Abstract
This study examined how to improve system performance by equipping multiple antennae at a base station (BS) and all terminal users/mobile devices instead of a single antenna as in previous studies. Experimental investigations based on three NOMA downlink models involved (1) a singleinputsingleoutput (SISO) scenario in which a single antenna was equipped at a BS and for all users, (2) a multiinputsingleoutput (MISO) scenario in which multiple transmitter antennae were equipped at a BS and a single receiver antenna for all users and (3) a multiinputmultioutput (MIMO) scenario in which multiple transmitter antennae were equipped at a BS and multiple receiver antenna for all users. This study investigated and compared the outage probability (OP) and system throughput assuming all users were over Rayleigh fading channels. The individual scenarios also each had an eavesdropper. Secure system performance of the individual scenarios was therefore also investigated. In order to detect data from superimposed signals, successive interference cancellation (SIC) was deployed for users, taking into account perfect, imperfect and fully imperfect SICs. The results of analysis of users in these three scenarios were obtained in an approximate closed form by using the GaussianChebyshev quadrature method. However, the clearly and accurately presented results obtained using Monte Carlo simulations prove and verify that the MIMONOMA scenario equipped with multiple antennae significantly improved system performance.
Introduction
The explosive growth of mobile devices and the Internet of Things (IoT) is facing a trend of increased wireless network traffic in future networks. Researchers have confirmed nonorthogonal multiple access (NOMA) as the candidate to become the fifth generation (5G) wireless communication technology [1–4]. Liu et al. [5] demonstrated that the NOMA system has a better ergodic sum rate (ESR) than the orthogonal multiple access (OMA) system. The key technologies of NOMA are lower latency, enhanced fairness between users and a better efficiency spectrum, because all user equipment (UE) is served in the same time slot or frequency by sharing the spectrum with different allocation power coefficients based on the UE channel conditions in the same power domain. The base station (BS) sends a superimposed signal to all UE in the same time slot. For example, the downlink NOMA system consists of nearby UE with strong channel conditions and distant UE with poor channel conditions [6–10]. At the UE, the signals received are decoded by applying successive interference cancellation (SIC) until their own information is successfully detected [11, 12]. For example, the nearby UE decodes the data symbol of the distant UE first and then decodes its own data symbol after subtracting the decoded data symbol of the distant UE. In addition, the distant UE only decodes its own data symbol by treating the nearby UE data symbol as noise. In [6], the authors studied the outage probability (OP) and ergodic sum capacity (ESC) of NOMA systems with randomly distributed UE in the neighbourhood of the BS and verified that the performance of a NOMA system considerably outperformed an OMA system when the allocation power scheme was deployed.
In order to improve system performance, researchers have proposed many different technologies. One of them is cooperative communications, which deploys relays as an effective solution in order to combat fading. The authors studied fullduplex (FD) relays to avoid wasting time slot/frequency by replacing halfduplex (HD) relays [13]. The authors also proposed using N−1 FD relays to support the Nth users with the poorest channel conditions [14]. The authors indicated that system performance could be enhanced by increasing the m coefficient of the Nakagamim fading channels compared to the Rayleigh fading channels. As expected, in accordance with capability and reality, some wireless technologies combined with NOMA were proposed in order to scale up system performance: cooperative communication [15, 16], full duplex [17], cognitive radio (CR) [18, 19], millimetre wave [20], visible light communication [21], etc.
Other, different protocols, such as HD, FD, decodeandforward (DF) and amplifyandforward (AF) with fixed gain (FG) or variable gain (VG), were also studied in order to find a better protocol to implement with NOMA technology. In [22], the DF protocol was deployed. The advantage of DF protocol is forwarded signals without including the data symbols of the previous UE and simplicity in analysis and simulation. The authors also considered deploying AF protocol with fixed gain (FG) or variable gain (VG). However, the authors demonstrated that DF protocol is better than AF protocol depending on certain parameters, and conversely, AF protocol is better than DF with other parameters. The authors therefore proposed a mechanism for switching adaptive protocols according the parameters in order to optimize system performance.
However, previous studies have commonly assumed that only a single antenna was equipped on network nodes. Recently, researchers have proposed multiple antenna technology as a powerful option for enhancing system performance [23–26]. The authors investigated the system performance of a NOMA network with multiple antennae and an energy harvesting (EH) relay on the OP performance [27]. Although the system performance can be potentially improved by equipping more antennae, the improvement is limited by the cost of radio frequency (RF) technology at the UE. In order to avoid expensive hardware costs and keep the throughput profits from multiple antennae, a transmit antenna selection (TAS) protocol was verified and admitted as a powerful option [28]. In [5], the authors investigated OP in a dualhop relay over a MIMONOMA network with TAS and maximum ratio combining (MRC) protocols over the Rayleigh fading channels. In the results of the study, the authors recognized that the system performance could be improved increasing the number of antennae.
On the other hand, physical layer security (PLS) is a topic popular not only in wireless communications but also network security. PLS can see secret communications by exploiting the entropy and confuse time of the wireless channels without the use of an encoding algorithm [29]. Zhang et al. investigated the secrecy system performance of a SISONOMA system and verified the secrecy sum rate of a NOMA system as superior to a traditional OMA system [30]. The authors investigated the PLS of NOMA systems in massive networks where all UE and eavesdroppers were located at random positions [31] and obtained new, precise asymptotic expressions for secrecy outage probability (SOP) [32]. The authors in [33] assumed that the BS had full channel state information (CSI) in both the main channels of trusted UE and the wiretap channels of nontrusted UE and proposed optimal antenna selection (OAS) and suboptimal antenna selection (SAS) protocol schemes in order to improve the secrecy performance of a MIMO system compared to an ordinary spacetime transmission (STT) protocol. Precise asymptotic expressions in closed form for the SOP of a MIMO system with underlay was obtained in [34]. The results indicated that both SAS and OAS protocols could considerably improve secrecy performance. In [35], Lei et al. investigated the secrecy performance of two types of UE over downlink MOMA systems in which SISO and MISO schemes were applied with different TAS methods. However, the authors have assumed that the UE only had one receiver antenna. From previously studied results, we are considering an investigations of nonsecrecy outage probability (NOP) and SOP of UE over three NOMA schemes SISO, MISO and MIMO systems with TAS protocol as motivations.
