issue at all. However, based on the definitions of secrecy rate
and SOP, we can consider that the outage probability is a
special case of SOP when the capacity of the eavesdropper
and 𝑅
ℓ
are zero.
To compare with the secrecy outage probability, we define
a special outage probability.
Definition 1. Special outage probability is the probability
when the capacity of user ℓ less than 𝑅
ℓ
.
Theorem 1. Given the ordered channel gains, the special
outage probability of user 𝑛 is
𝑃
𝑛
(𝑅
𝑛
) = 𝜑
𝑛
𝑀 −𝑛
𝑞=0
𝑀 − 𝑛
𝑞
(−1)
𝑞
𝑛 + 𝑞
(
1 − 𝑒
−𝑦
𝑛
)
𝑛+𝑞
, (16)
where 𝑦
𝑛
=
𝜖
𝑛
(
2
𝑅
𝑛
−1
)
𝑎
𝑛
𝜌
𝑏
.
Proof. Based on the definition of special outage probability
and Lemma 1, it is easy to obtain (16).
Theorem 2. Given the ordered channel gains, in the case of
𝑎
𝑚
𝑎
𝑛
> 2
𝑅
𝑚
− 1, the special outage probability of user 𝑚 is
𝑃
𝑚
(𝑅
𝑚
) = 𝜑
𝑚
𝑀 −𝑚
𝑞=0
𝑀 − 𝑚
𝑞
(−1)
𝑞
𝑚 + 𝑞
(
1 − 𝑒
−𝑦
𝑚
)
𝑚+𝑞
, (17)
where 𝑦
𝑚
=
𝜖
𝑚
(
2
𝑅
𝑚
−1
)
(
𝑎
𝑚
−𝑎
𝑛
(
2
𝑅
𝑚
−1
))
𝜌
𝑏
.
Proof. The proof is akin to Theorem 1.
To derive the diversity order, we have 𝑑
𝑠
=
− lim
𝜌
𝑏
→∞
log 𝑃
∞
/log 𝜌
𝑏
, where 𝑃
∞
denotes the asymptotic
special outage probability.
Corollary 1. The asymptotic special outage probability of the
user 𝑛 is
𝑃
∞
𝑛
(
𝑅
𝑛
)
=
𝜑
𝑛
𝑛
𝑦
𝑛
𝑛
.
(18)
Proof. In the case of 𝑦 → 0, we have 1−𝑒
−𝑦
≈ 𝑦. By applying
this result to (16), (18) can be obtained.
Remark 2. By substituting (18) into 𝑑
𝑠
, the diversity order of
user 𝑛 can be obtained as 𝑛.
Corollary 2. The asymptotic special outage probability of user
𝑚 is
𝑃
∞
𝑚
(
𝑅
𝑚
)
=
𝜑
𝑚
𝑚
𝑦
𝑚
𝑚
.
(19)
Proof. The proof is akin to Corollary 1.
Remark 3. By substituting (19) into 𝑑
𝑠
, the diversity order of
user 𝑚 can be obtained as 𝑚.
It is worth noting that, based on the design of the RIS,
there are no influences on our results by the reflected links and
the number of RIS elements. Compared with the transmission
strategy proposed in [10], the design of the RIS ensures that
the BS can always transmit signals without secrecy outage.
The proposed design of the RIS provides insightful guidelines
for a novel solution when the channel gains of the Eve are
better than that of LUs. Moreover, the proposed design of the
RIS not only can solve the issue that increasing the number
of RIS elements harms the secrecy performance [7], but also
can avoid the need for wiretap codes [11], and the system is
able to use conventional channel codes to achieve secrecy.
