Secure outage probability of cognitive radio network relying non-orthogonal multiple access scheme

This paper studies the secondary network relying relay selection to transmit signal from the secondary source (base station) to two destinations. Especially, two destinations are required non-orthogonal multiple access (NOMA) scheme and it benefits to implementation of the Internet of Things (IoT) systems. However, eavesdropper overhears signal related link from selected relay to destination. This paper measure secure performance via metric, namely secure outage probability (SOP). In particular, signal to noise ratio (SNR) criterion is used to evalute SOP to provide reliable transmission to the terminal node. Main results indicates that the considered scheme provides performance gap among two signals at destination. The exactness of derived expressions is confirmed via numerical simulation.


INTRODUCTION
Recently, spectrum occupation results in lack of resource, and cognitive radio (CR) is introduced as technique to provide network with higher the spectrum efficiency. In CR, users in the secondary network are able to dynamically access the licensed spectrum provided by primary network [1][2][3]. In the current researches about the cognitive radio networks, two popular kinds of CR network are employed, such as underlay and/or overlay spectrum sharing strategies. The unused licensed spectrum is reused by secondary users to access network and collision is prohibited with the primary transmission, and such is called as the overlay strategy [4][5][6]. By controlling the interfere to the primary receivers under a threshold, the secondary network applying the underlay strategy, and it then shares with the primary users in term of licensed spectrum. Advantages of CR can be determined in various works such as [7][8][9].
To further improve the spectrum efficiency, non-orthogonal multiple access (NOMA) technique is also introduced, it benefits to wireless network by allocating multiple users using different levels of transmit power to access the same frequency [10][11][12]. The receiver need to distinguish the different signals, and it requires operation of successive interference cancellation (SIC) [12]. The NOMA based CR networks are studied recently. Taking advantages of both the cognitive radio and NOMA techniques, it will benefit from the spectrum efficiency and massive users connected [13].
It is necessary to put our concerns on information security for the cognitive radio networks in which NOMA is activated. Advantage of spectrum sharing nature in the cognitive radio benefits to CR-NOMA networks, however it is vulnerable to be eavesdropped since the eavesdropper can pretend to be a secondary user. Recently, secure performance is presented in the context of NOMA implementation [15][16][17][18][19][20][21][22][23][24]. Motivated by these papers, we consider secure outage probability of CR-NOMA.

SYSTEM MODEL
In Figure 1, the considered system contains secondary network including base station BS, and N relay nodes, two destination U 1 , U 2 , one eavesdropper E, We denote h u as channel dedicating to node u, and it follows Rayleigh fading model with channel gain λ u . It is noted that P S is transmit power at BS and it is limited under constraint with the primary network. The interference channel from BS to the primary network is h SP .  The transmit power constraint is limited at BS whereP S and I is maximum average transmit power at the BS and interference temperature constraint (ITC) at primary destination P D . In the first time, R receives the following signal where h Rn is the channel between BS and R, the AWGN noise terms at R is n Rn indicates. To decode signal x 1 and x 2 at R, the signal-to-interference-plus-noise ratio (SINR) and signal-to-noise ratio (SNR) can be respectively written as (3) where ρ S = P S σ 2 is the transmit SNR at the BS. At relay relying SIC, the received SINR at R to detect its own message x 2 is given by (4).
In the next phase, signal √ a 1 P Rx1 + √ a 2 P Rx2 is forwarded by relay R to U i , i = 1, 2. We denote P R is the transmit power at R. U i receives the following signal Ì ISSN: 2302-9285 in which the AWGN noise terms is n Ui measured at U i . h Rn is denoted as the channel between Relay and U i . Further, principle of NOMA applied to U i with higher power factor, it can detectx 1 by consideringx 2 as a background noise with the following SINR where ρ R = P R σ 2 is the transmit SNR at R. To continue detecting procedure, U 2 needs SIC to decodex 1 while considering its own datax 2 as a noise. The SINR is written as (7).
In this regard, U 2 detects its own signal and the corresponding SNR is given as (8).
Unfortunately, eavesdropper steals information from the selected relay, the received signal at E is given as (9) where n E is the AWGN noise terms at E. The channel between Relay and E is h E . Then, SNR at E is given as where ρ E = P E σ 2 is transmit SNR at E. The secrecy capacity for U 1 is computed as (11).
Similarly, the secrecy capacity for U 2 is formulated as (12).
The criteria to select relay n * is defined related the best channel given as (13).
Since it is difficult to derive a closed-form expression for (18), an accurate approximation can be achieved for it. Using Gaussian-Chebyshev quadrature, we have (19) where then, A 2 is calculated as (21).
Although, it is difficult to derive a closed-form expression for (22), we can obtain an accurate approximation for it. Using Gaussian-Chebyshev quadrature, it can be obtained that where ζ v = cos 2v−1 2V π and Ξ (x) = Substituting (23) and (19) to (16), A is given by (24).
The similar, using Gaussian-Chebyshev quadrature, B is given by (26).
After several steps, P SOP 2 can be obtained as (30).

NUMERICAL RESULTS
We set main parameters as: power allocation factor a 1 = 0.9 and target rates a 2 = 0.1. R 1 = R 2 = 1. SNR of interference from primary network ρ I = 15 dB. SNR of eavesdropper ρ E = 1 dB. The channel gains are λ SP = 0.1, λ SR = 1, λ RD1 = 0.9, λ RD2 = 0.6 and λ RE = 0.1. Q = V = 1000. Figure 2 shows SOP performance versus transmit SNR. We consider many cases related to NOMA, OMA. As can be seen from such outage performance, the error floor curves are very matched with exact curves at high ρ. Signal x 1 exhibits better SOP compared with x 2 , and it can be explained that different power allocated to each signal. The simulation results are also very tight with the analytical results. Similar trend of SOP can be reported in Figure 3. Figure 3 depicts SOP curves versus ρ I . When ρ I is greater than 30 (dB), SOP curves meet saturation.

CONCLUSION
The paper studied cognitive radio using NOMA and relay selection. Secure performance is considered as existence of an eavesdropper. Moreover, the exact SOP is derived for two destinations. The derivations and analysis results showed that the proposed relay selection can effectively enhance the secure performance.