Some important and recent studies similar to this research are [36–43]. In the excellent work [36], the authors investigated SOP for the cooperative NOMA (CNOMA) in CR networks and took into account the impact of distance of users from the BS on secrecy performance. Although the authors investigated CRNOMA with multiple users and multiple eavesdroppers, the power allocation (PA) coefficients for users were fixed. Our objective is to implement a PA strategy to ensure QoS for users. Zhou et al. [37] improved the secrecy performance of NOMA system based on CR networks using simultaneous wireless information and power transfer (SWIPT). Another work [38] investigated the popular PLS topic in order to find a way to minimize power over MISONOMA systems. In the work in [39], the authors fully surveyed the special issues in PLS, such as PLS fundamentals, CNOMA for PLS, cooperative jamming for PLS and hybrid CNOMA for PLS. In another work [40], the authors investigated the secrecy performance of random MIMONOMA with homogeneous Poisson point processes (HPPPs) on both the BS and users over α−μ fading channels. The authors obtained analysis results and verified them with Monte Carlo simulation results. From the obtained results, the author indicated that SOP performance was impacted by the number of users, the pathloss exponent and the number of antennae. In the work [41], the authors investigated a MIMONOMA system based on TAS protocol for two users over Nakagamim fading channels. Although the work was interesting, the authors, however, did not consider the PA issue, such as in [36]. Feng et al. [42] considered PA issue in order to maximize the QoS for strong user while guaranteeing the QoS for weak user. In another work [43], the authors investigated two sourcedestination pairs through twostage secure relay selection (TSSRS) in order to maximize the SOP of one sourcedestination pair, while guaranteeing the SOP of the other pair. The author, however, only equipped a single antenna for all source, relay, destination, and eavesdropper.
The featured study investigated certain issues that form its primary contribution:
This paper investigated and compared the system performance in different scenarios of SISO, MISO and MIMO architecture with deploying TAS protocol in order to determine which had the best quality of service (QoS) for users. Although the investigations were based on two users, this study can be extended to N users, such as in [14].
This paper investigated system performance on various criteria such as nonsecure outage probability (NOP), secure outage probability (SOP) and system throughput. Novel closed forms and approximate forms were obtained from these investigations.
This paper also investigated the impact of antennae on system performance. The impact of perfect/imperfect/fully imperfect SIC were also considered.
Finally, the analysis results were demonstrated and verified with Monte Carlo simulation results.
This study was presented in the following structure. In experimental model section, three models, SISO, MISO and MIMONOMA, considering imperfect SIC, respectively, are proposed and analysed. In the next section, system performance is analysed and the closed form expressions based on NOP, SOP and system throughput are obtained. In Section 4, the study proposes the system parameters for investigations and simulations using Monte Carlo simulations in Matlab software^{Footnote 1}. The results are presented in the figures. A detailed discussion based on the obtained results is given as figures. Finally, a summary of the study’s results is presented in the “Conclusion” section.
This paper uses some notations such as
referred the matrix.
referred the maximum function.
referred the probability.
referred the mean function.
referred probability density function (PDF).
referred cumulative distribution function (CDF).
Experimental models
This study investigated the system performance on NOP and SOP of two types of UE over three individual downlink scenarios: (1) SISO, (2) MISO and (3) MIMONOMA with the TAS protocol.
The system model proposed is shown in Figure 1. Two users wait for serving signals from the BS. The BS sends a superimposed signal \({\vartheta } = \sqrt {{P_{0}}} \left ({\sqrt {{\alpha _{{1}}}} {x_{{1}}} + \sqrt {{\alpha _{{2}}}} {x_{{2}}}} \right)\) to both U_{1} and U_{2} in the same time slot and the same power domain, where α_{1} and α_{2} are the allocation power coefficients of the users U_{1} and U_{2}, respectively. According to the terms of NOMA theory, user U_{2} with poor channel conditions was prioritized to allocate a larger power coefficient than user U_{2}, whereas α_{1}<α_{2} and α_{1}+α_{2}=1.
As a feature study, Ding et al. [6] proposed a downlink NOMA system with random χ users and proposed a PA strategy for random χ users as \({\alpha _{i}^{[6]}} = \frac {{\chi  i + 1}}{1 + \cdots + \chi }\) for i∈χ, where U_{1} to U_{χ} were the poorest user to the strongest user. But U_{1} to U_{χ} in our model were the strongest user to the poorest user. We therefore proposed a Dings’ modified PA strategy as \({\alpha _{i}^{[6]}} = \frac {{i}}{1 + \cdots + \chi }\). To simplify, for this study, we assumed χ=2. We realized, however, that this PA strategy fixed a PA coefficient depending on the number of χ users. For example, χ=2, we obtained the PA coefficients as \(\alpha _{1}^{[6]} = 0.3333\) and \(\alpha _{2}^{[6]} = 0.6667\) without considering the strong channel conditions or slight differences between the two users. Tran et al. [14] proposed the PA strategy for strongest user and poorest user as \(\alpha _{i}^{[14]} = {\sigma _{0,\chi  i + 1}^{2}} \left / {{\sum \nolimits }_{k = 1}^{\chi } {\sigma _{0,k}^{2}} } \right. \). This study therefore respectively obtained the PA factors for U_{1} and U_{2} as \({\alpha _{1}^{[14]}} = {\sigma _{0,2}^{2}} \left / {{\sum \nolimits }_{k = 1}^{2} {\sigma _{0,k}^{2}} } \right.\) and \({\alpha _{2}^{[14]}} = {\sigma _{0,1}^{2}} \left / {{\sum \nolimits }_{k = 1}^{2} {\sigma _{0,k}^{2}} } \right.\) with assuming the BS own fully CSIs. We investigated and compared both of these PA strategies as a contribution.
SISO scenario
A common scenario in previous studies, such as [6, 14], the BS and users were equipped only with a single antenna. Therefore, a single connection from the BS to each user was denoted by \(h_{0,i}^{(1,1)}\) for i={1,2,E} where the channel \(h_{0,i}^{(1,1)}\) followed \({h_{0,i}^{(1,1)}} = d_{0,i}^{ r}\), where d_{0,i} refers to the distance from the BS to U_{i} and r refers the pathloss exponent factor [44]. This study assumed all users were over Rayleigh fading channels.
The received signals at both U_{1} and U_{2} were respectively expressed as follows:
where P_{0} refers to the BS’s transmission power and the subsequent signaltonoiseratio (SNR) ρ_{0}=P_{0}/N_{0}, and n_{i} for i={1,2,E} refers to additive white Gaussian noises (AWGNs) followed by n_{i}∼CN(0,N_{0}) with zero mean and variance N_{0}.
By deploying the SIC as [43] after reversing user arrangement, the user U_{i} obtains the signaltointerferenceplusnoise ratios (SINRs) when it decodes x_{j} symbol as follows:
where i={1,2} and j={2,1}. There are two SIC phases at U_{1}. The first SIC phase obtains the SINR as (2) for i=1 and j=2 when U_{1} decodes U_{2}s’ x_{2} symbol and removes x_{2} symbol from the received signal. U_{1} then decodes its own x_{1} symbol and obtains SINS as (3) for i=j=1 at the second SIC phase. On another hand, it is important to note that the user U_{2} only detects its own x_{2} symbol by treating x_{1} symbol as noise and obtains the SINR as (3) for i=j=2. In addition, this paper assumed that the users deployed imperfect SIC [45] denoted by coefficients 0≤Λ^{2}≤1 and 0≤Δ^{2}≤1. For clarity, when U_{2} detects its own x_{2} symbol by treating x_{1} as interference, then Λ^{2}=0 refers to perfect SIC, Λ^{2}=1 refers to fully imperfect SIC, and otherwise referred imperfect SIC.
The instantaneous bit rate of U_{i} is obtained when it detects x_{j} symbol expressed as follows:
where i={1,2,E} and j={2,1}.
MISO scenario
Previous research results have indicated that system performance improved by equipping more antennae. We in this subsection assumed that BS was equipped a multiple transmitter antenna, denoted by S for S>1 and followed by matrix channel as \({\bf {H}}_{0,i}^{\left ({\text {MISO}} \right)} = \left [ {h_{0,i}^{\left ({1,1} \right)} \cdots h_{0,i}^{\left ({s,1} \right)} \cdots h_{0,i}^{\left ({S,1} \right)}} \right ]\) for s∈S. Vector transmitter antennae on BS send a broadcast beamforming superimposed signal to all users as in [46].
Therefore, the vector beamforming received signals from S antennae on the BS to the user U_{i} for i={1,2,E} are expressed as follows:
where \({h_{{0},{i}}^{\left (s,1 \right) }} \in {\bf {H}}_{0,i}^{\left ({\text {MISO}} \right)}\) refers the channel from the sth transmitter antenna for vector s=[1⋯S] on BS to the receiver antenna on U_{i} for i={1,2,E}.