IV. NUMERICAL RESULTS
In this section, we show the performance of the proposed
method for enhancing the PLS. We assumed 𝑎
𝑚
= 0.6, 𝑎
𝑛
=
0.4, and 𝐾 = 20. The bandwidth (𝐵𝑊) is set to 1 MHz, and
𝜎
2
𝑖
= −174 + 10 log
10
(𝐵𝑊) dBm. The power attenuation at
the reference distance is set to −30 dB for each link. Targeted
rates/secrecy rates are set to 𝑅
𝑚
= 1 Mbps and 𝑅
𝑛
= 1.5 Mbps,
respectively. The path loss exponents are set to 𝛼
𝑑,𝑖
= 4 and
𝛼
1
= 𝛼
𝑟 ,𝑖
= 2.2, respectively. The distance between the BS and
the RIS is 𝑑
1
= 80m. The length of the RIS to the user 𝑚, user
𝑛 and the Eve are set to 𝑑
𝑟 ,𝑚
= 160m and 𝑑
𝑟 ,𝑛
= 80m, and
𝑑
𝑟 ,𝑒
= 160m, respectively. The length of the BS to the user 𝑚,
user 𝑛 and the Eve are set to 𝑑
𝑑,𝑚
= 150m and 𝑑
𝑑,𝑛
= 100m,
and 𝑑
𝑑,𝑒
= 200m, respectively.
As the benchmark, we consider the traditional NOMA
and signal-enhance scheme, where the RIS is deployed to
enhance the signals received by the user who has a good direct
link [12].
Fig. 1 shows the special outage probability versus the
transmit power for different distances. We can observe that
by reducing the distance between the BS and LUs result in
the decreased special outage probability. This is because that
the smaller distance of the direct link leads to a lower path
loss. Another observation is that even the channel gain of the
direct link for the LUs is poorer than that of the Eve’s, the
considered network also works.
4
-10 -5 0 5 10 15 20 25 30
P (dBm)
10
-3
10
-2
10
-1
10
0
Special outage probability
Simulation
Analysis, n=2
Analysis, m=1
Asymptotic, n=2
Asymptotic, m=1
d
d,1
=150 m
d
d,2
=250 m
d
d,1
=100 m
d
d,2
=150 m
Fig. 1: The special outage probability
versus the transmit power for different
distances in the case of 𝑀 = 2.
-10 -5 0 5 10 15 20 25 30
P (dBm)
10
-3
10
-2
10
-1
10
0
Special outage probability
Simulation
Asymptotic
exact, m=1, n=2
exact, m=1, n=3
exact, m=2, n=3
n=3
m=1
m=2
n=2
Fig. 2: The special outage probability
versus the transmit power for different
selected user pair in the case of 𝑀 = 3.
-10 -5 0 5 10 15 20
P (dBm)
10
-4
10
-3
10
-2
10
-1
10
0
Secrecy outage probability
10
-4
10
-3
10
-2
10
-1
10
0
Special outage probability
Signal-enhance, n=2
Signal-enhance, m=1
Prospsed, n=2
Prospsed, m=1
Traditional, n=2
Traditional, m=1
K=4, 6
Fig. 3: The comparison between the pro-
posed scheme and the signal-enhance
scheme, as well as traditional NOMA.
Fig. 2 plots the special outage probability versus the transmit
power for different selected user pair. It can be observed that,
for the paired NOMA users, the slope of the special outage
probability for user 𝑛 is higher than that for user 𝑚, this
is because that the channel gains of user 𝑛 is better than
the channel gains of user 𝑚, and the diversity orders of the
LUs are decided by the index of the ordered channel gains.
This phenomenon is also affirmed by the understandings in
Remark 2 and Remark 3.
Fig. 3 shows the SOP of traditional NOMA and signal-
enchance scheme, as well as the special outage probability of
the proposed scheme versus the transmit power when 𝐾 = 4
and 𝐾 = 6, and we compared the traditional NOMA and the
signal-enhance scheme with our proposed scheme. For user 𝑛,
we observe that the SOP of the signal-enhance scheme is better
than the special outage probability of the proposed scheme in
low-SNR regions, while the the special outage probability of
the proposed scheme is better than the SOP of the signal-
enhance scheme in the high-SNR regions. We also observe
that with the increase of 𝐾, the SOP of the signal-enhance
scheme decrease, while the special outage probability of the
proposed scheme for both 𝐾 = 4 and 𝐾 = 6 are equal. It is
because that the SOP tends to the floor with the transmit SNR
increasing. However, with the increase of transmit power, the
the special outage probability of proposed scheme decrease.
Moreover, the number of the RIS elements 𝐾 has no effect
on the the special outage probability of the proposed scheme.