The TAS protocol in this subsection deployed. The user U_{i} for i={1,2,E} therefore obtains SINRs when it detects x_{j} symbol for j={2,1} with implementing TAS protocol expressed as follows:
The instantaneous bit rate of U_{i} for i={1,2,E} obtained when it decoded the x_{j} symbol for j={2,1} expressed as follows:
MIMO scenario
We in this subsection assumed the BS and all users were equipped a multiple antenna like [23, 24]. The number antennae on BS denoted by S>1, as in the previous subsection, while the number antennae on U_{i} denoted by U>1. The NOMA network therefore existed S×U channels from S transmitter antennae at the BS to the U receiver antennae at user U_{i} is expressed as follows:
Each sth transmitter antenna on BS sends a broadcast beamforming superimposed signal to all U antennae at user U_{i}. The received signals at each user U_{i} are expressed as follows:
where \(h_{0,i}^{(s,u)} \in \bf {H}_{0,i}^{\left ({MIMO} \right)}\) is the transmission channel from sth transmitter antenna at the BS to the uth receiver antenna at user U_{i} for vector s=[1⋯S] and vector u=[1⋯U].
The TAS protocol was deployed at the user in this subsection. After selecting the best pairing antenna with one at the BS and one at U_{i}, the SINRs therefore obtained at the U_{i} when it detects the x_{j} symbol for i={1,2,E} and j={2,1} are expressed as follows:
The instantaneous bit rate of U_{i} achieved over MIMO scheme when it decodes the x_{j} symbol is expressed as follows:
System performance analysis
In this section, NOP, SOP and system throughput, respectively, is analysed. Note that the users were over Rayleigh fading channels with a respective probability density function (PDF) and cumulative distribution function (CDF) expressed as \({f_{{{\left  h_{0,i} \right }^{2}}}} = {\exp \left ({{  x} \left / {\sigma _{0,i}^{2}}\right. } \right)} \left / {\sigma _{0,i}^{2}}\right.\), and \({F_{{{\left  h_{0,i} \right }^{2}}}} = 1  \exp \left ({{  x} \left / {\sigma _{0,i}^{2}}\right.} \right)\) where x refers to a random independent variable followed by x≥0, and \({\sigma _{0,i}^{2}}\) refers to the mean of the channel followed by \(\sigma _{0,i}^{2} = E\left [ {{{\left  h_{0,i} \right }^{2}}} \right ]\).
Nonsecrecy outage probability (NOP) without eavesdropper E
Ding et al. [6] investigated a NOMA system with random χ users. In addition, the authors demonstrated the NOP at ith user for i∈χ occurred when it cannot successfully decode at least one of data symbols x_{j} for j={χ,...,i} [14]. Although, the system model (Fig. 1) in this study had only two users. However, the system model can be expanded with a massive χ users like [6] and [14]. The NOPs at the two users over three individual schemes were respectively presented in terms as follows.
Theorem 1 As shown in Fig. 1, the NOP of signal transmission at U_{i} occurred when U_{i} cannot successfully decode x_{j} symbol for i={1,2} meanwhile j={2,i}. Specifically, the NOP at U_{1} and U_{2} will occur when there is one of corresponding the following cases:
NOP at U_{1}: The NOP of signal transmission at U_{1} will occur when it cannot successfully decode either x_{2} or x_{1} symbol. For clarity, U_{1} over three individual scenarios firstly decoded the x_{2} symbol of U_{2} and obtained SINRs that are given by (2), (6), and (11), respectively. After detecting the x_{2} symbol, the U_{1} removed the x_{2} symbol from the received signal and then U_{1} decoded its own x_{1} symbol and obtained SINRs as shown in (3), (7), and (12). After SICing with assuming the SINRs obtained with implementing perfect, imperfect and fully imperfect SICs, the instantaneous bit rate thresholds then obtained when U_{1} decoded x_{j} symbol for j={2,1}, which were given by (4), (8), or (13). Next processing, the instantaneous bit rate thresholds \(R_{1  {x_{j}}}^{(\Psi)}\) for Ψ={SISO,MISO,MIMO} were compared with U_{j}s’ bit rate thresholds denoted by \(R_{j}^{*}\). The NOP at U_{1} therefore occurred as a result when the instantaneous bit rate threshold \(R_{1  {x_{j}}}^{(\Psi)}\) cannot reach to the bit rate threshold \(R_{j}^{*}\) of user U_{j} for j={2,1}. In other words, the NOP at U_{1} over three individual scenarios was then expressed as follows:
$$\begin{aligned} \Theta_{1}^{\left(\Psi \right)} &{}= 1  \prod\limits_{j = 2}^{1} {\Theta_{1  {x_{j}}}^{\left(\Psi \right)}} = 1  \Theta_{1  {x_{2}}}^{\left(\Psi \right)}\Theta_{1  {x_{1}}}^{\left(\Psi \right)} \\ \end{aligned} $$$$ \begin{aligned} & = 1  \left\lbrace \Pr \left\{ {R_{i  {x_{2}}}^{\left(\Psi \right)} \ge R_{2}^{*}} \right\} \text{ and} \Pr \left\{ {R_{i  {x_{1}}}^{\left(\Psi \right)} \ge R_{1}^{*}} \right\} \right\rbrace. \end{aligned} $$(14)It is worth to noting that the upper limit of Eq. (14) with j=2 indicates the number of current χ users have joined the network. By replacing the upper limit small value j with a massive χ, Eq. (14) can be used to investigate NOP at the strongest user over NOMA network with a massive χ user scenario like [6] or [14]. However, the aim of this study was to examined the impact of antennae on system performance, which is presented in the following sections. We were therefore limited to only χ=2 users without losing NOMA key features.
NOP at U_{2}: The NOP of signal transmission at U_{2} will occur when it cannot successfully decode its own x_{2} symbol by treating x_{1} symbol as interference, assuming that the SINRs obtained with perfect, imperfect and fully imperfect SIC, respectively, over three individual scenario as shown (2) for the SISO scenario, (6) for the MISO scenario and (11) for the MIMO scenario, where i=j=2. After SICing, the instantaneous bit rate threshold is obtained when U_{2} decodes its own symbol x_{2}, which is similarly given by (4), (8), and (13), where i=j=2. Further processing, the instantaneous bit rate thresholds \(R_{2  {x_{2}}}^{(\Psi)}\) for Ψ={SISO,MISO,MIMO} were compared with its own bit rate threshold \(R_{2}^{*}\). The NOP at U_{2} therefore occurred as a result when \(R_{2  {x_{2}}}^{(\Psi)}\) cannot reach to the bit rate threshold \(R_{2}^{*}\). In other words, the NOP at U_{2} over three individual scenarios is expressed as follows:
$$ \begin{aligned} {\Theta_{2}^{(\Psi)}} &= 1  {\Theta_{2x_{2}}^{(\Psi)}} = \Pr \left\{ {{R_{2  {x_{2}}}^{(\Psi)}} < R_{2}^{*}} \right\} \\&= 1  \Pr \left\{ {{R_{2  {x_{2}}}^{(\Psi)}} \geq R_{2}^{*}} \right\}. \end{aligned} $$(15)
NOP at users over SISO scheme
Remarks 1
Through NOP conditions as shown in (14) and (15) in Theorem 1, this study obtained the NOP at users over SISO scheme.
First, the NOP at U_{1} over SISO scenario is obtained and expressed in closed form as follows:
where \(\gamma _{j}^{*} = {2^{R_{j}^{*}}}\) for j={2,1}.
Through observation (17), it was simple to note that when the SIC was perfect, U_{1} certainly detected the x_{1} symbol but could not successfully detect its own x_{1} symbol. U_{1} therefore obtained the worst QoS. This issue was verified by analysis and simulation results shown in the next section.
Second, the NOP at U_{2} over SISO scenario is easily obtained and expressed in closed form as follows:
where Λ^{2}=0 for perfect SIC, 0<Λ^{2}<1 for imperfect SIC and Λ^{2}=1 for fully imperfect SIC. By observation (19), it was simple to note that when the SIC was perfect, U_{2} obtained the best QoS.
See Appendix for proof.
NOP at users over MISO scheme
Remarks 2
The MISO scheme in this study assumed that the BS was equipped with a multiple transmitter antenna while the users were still equipped with a single receiver antenna, such as [35]. It was important to remember that S denoted the number of transmitter antennae at the BS, where S>1. Therefore, The matrix channels from the BS to the user U_{i} are \({\bf {H}}_{0,i}^{\left (\mathrm {{MISO}} \right)} = \left [ {h_{0,i}^{\left ({1,1} \right)} \cdots h_{0,i}^{\left ({s,1} \right)} \cdots h_{0,i}^{\left ({S,1} \right)}} \right ]\) for s∈S. TAS protocol was deployed.
The NOP conditions at U_{1} over MISO scenario are given as follows:
Equation (20) is obtained in closed form as follows:
In addition, the approximation in closed form of NOP at U_{1} is obtained by using the PDF as shown in (57) and (58) for \({0 \le x < \frac {{{\Delta ^{2} \alpha _{2}}}}{{{\Lambda ^{2} \alpha _{1}}}}}\) and expressed as follows:
However, the NOP conditions at the U_{2} over MISO scenario are given as follows:
Equation (23) is obtained in closed form as follows:
Similarly, the approximation in closed form of NOP at U_{2} over MISO scenario is obtained by using the PDF as shown (57) for \({0 \le x < \frac {{{\Delta ^{2} \alpha _{2}}}}{{{\Lambda ^{2} \alpha _{1}}}}}\) and expressed as follows:
See Appendix for the proof.
NOP at users over MIMO scheme
Remarks 3
In this section, this study assumed that the BS and all users are equipped with a multiple antenna. It is important to remember that U denotes the number of receiver antennae at the users, where U>1, while S denotes the number of transmitter antennae at the BS, where S>1. Therefore, the matrix channels from the sth transmitter antenna at the BS to the uth receiver antenna at the ith user for vectors s=[1⋯S], and u=[1⋯U] are \(\bf {H}_{0,i}^{\left ({MIMO} \right)}\). The TAS protocol was deployed in this scenario.
The NOP conditions at U_{1} over MIMO scenario are given as follows:
Equation (26) is obtained in closed form as follows:
In addition, the approximation in closed form of NOP at U_{1} over MIMO scenario can be also obtained by using the PDF as shown (55) and expressed as follows:
However, the NOP conditions at U_{2} over MIMO scenario are given as follows:
Equation (29) is obtained in closed form as follows:
Similarly, the approximation in closed form of NOP at U_{2} over MIMO scenario is also obtained by using the PDF (55) and expressed as follows:
See Appendix for the proof.
Secrecy outage probability (SOP) with eavesdropper E
In this investigation, this study assumed an eavesdropper E existed in the NOMA network. Eavesdropper E over three individual scenarios received signals by substituting its own channel \({h_{0,E}^{\left (\Psi \right)}}\) into (1), (5) or (10). The eavesdropper E can also detect x_{1} or x_{2} data symbol of U_{1} or U_{2} when it eavesdrops U_{1} or U_{2}. For clarity, the SINRs are obtained at the eavesdropper E when it decodes the x_{2} symbol by substituting \({\left  {h_{0,E}^{\left (\Psi \right)}} \right ^{2}}\) into (2), (6) or (11). However, the eavesdropper E can also detect and remove x_{2} symbol from received signal and then detect the x_{1} symbol when it eavesdropped U_{1}. The instantaneous bit rate threshold of the eavesdropper E over three individual scenarios is therefore obtained from (4), (8) or (13).
The secure instantaneous bit rate of U_{i} for i={1,2} over SISO, MISO and MIMO schemes, respectively, are expressed as follows:
and
Theorem 1
SOP at the U_{i} over three individual scenarios is the probability that the secure instantaneous bit rate given by (30), (32) or (34) cannot reach the U_{i}s’ bit rate threshold \(R_{i}^{*}\). In other words, the SOP at U_{i} for i={1,2} can be respectively expressed as follows:
and
where Ψ={SISO,MISO,MIMO}.
SOP at users over SISO scheme
Remarks 4
In this subsection, this study investigated the SOP at legitimate user U_{i} for i={1,2} over SISO scheme as the system model in [42]. As with (33) in Theorem 2, the SOP at U_{1} over SISO scheme can be rewritten, solved and expressed in closed form as follows:
where \({\theta _{1}} = \frac {{\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1}}{{\left ({{\Delta ^{2} \alpha _{2}}  {\Lambda ^{2} \alpha _{1}}\left ({\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1} \right)} \right){\rho _{0}}\sigma _{0,1}^{2}}} + \frac {{{\beta }}}{{\left ({{\Delta ^{2} \alpha _{2}}  {\Lambda ^{2} \alpha _{1}}{\beta }} \right){\rho _{0}}\sigma _{0,E}^{2}}}, {\beta } = \frac {{\varpi \left ({{\lambda } + 1} \right)}}{2}, {\lambda } = \cos \left ({\frac {{2w  1}}{2W}\pi } \right)\), and \(\varpi = \frac {1}{{{\Lambda ^{2} (1\alpha _1)}\gamma _{1}^{*}}}  1+ \Xi < \frac {{{\Delta ^{2} \alpha _{2}}}}{{{\Lambda ^{2} \alpha _{1}}}}\), while Ξ is the approximate coefficient with \(0 < \Xi < \frac {{{\Delta ^{2}}{\alpha _{2}}}}{{{\Lambda ^{2}}{\alpha _{1}}}}  \left ({\frac {1}{{{\Lambda ^{2}}\left ({1  {\alpha _{1}}} \right)\gamma _{1}^{*}}}  1} \right)\) for the case of imperfect SIC, otherwise Ξ=0. It is important to note that W referred the accuracy coefficient. Meanwhile the coefficient W is an increasingly large value, the SOP analysis results at U_{1} become increasingly more accurate.
However, the SOP at U_{2} over SISO scheme is also rewritten, solved and expressed in closed form as follows:
where \({\theta _{2}} = \frac {{\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1}}{{\left ({{\Delta ^{2} \alpha _{2}}  {\Lambda ^{2} \alpha _{1}}\left ({\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1} \right)} \right){\rho _0}\sigma _{0,2}^{2}}} + \frac {{{\beta }}}{{\left ({{\Delta ^{2} \alpha _2}  {\Lambda ^{2} \alpha _1}{\beta }} \right){\rho _0}\sigma _{0,E}^{2}}}\).
See Appendix for the proof.
SOP at user over MISO scheme
Remarks 5
In this subsection, this study investigated the SOP at user U_{i} for i={1,2} over MISO scheme. As with (35) in Theorem 2, the SOP at U_{1} over MISO scheme can be rewritten and expressed as follows:
where vector s=[1⋯S].
The CDF of \(\gamma _{i  {x_j}}^{\left (\mathrm { {MISO}}\right)}\) for i={1,2,E} are respectively given by (57) for j=2 or (58) for j=1.
The PDF of \(\gamma _{i  {x_j}}^{\left (\mathrm { {MISO}}\right)}\) for i={1,2,E} and j={2,1} is respectively given by (59) or (60).
Through Eq. (41), the SOP at U_{1} can be obtained and expressed in closed form as follows:
where G_{1} was given by (70).
The SOP at U_{2} as with (36) in Theorem 2 can be obtained and expressed in closed form as follows:
where \(\Phi _{i} = \frac {{s\left ({\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1} \right)}}{{\left ({{\Delta ^{2} \alpha _2}  {\Lambda ^{2} \alpha _1}\left ({\gamma _{2}^{*}\left ({1 + {\beta }} \right)  1} \right)} \right){\rho _0}\sigma _{0,i}^{2}}} + \frac {{{\beta }}}{{\left ({{\Delta ^{2} \alpha _2}  {\Lambda ^{2} \alpha _1}{\beta }} \right){\rho _0}\sigma _{0,E}^{2}}}\) for i={1,2}, and G_{2} is given by (70) by substituting i=2.
See Appendix for the proof.
SOP at user over MIMO scheme
Remarks 6
In this subsection, this study investigated the SOP at U_{i} for i={1,2} over MIMO scenario. The users were also equipped a multiple receiver antenna. As with (35) in Theorem 2, the SOP of U_{1} therefore can therefore be obtained a nd expressed in closed form as follows:
However, the SOP at U_{2} over MIMO scenario can be obtained and expressed in closed form as follows:
See Appendix for the proof.
System throughput
The system throughput is the sum of achievable received bit rate at all U_{i} for i={1,2} which was denoted by \(P_{sys}^{\left (\Psi \right)} \) [47]. The system throughput can therefore be computed and expressed as follows:
where Ψ={SISO,MISO,MIMO}.
Numerical results and discussions
In this section, this study presents the analysis results and Monte Carlo simulation results obtained from the investigation in the previous sections. Due to Kong et al. [40] confirmed that the factors include number of users, pathloss exponent, the number of antennae impacted on system performance. We therefore set the parameters for two users U_{1} and U_{2} and an eavesdropper E, pathloss exponent r=4 [41]. We subjected the parameters to analysis and simulate as shown in Table 1.
Note that in all figures, the markers indicate the analysis results while the solid or dashed lines indicate the Monte Carlo simulation results. The simulation results were based on the investigation of 10^{6} random samples. Monte Carlo simulation results were used to compare and verify the analysis results.
Results and discussions for perfect/imperfect/fully imperfect sIC
In this subsection, this study investigated the NOP and SOP performance at the users over SISO scheme. The PA coefficients were also given by two PA strategies by [6] with (\(\alpha _{1}^{[6]} = 0.33333, \alpha _{2}^{[6]} = 0.66667\)), and [14] with (\(\alpha _{1}^{[14]} = 0.45136, \alpha _{2}^{[14]} = 0.54864\)). Figure 2a plots the NOP results at U_{1} with implementing perfect (Λ^{2}=0), imperfect (Λ^{2}=0.5) and fully imperfect (Λ^{2}=1) SICs, respectively. It is interesting to observe the results obtained through Fig. 2a. Although U_{1} implemented perfect SIC, the NOP performance at U_{1}, however, obtained the worst results. For clarity, U_{1} had two SIC phases. The first SIC phase decoded the x_{2} symbol and then removing x_{2} symbol from the received signal. For the perfect SIC case with coefficient Λ^{2}=0,U_{1} therefore easily decoded the x_{2} symbol with only interference n_{1}. In the second SIC phase, U_{1} cannot successfully decode its own x_{1} symbol due to \(\Pr \left \lbrace R_{1x_1}^{\mathrm {(SISO)}} \ge R_{1}^{*} \right \rbrace = 0\). The NOP results at U_{1} obtained through PA strategy given by [14] better than another strategy given by [6] at the SNRs ρ_{0}→∞.