For user 𝑚, we observe that the SOP of the signal-enhance
scheme is 1 for both 𝐾 = 4 and 𝐾 = 6. It is caused by the
following reasons: the RIS is designed to enchane the signals
received by user 𝑛, therefore it acts the same role to user 𝑚
and Eve. Moreover, the distance from the RIS to user 𝑚 and
EVe are equal. We also observe that the the special outage
probability of the proposed scheme for 𝐾 = 4 and 𝐾 = 6 are
equal. Furthermore, the SOP curves of the traditional NOMA
network are plotted for comparison. We can observe that RIS-
aided NOMA networks have superior secrecy performance
than traditional NOMA networks. It is because that the design
of RIS can eliminate the risk of eavesdropping on information
at the Eve.
V. CONCLUSION
This letter investigated the secrecy performance of RIS-
aided NOMA networks. We first proposed a novel design of
the RIS to improve the secrecy performance. Then, we derived
the analytical expressions of the special outage probability.
Also, the diversity orders were provided. Numerical results
were provided to verify the accuracy of the analytical results.
The secrecy performance of traditional NOMA and signal-
enhance RIS-aided NOMA scenarios have been compared,
which concluded that the proposed scheme has superior se-
crecy performance than traditional NOMA and signal-enhance
scheme.
REFERENCES
[1] Y. Liu, Z. Qin, M. Elkashlan, Z. Ding, A. Nallanathan, and L. Hanzo,
“Non-orthogonal multiple access for 5G and beyond,” Proc. IEEE., vol.
105, no. 12, pp. 2347–2381, Dec. 2017.
[2] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless
network via joint active and passive beamforming,” IEEE Trans. Wireless
Commun., vol. 18, no. 11, pp. 5394–5409, Aug. 2019.
[3] Y. Liu, X. Liu, X. Mu, T. Hou, J. Xu, Z. Qin, M. Di Renzo, and N. Al-
Dhahir, “Reconfigurable intelligent surfaces: Principles and opportuni-
ties,” arXiv preprint arXiv:2007.03435, Jul. 2020.
[4] T. Hou, Y. Liu, Z. Song, X. Sun, and Y. Chen, “MIMO-NOMA networks
relying on reconfigurable intelligent surface: A signal cancellation based
design,” IEEE Trans. Commun., vol. 68, no. 11, pp. 6932–6944, Nov.
2020.
[5] L. Yang, J. Yang, W. Xie, M. O. Hasna, T. Tsiftsis, and M. D. Renzo,
“Secrecy performance analysis of RIS-aided wireless communication
systems,” IEEE Trans. Veh. Technol., pp. 1–1, Jul. 2020.
[6] A. U. Makarfi, K. M. Rabie, O. Kaiwartya, K. Adhikari, X. Li,
M. Quiroz-Castellanos, and R. Kharel, “Reconfigurable intelligent
surfaces-enabled vehicular networks: A physical layer security perspec-
tive,” arXiv preprint arXiv:2004.11288, 2020.
[7] L. Yang and Y. Yuan, “Secrecy outage probability analysis for RIS-
assisted NOMA systems,” Electronics Letters, Oct. 2020.
[8] Y. Liu, Z. Qin, M. Elkashlan, Y. Gao, M. Elkashlan, and L. Hanzo,
“Enhancing the physical layer security of non-orthogonal multiple access
in large-scale networks,” IEEE Trans. Wireless Commun., vol. 16, no. 3,
pp. 1656–1672, Mar. 2017.
[9] S. Gong, S. Ma, C. Xing, Y. Li, and L. Hanzo, “Multi-antenna aided
secrecy beamforming optimization for wirelessly powered HetNets,”
IEEE Trans. Wireless Commun., vol. 19, no. 8, pp. 5261 – 5277, Aug.
2020.
[10] P. K. Gopala, L. Lai, and H. El Gamal, “On the secrecy capacity of
fading channels,” IEEE Trans. Inf. Theory, vol. 54, no. 10, pp. 4687–
4698, Oct. 2008.
[11] H. Mahdavifar and A. Vardy, “Achieving the secrecy capacity of wiretap
channels using polar codes,” IEEE Trans. Inf. Theory, vol. 57, no. 10,
pp. 6428–6443, 2011.
[12] T. Hou, Y. Liu, Z. Song, X. Sun, Y. Chen, and L. Hanzo, “Reconfig-
urable intelligent surface aided NOMA networks,” IEEE J. Sel. Areas
Commun., vol. 38, no. 11, pp. 2575–2588, Nov. 2020.