Figure 2b also plots the SOP results at U_{1} and also implemented with perfect/imperfect/fully imperfect SIC at U_{1}. The SOP results at U_{1} assuming perfect SIC (Λ^{2}=0) still obtained the worst results at low SNRs ρ_{0}<40 dB. The SOP results at U_{1} assuming fully imperfect SIC (Λ^{2}=1) still obtained the best results at SNRs ρ_{0}<50 dB. However, the SOP results at U_{1} assuming imperfect SIC (Λ^{2}=0.5) obtained the best results when SNRs ρ_{0} increased, e.g. SNR ρ_{0}=60 dB. In summary, the SOP results obtained at U_{1} assuming perfect/imperfect/fully imperfect SIC were approximately together at SNRs ρ_{0}→∞.
In other investigations, Fig. 2c and d plot NOP and SOP results obtained at U_{2} based on two PA strategies by [6, 14]. Due to \(\alpha _{2}^{[6]} = 0.66667\) given by [6] bigger than \(\alpha _{2}^{[14]} = 0.54864\) given by [14], the obtained NOP results from [6] indicated with blue markers outperform the results indicated with black markers obtained from [14] at SNRs ρ_{0}. In addition, the NOP results obtained at U_{2} with the same PA strategy were approximately obtained together based on all three perfect, imperfect and fully imperfect SICs. By observation, equation (38) for (Λ^{2}=0), (Λ^{2}=0.5), and (Λ^{2}=1), we noted that U_{1} was allocated a small power factor \(\alpha _{1}^{[6]} = 0.33333\) or \(\alpha _{1}^{[14]} = 0.45136\) because the channel conditions for U_{1} was better than channel conditions for U_{2}. The x_{1} symbol therefore lightly impacted when U_{2} decoded its own x_{2} symbol by treating the x_{1} symbol as noise.
Figure 2d plots the SOP results at U_{2} assuming perfect/imperfect/fully imperfect SICs. The difference to Fig. 2b is easily seen. Assuming perfect SIC (Λ^{2}=0), the results obtained at U_{2} were the best results compared to other SICs because U_{2} implemented perfect SIC with no impact from internal interference x_{1}. Assuming imperfect (Λ^{2}=0.5) and fully imperfect (Λ^{2}=1) SICs at U_{2} because of the impact from internal interference x_{1} and external eavesdropper E, U_{2}’s SOP results from (38) therefore obtained \(\Theta _{1}^{(\text {SISO})} \to 1\) as the secrecy instantaneous bit rate reached \(\tilde R_{2  {x_2}}^{\left (\Psi \right)} \to 0\) with SNRs ρ_{0}→∞. It is worth noting that the results of analysis given by (38) plotted with various markers were proved and verified using Monte Carlo simulation results given by (37) and plotted with black solid and blue crossedsolid lines.
Results and discussions for the NOP and SOP
From the analysis and simulation results obtained in the previous subsection 4.1, we observed that fully imperfect SIC showed a balanced QoS between U_{2} and U_{1}. The investigations are therefore discussed below assuming fully imperfect (Λ^{2}=1) at the users.
Figure 3a plots the NOP results at U_{1} over the three SISO, MISO and MIMO scenarios in four investigations as follows: a SISO scenario equipped with a single antenna at the BS and users (S=U=1); a MISO scenario equipped with a double transmitter antenna at the BS (S=2) and a single receiver antenna at the users (U=1); another MISO scenario equipped with a triple transmitter antenna at the BS (S=3) and a single receiver antenna at the users (U=1); and a MIMO scenario equipped with a double antenna at the BS and users (S=U=2). From the analysis and simulations results, the NOMA system performance progressively improved by equipping more antennae. For example, the MISO scenario with S=3 obtained better performance than the MISO scenario with S=2. However, it was interesting when the MIMO system with S=2 obtained better performance than the MISO system with S=3. For clarity, the MISO scheme with S=3 and U=1 had three channels from the BS to each user, while the MIMO scenario with S=2 and U=2 had four channels from the BS to each user. The TAS protocols over the MIMO scheme selected the best channel from four channels, while only three channels over the MISO scheme. The different markers plot the results of analysis given by (17), (21) and (27) for the SISO, MISO and MIMO scenarios, respectively. The different crossed markers plot the approximated results given by (22) and (28). The other lines indicate the Monte Carlo simulation results. The analysis and approximated results are generally close and were verified by the Monte Carlo simulation results. Monte Carlo simulations were investigated based on the statistical results of 10^{6} experimental iterations, as in previous studies. The analysis, approximation and simulation results matched closely, as shown in Fig. 3a.
In addition, Fig. 3b plots the SOP results at U_{1} over three individual schemes with the same parameters as Table 1. However, this investigation included an eavesdropper E. It is easy to observe that the diamond markers in Fig. 3b are close to the SOP results at U_{2} based on fully imperfect SIC, plotted as circle markers in Fig. 2b. The initial impact of an eavesdropper on system performance was negligible, e.g. at low SNRs ρ_{0}≤40 dB, because the eavesdropper found it difficult to successfully decode the x_{1} symbol at U_{1} when it eavesdropped U_{1}. As the SNRs ρ_{0}→∞ increased, the system performance deteriorated because the eavesdropper E easily decoded the x_{1} symbol at U_{1}. In all investigations, the MIMO scenario with S=U=2 obtained better results than the other scenarios. The SOP analysis results were plotted with various markers given by (38), (42) and (46) for the SISO, MISO and MIMO scenarios, respectively, while the various lines plot the Monte Carlo simulation results given by (37), (41) and (45). Monte Carlo simulations, in particular, were investigated based on the statistical results of 10^{7} experimental iterations instead of only 10^{6} experimental iterations as in previous investigation, see Fig. 3a. Due to the existence of the eavesdropper E, statistics over only 10^{6} samples were not guaranteed to be accurate.
This paper also investigated the NOP and SOP at U_{2} over three individual SISO, MISO and MIMO scenarios as shown in Figs. 4a and b, respectively. The NOP and SOP results at U_{2} were indicated with diamond markers in Fig. 4a and b, the results assuming fully imperfect SIC, plotted with circle markers as shown in Fig. 2c and d. QoS at U_{2} over the MIMO scenario significantly improved compared to its results in other scenarios. Figure 4a plots various markers for NOP analysis results at U_{2} given by (19), (24) and (30) for SISO, MISO and MIMO scenarios, respectively, while crossed markers show the approximated results given by (25) and (31). The various lines plot the Monte Carlo simulation results given by (18), (23) and (29), respectively. The NOP results at U_{2} obtained \(\Theta _{2}^{\left (\Psi \right)} \to 0\) when ρ_{0}→∞.
Figure 4b shows the plotted results for SOP at U_{2} over three individual scenarios and indicate that the MIMO scenario significantly improved QoS at U_{2} at low SNRs, e.g. ρ_{0}<40 dB. As SNRs ρ_{0} increased to 40 dB and over, the SOP performance at U_{2} over all three scenarios developed progressively worse results, e.g. SNRs ρ_{0}≥40 dB, and then approximated each other for SNRs ρ_{0}≥100 dB. At high SNRs, e.g. ρ_{0}>100 dB, SOP at U_{2} over the three individual scenarios tended to 100% outage (\(\tilde \Theta _{2}^{\left (\Psi \right)} \to 1\)) as a result of the secrecy instantaneous bit rate threshold tending to zero (\(\tilde R_{2  {x_2}}^{\left (\Psi \right)} \to 0\)) and the instantaneous bit rate threshold of the eavesdropper E tending to 1 (\(R_{E  {x_2}}^{\left (\Psi \right)} \to 1\)) when it eavesdropped U_{2} with SNRs ρ_{0}→∞. The results of analysis for the three individual scenarios plotted by various markers given by (40), (44) and (48) were also proved and verified using Monte Carlo simulation results given by (39), (43) and (47), plotted as various lines.
Results and discussions for the system throughput
From the results obtained for NOP and SOP at U_{1} and U_{2} over three individual scenarios, we plotted the achievable throughput and secrecy throughput for both U_{1} and U_{2}, shown in Figs. 5a, b and 6a, b. Figure 5a and b plot the throughput and secure throughput of U_{1} over three individual scenarios, while Figs. 6a and b plot the throughput and secure throughput of U_{2} over the same scenarios. We can easily see that the throughput results achieved at U_{i} over the MIMO scenario were slightly better, because the NOP results at U_{i} over the MIMO scenario obtained better results than other scenarios. As SNR ρ_{0}→∞ increased, the NOP results therefore tended approximately to zero (\(\Theta _{i}^{\left (\Psi \right)} \to 0\)), as shown in Figs. 3a and 4a. Throughput therefore tended to the user’s bit rate threshold \(R_{i}^{(*)}=0.1\) bit per channel user (BPCU), as shown in Figs. 5a and 6a.
Figures 5b and 6b plot the secrecy throughput of U_{i} over the three individual scenarios. However, it was interesting to observe in Fig. 6b that secrecy throughput improved at U_{2} as SNR ρ_{0}→40 dB increased and thereafter reduced to approximately zero instead of tending toward its bit rate threshold \(R_{i}^{(*)}\), as in Figs. 5a and b. Because of the eavesdropper E, the approximately obtained instantaneous bit rate threshold at U_{2} for SNRs ρ_{0}→∞, the SOP results at U_{2} over three individual scenarios tended to 100% outage (\(\tilde \Theta _{2}^{\left (\Psi \right)} \to 1\)) while the secrecy bit rate threshold at U_{2} therefore tended to zero (\(\tilde R_{2  {x_2}}^{\left (\Psi \right)} \to 0\)).
Results and discussions for the impacts of antennae
In this section, the impact of antennae on system performance is investigated. All the parameters of investigation in Table 1 were reapplied, however, with fixed SNRs ρ_{0}={20,50} dB.
Figure 7a and b plot the SOP and NOP results at U_{1} and U_{2}, respectively. As a contribution, it is worth noting that the BS, U_{1} and U_{2} were equipped with multiantenna technology differentiated from each other instead of being equipped with the same number of antennae as in the previous studies [23] with M=N=3 or [24] with M=N=2, where M and N denote the number of antennae at the BS and users, respectively. By observation, we can conclude that the system performance improved by equipping more antennae at either the BS or user or both. In addition, the SNR ρ_{0} also significantly impacted the system’s performance. For example, the NOP and SOP results at users obtained with SNR ρ_{0}=50 dB outperformed the results at SNR ρ_{0}=20 dB. Figure 8a and b plot the throughput and secrecy throughput results at the users over the three individual scenarios based on the results as shown in Figs. 7a and b.
Conclusion
In this paper, a MINONOMA system was proposed equipping multiple antennae not only at the BS but also at all users. The TAS protocol was also deployed. An analysis and approximation of NOP and SOP at the users were investigated and the results obtained were expressed in closed form, which were proved and verified using Monte Carlo simulation results based on 10^{6} random samples of experiments. In the SOP results, the secrecy system performance was impacted because of an eavesdropper. However, the analysis, approximation and simulation results indicated that secrecy system performance can be significantly enhanced by increasing the number of antennae or the SNRs. This paper therefore demonstrated that multiple antennae combined with the TAS protocol and reasonable PA were an effective strategy for improving secrecy system performance and resisting eavesdropping.
Appendix
The proof of Remark 1
The CDF of \({\gamma _{i  {x_j}}^{\left (\mathrm { {SISO} }\right)}}\) where i={1,2,E} and j={2,1} can be respectively expressed as follows:
and,
However, the PDF of \({\gamma _{i  {x_j}}^{\left (\mathrm { {SISO}}\right)}}\) where i={1,2,E} and j={2,1} can be also respectively expressed as follows:
and
By substituting (2) or (3) into (4) and combining with the outage conditions in (16) or (18), we obtain the expressions as follows:
By applying the PDF, the NOP at U_{i} for i={1,2} obtained when it cannot successfully detect the x_{j} symbol for j={2,1} is expressed as
By substituting (55) and (56) for i=1 and j={2,1} into (16), we easily obtain the closed form of the NOP at U_{1} over SISO scheme as shown in (17). Similarly, the NOP at U_{2} over SISO scheme can be obtained in closed form as shown in (19) by substituting (55) for i=j=2 into (18).
The proof of Remark 2
The CDF of \({\gamma _{i  {x_j}}^{\left (\mathrm { {MISO} }\right)}}\), where i={1,2,E},j={2,1} and s=[1⋯S], can be respectively expressed as follows:
and
However, the PDF of \(\gamma _{i  {x_j}}^{\left (\mathrm { {MISO} }\right)}\) for i={1,2,E} and j={2,1} is expressed as follows:
By substituting (6) or (7) into (8) and combining with the outage conditions in (20) or (23), we obtain the expression as follows:
The NOP at U_{i} for i={1,2} obtained when it cannot successfully decode the x_{j} symbol for j={2,1} expressed as follows:
By substituting (62), where i=1 and j={2,1} into (20), we easily obtain the closed form of the NOP at U_{1} over MISO scheme as shown in (21). Similarly, the NOP at U_{2} over MISO scheme can be also obtained in closed form as shown in (24) by substituting (62), where i=2 and j=2, into (23).
The proof of Remark 3
By substituting (11) or (12) into (13) and combining with the outage conditions in (26) or (29), we obtain the expression as follows:
The NOP at U_{i} for i={1,2} when it cannot successfully decode the x_{j} symbol for j={2,1} over MIMO scheme is expressed as follows:
By substituting (64), where i=1 and j={2,1}, into (26), we easily obtain the closed form of the NOP at U_{1} over MIMO scheme as shown in (27). Similarly, the NOP at U_{2} over MIMO scheme can be also obtained in closed form as shown in (30) by substituting (64), where i=2 and j=2, into (29).
The proof of Remark 4
By substituting (4) for i={1,E} and j=1 into (32) and combining secure outage conditions in (37), we obtain the SOP at U_{1} when it cannot successfully detect x_{1} symbol expressed as follows:
By applying the PDF, Eq. (65), where \(R_{1  {x_1}}^{(\mathrm { {SISO}})} > R_{E  {x_1}}^{(\mathrm { {SISO}})}\), can be solved and expressed in closed form as follows:
Similarly, we obtain the SOP at U_{i} when it cannot successfully detect the x_{2} symbol by substituting (4) into (32) and combining secure outage conditions in (37) or (39) as follows:
We attempted to solve (67) using integrals with condition \(R_{1  {x_1}}^{(\mathrm { {SISO}})} > R_{E  {x_1}}^{(\mathrm { {SISO}})}\). However, it is difficult to obtain the SOP at U_{i} when it cannot decode x_{2} symbol in closed form. The authors in [35] proposed applying the GaussianChebyshev quadrature method to obtain an approximation expression. The SOP at U_{i} when it cannot decode the x_{2} symbol therefore obtained and expressed as follows:
The proof of Remark 5
By using the CDF (57–58), and PDF (59–60), the SOP at U_{1} is expressed as follows:
where \(\Omega = \gamma _{2}^{*}\left ({1 + x} \right)  1\), and μ=Δ^{2}α_{2}−Λ^{2}α_{1}x.
Using CDF (57) and GaussianChebyshev quadrature, G_{i} for i={1,2} can be obtained as follows:
By substituting (70) into (69), we obtained the closed form of the SOP at U_{1} over the MISO scheme as shown in(42). From (70), we can also obtain the closed form of the SOP at U_{2} over the MISO scheme as shown in the (44) with outage conditions as shown in (43).
The proof of Remark 6
From (57), the CDF of \(\gamma _{i  {x_2}}^{\left (\text {MIMO} \right)}\) was rewritten and expressed as follows:
The SOP at U_{1} is expressed as follows:
Using CDF (72) and GaussianChebyshev quadrature method, K_{i} for i={1,2} can be obtained and expressed as follows:
By substituting (73) into (72), we obtained the closed form of the SOP at U_{1} over the MIMO scheme as shown in(46). From (73), we can also obtain the closed form of the SOP at U_{2} over the MIMO scheme as shown in the (48) with outage conditions as shown in (47).
Availability of data and materials
Data sharing is not applicable to this article as no datasets were generated or analysed during the study.
Notes
This study used Matlab software version R2017b made by The MathWorks, Inc. based at 3 Apple Hill Drive Natick, MA 01760 USA 5086477000.
Abbreviations
 5G:

Fifth generation network
 BS:

Base station
 CDF:

Cumulative distribution function
 MIMO:

Cumulative distribution function
 MISO:

Multiinputsingleoutput
 NOMA:

Nonorthogonal multiple access
 NOP:

Nonsecrecy outage probability
 PA:

Power allocation
 PDF:

Probability density function
 PLS:

Physical layer security
 SIC:

Successive interference cancellation
 SINRs:

Signaltointerferenceplusnoiseratios
 SISO:

Singleinputsingleoutput
 SNRs:

Signaltonoiseratios
 SOP:

Secrecy outage probability
 TAS:

Transmit antenna selection
References
Z. Ding, et al., Application of nonorthogonal multiple access in LTE and 5G networks. IEEE Commun. Mag.55(2), 185–191 (2017). https://doi.org/10.1109/mcom.2017.1500657cm.
L. Dai, B. Wang, Y. Yuan, S. C. I. Han, Z. Wang, Nonorthogonal multiple access for 5G: solutions, challenges, opportunities, and future research trends. IEEE Commun. Mag.53(9), 74–81 (2015). https://doi.org/10.1109/mcom.2015.7263349.
Y. Saito, A. Benjebbour, Y. Kishiyama, T. Nakamura, in 2015 IEEE 81st Vehicular Technology Conference (VTC Spring). Systemlevel performance of downlink nonorthogonal multiple access (NOMA) under various environments (Glasgow, 2015), pp. 1–5. https://doi.org/10.1109/VTCSpring.2015.7146120.
Z. Ding, P. Fan, H. Poor, Impact of user pairing on 5G nonorthogonal multipleaccess downlink transmissions. IEEE Trans. Veh. Technol.65(8), 6010–6023 (2016). https://doi.org/10.1109/tvt.2015.2480766.
Y. Liu, G. Pan, H. Zhang, M. Song, On the capacity comparison between MIMONOMA and MIMOOMA. IEEE Access. 4:, 2123–2129 (2016). https://doi.org/10.1109/access.2016.2563462.
Z. Ding, Z. Yang, P. Fan, H. Poor, On the performance of nonorthogonal multiple access in 5G systems with randomly deployed users. IEEE Signal Process. Lett.21(12), 1501–1505 (2014). https://doi.org/10.1109/lsp.2014.2343971.
S. Timotheou, I. Krikidis, Fairness for NonOrthogonal Multiple Access in 5G Systems. IEEE Signal Process. Lett.22(10), 1647–1651 (2015). https://doi.org/10.1109/lsp.2015.2417119.
J. Cui, Z. Ding, P. Fan, A novel power allocation scheme under outage constraints in NOMA systems. IEEE Signal Process. Lett.23(9), 1226–1230 (2016). https://doi.org/10.1109/lsp.2016.2591561.
Z. Yang, Z. Ding, P. Fan, N. AlDhahir, A general power allocation scheme to guarantee quality of service in downlink and uplink NOMA systems. IEEE Trans. Wirel. Commun.15(11), 7244–7257 (2016). https://doi.org/10.1109/twc.2016.2599521.
S. Islam, N. Avazov, O. Dobre, K. Kwak, Powerdomain nonorthogonal multiple access (NOMA) in 5G systems: potentials and challenges. IEEE Commun. Surv. Tutor.19(2), 721–742 (2017). https://doi.org/10.1109/comst.2016.2621116.
K. Higuchi, A. Benjebbour, Nonorthogonal multiple access (NOMA) with successive interference cancellation for future radio access. IEICE Trans. Commun.98(3), 403–414 (2015). https://doi.org/10.1587/transcom.e98.b.403.
K. Higuchi, Y. Kishiyama, Nonorthogonal multiple access using intrabeam superposition coding and successive interference cancellation for cellular MIMO downlink. IEICE Trans. Commun.98(9), 1888–1895 (2015). https://doi.org/10.1587/transcom.e98.b.1888.
Z. Zhang, Z. Ma, M. Xiao, Z. Ding, P. Fan, Fullduplex devicetodevice aided cooperative nonorthogonal multiple access. IEEE Trans. Veh. Technol., 1–1 (2016). https://doi.org/10.1109/tvt.2016.2600102.
T. N. Tran, M. Voznak, Multipoints cooperative relay in NOMA system with N1 DF relaying nodes in HD/FD mode for N user equipments with energy harvesting. Electronics. 8(2), 167 (2019). https://doi.org/10.3390/electronics8020167.
Y. Liu, Z. Ding, M. Elkashlan, H. Poor, Cooperative nonorthogonal multiple access with simultaneous wireless information and power transfer. IEEE J. Sel. Areas Commun.34(4), 938–953 (2016). https://doi.org/10.1109/jsac.2016.2549378.
J. Men, J. Ge, C. Zhang, Performance analysis of nonorthogonal multiple access for relaying networks over Nakagamim fading channels. IEEE Trans. Veh. Technol.66(2), 1200–1208 (2017). https://doi.org/10.1109/TVT.2016.2555399.
C. Zhong, Z. Zhang, Nonorthogonal multiple access with cooperative fullduplex relaying. IEEE Commun. Lett.20(12), 2478–2481 (2016). https://doi.org/10.1109/lcomm.2016.2611500.
Y. Liu, Z. Ding, M. Elkashlan, J. Yuan, Nonorthogonal multiple access in largescale underlay cognitive radio networks. IEEE Trans. Veh. Technol.65(12), 10152–10157 (2016). https://doi.org/10.1109/tvt.2016.2524694.
T. N. Tran, D. Do, M. Voznak, On outage probability and throughput performance of cognitive radio inspired NOMA relay system. Adv. Electr. Electron. Eng.16(4) (2018). https://doi.org/10.15598/aeee.v16i4.2801.
F. Lu, M. Xu, L. Cheng, J. Wang, J. Zhang, G. Chang, Nonorthogonal multiple access with successive interference cancellation in millimeterwave radiooverfiber systems. J. Light. Technol.34(17), 4179–4186 (2016). https://doi.org/10.1109/jlt.2016.2593665.
H. Marshoud, V. Kapinas, G. Karagiannidis, S. Muhaidat, Nonorthogonal multiple access for visible light communications. IEEE Photon. Technol. Lett.28(1), 51–54 (2016). https://doi.org/10.1109/lpt.2015.2479600.
T. N. Tran, M. Voznak, HD/FD and DF/AF with fixedgain or variablegain protocol switching mechanism over cooperative NOMA for greenwireless networks. Sensors. 19(8) (2019). https://doi.org/10.3390/s19081845.
Y. Liu, M. Elkashlan, Z. Ding, G. Karagiannidis, Fairness of user clustering in MIMO nonorthogonal multiple access systems. IEEE Commun. Lett., 1–1 (2016). https://doi.org/10.1109/lcomm.2016.2559459.
Z. Ding, F. Adachi, H. Poor, The application of MIMO to nonorthogonal multiple access. IEEE Trans. Wirel. Commun.15(1), 537–552 (2016). https://doi.org/10.1109/twc.2015.2475746.
Y. Yu, H. Chen, Y. Li, Z. Ding, L. Song, B. Vucetic, Antenna selection for MIMO nonorthogonal multiple access systems. IEEE Trans. Veh. Technol.67(4), 3158–3171 (2018). https://doi.org/10.1109/tvt.2017.2777540.
J. Men, J. Ge, Nonorthogonal multiple access for multipleantenna relaying networks. IEEE Commun. Lett.19(10), 1686–1689 (2015). https://doi.org/10.1109/lcomm.2015.2472006.
W. Han, J. Ge, J. Men, Performance analysis for NOMA energy harvesting relaying networks with transmit antenna selection and maximalratio combining over Nakagamim fading. IET Commun.10(18), 2687–2693 (2016). https://doi.org/10.1049/ietcom.2016.0630.
X. Liu, X. Wang, in 2016 IEEE 83rd Vehicular Technology Conference (VTC Spring). Efficient antenna selection and user scheduling in 5G massive MIMONOMA system (Nanjing, 2016), pp. 1–5. https://doi.org/10.1109/VTCSpring.2016.7504208.
M. Bloch, J. Barros, M. Rodrigues, S. McLaughlin, Wireless informationtheoretic security. IEEE Trans. Inf. Theory. 54(6), 2515–2534 (2008). https://doi.org/10.1109/tit.2008.921908.
Y. Zhang, H. Wang, Q. Yang, Z. Ding, Secrecy sum rate maximization in nonorthogonal multiple access. IEEE Commun. Lett.20(5), 930–933 (2016). https://doi.org/10.1109/lcomm.2016.2539162.
Z. Qin, Y. Liu, Z. Ding, Y. Gao, M. Elkashlan, in 2016 IEEE International Conference on Communications (ICC). Physical layer security for 5G nonorthogonal multiple access in largescale networks (Kuala Lumpur, 2016), pp. 1–6. https://doi.org/10.1109/ICC.2016.7510755.
Y. Liu, Z. Qin, M. Elkashlan, Y. Gao, F. Lajos Hanzo, Enhancing the physical layer security of nonorthogonal multiple access in largescale networks. IEEE Trans. Wirel. Commun.16(3), 1656–1672 (2017). https://doi.org/10.1109/TWC.2017.2650987.
J. Zhu, Y. Zou, G. Wang, Y. D. Yao, G. K. Karagiannidis, On secrecy performance of antennaselectionaided MIMO systems against eavesdropping. IEEE Trans. Veh. Technol.65(1), 214–225 (2016). https://doi.org/10.1109/TVT.2015.2397195.
H. Lei, C. Gao, I. S. Ansari, Y. Guo, Y. Zou, G. Pan, K. Qaraqe, Secrecy outage performance of transmit antenna selection for MIMO underlay cognitive radio systems over Nakagamim channels. IEEE Trans. Veh. Technol.66(3), 2237–2250 (2017). https://doi.org/10.1109/TVT.2016.2574315.
H. Lei, et al., On secure NOMA systems with transmit antenna selection schemes. IEEE Access. 5:, 17450–17464 (2017). https://doi.org/10.1109/access.2017.2737330.
B. Li, X. Qi, K. Huang, Z. Fei, F. Zhou, R. Hu, Securityreliability tradeoff analysis for cooperative NOMA in cognitive radio networks. IEEE Trans. Commun.67(1), 83–96 (2019). https://doi.org/10.1109/tcomm.2018.2873690.
F. Zhou, Z. Chu, H. Sun, R. Hu, L. Hanzo, Artificial noise aided secure cognitive beamforming for cooperative MISONOMA using SWIPT. IEEE J. Sel. Areas Commun.36(4), 918–931 (2018). https://doi.org/10.1109/jsac.2018.2824622.
F. Zhou, Z. Chu, Y. Wu, N. AlDhahir, P. Xiao, in 2018 IEEE International Conference on Communications Workshops (ICC Workshops). Enhancing PHY Security of MISO NOMA SWIPT Systems with a Practical NonLinear EH Model, (2018). https://doi.org/10.1109/iccw.2018.
F. Jameel, S. Wyne, G. Kaddoum, T. Duong, A comprehensive survey on cooperative relaying and jamming strategies for physical layer security. IEEE Commun. Surv. Tutor.21(3), 2734–2771 (2019). https://doi.org/10.1109/comst.2018.2865607.
L Kong, S Vuppala, G Kaddoum, Secrecy Analysis of Random MIMO Wireless Networks Over α μ Fading Channels. IEEE Transactions on Vehicular Technology. 67(12), 11654–11666 (2018). https://doi.org/10.1109/tvt.2018.2872884.
D. Tran, H. Tran, D. Ha, G. Kaddoum, Secure transmit antenna selection protocol for MIMO NOMA networks over Nakagamim channels. IEEE Syst. J., 1–12 (2019). https://doi.org/10.1109/jsyst.2019.2900090.
Y. Feng, S. Yan, Z. Yang, Secure transmission to the strong user in nonorthogonal multiple access. IEEE Commun. Lett.22(12), 2623–2626 (2018). https://doi.org/10.1109/lcomm.2018.2877320.
Y. Feng, S. Yan, C. Liu, Z. Yang, N. Yang, Twostage relay selection for enhancing physical layer security in nonorthogonal multiple access. IEEE Trans. Inf. Forensic. Secur.14(6), 1670–1683 (2019). https://doi.org/10.1109/tifs.2018.2883273.
Z. Yang, Z. Ding, P. Fan, N. AlDhahir, The impact of power allocation on cooperative nonorthogonal multiple access networks with SWIPT. IEEE Trans. Wirel. Commun.16(7), 4332–4343 (2017). https://doi.org/10.1109/twc.2017.2697380.
I. Abu Mahady, E. Bedeer, S. Ikki, H. Yanikomeroglu, Sumrate maximization of NOMA systems under imperfect successive interference cancellation. IEEE Commun. Lett.23(3), 474–477 (2019). https://doi.org/10.1109/lcomm.2019.2893195.
J. Xu, L. Liu, R. Zhang, Multiuser MISO beamforming for simultaneous wireless information and power transfer. IEEE Trans. Signal Process.62(18), 4798–4810 (2014). https://doi.org/10.1109/tsp.2014.2340817.
X. Yue, Y. Liu, S. Kang, A. Nallanathan, Performance analysis of NOMA with fixed gain relaying over Nakagami m fading channels. IEEE Access. 5:, 5445–5454 (2017). https://doi.org/10.1109/access.2017.2677504. Accessed 26 Aug 2019.
S. Arzykulov, T. Tsiftsis, G. Nauryzbayev, M. Abdallah, Outage performance of cooperative underlay CRNOMA with imperfect CSI. IEEE Commun. Lett.23(1), 176–179 (2019). https://doi.org/10.1109/LCOMM.2018.2878730.
Acknowledgments
We would especially like to thank the editors and anonymous reviewers for their helpful comments to improve this paper.
Funding
The research leading to these results received funding from the Czech Ministry of Education, Youth and Sports under grant No. SP2019/41 conducted at VSB  Technical University of Ostrava.
Author information
Authors and Affiliations
Contributions
Authors’ contributions
TNT is the first author who proposed the main concept, analysed and simulated the system and presented the final manuscript. MV is the second author. He is experienced in wireless communication research. He conducted a review and provide the first author with useful commentary. Both authors have read and approved the final manuscript.
Authors’ information
ThanhNam TRAN (ORCID:0000000270657951) was born on Oct. 15, 1988 in the Vinh Long province, Vietnam. He received his M.Sc. from the Military Technical Academy (MTA) in 2014. He works and lectures at the Faculty of Electronics and Telecommunications at Sai Gon University, Vietnam. He is a member of the Wireless Communication Research Group and Faculty of Electrical and Electronics Engineering at Ton Duc Thang University. He is currently pursuing his Ph.D. in communications technology at VSB – Technical University of Ostrava, Czech Republic. He received Prof. Miroslav Voznak as a supervisor. His major interests are NOMA, energy harvesting (EH), cognitive radio (CR) and physical layer security (PLS). He has significant skill in C++, python, and Matlab programming.
Miroslav VOZNAK (ORCID:0000000151357980) received his Ph.D. in telecommunications from the Faculty of Electrical Engineering and Computer Science at VSB – Technical University of Ostrava and completed his habilitations in 2002 and 2009. He was appointed Full Professor in 2017 in Electronics and Communications Technologies. He is an IEEE senior member and has served as a member on editorial boards for several publications, such as the Journal of Communications and the Advances in Electrical and Electronic Engineering Journal. His research interests focus generally on information and communications technology, particularly on quality of service and experience, network security, wireless networks and, in the last few years, big data analytics.
Corresponding author
Ethics declarations
Competing interests
The authors declare no conflict of interest in producing this article.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Tran, TN., Voznak, M. On secure system performance over SISO, MISO and MIMONOMA wireless networks equipped a multiple antenna based on TAS protocol. J Wireless Com Network 2020, 11 (2020). https://doi.org/10.1186/s136380191586y
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s136380191586y
Keywords
 Singleinputsingleoutput (SISO)
 Multiinputsingleoutput (MISO)
 Multiinputmultioutput (MIMO)
 Nonorthogonal multiple access (NOMA)
 Transmitter antenna selection (TAS)
 Secrecy outage probability (SOP)
 Imperfect SIC