Full Duplex Component-Forward Cooperative Communication for a Secure Wireless Communication System

: The technological breakthrough in the form of Internet of Things (IoT), Big data and connected world is increasing the demand of better spectrum utilization. Half-Duplex (HD) transmission is mostly used in the earlier communication systems. The high transmission demand requires the better utilization of the existing spectrum. There are several possible ways to overcome the problem of better spectrum usage. In-Band Full Duplex (IBFD) is one of the techniques that can double the Spectral Efﬁciency (SE) in a Beyond 5G (B5G) communication system. In this paper, our aim is to use the spectral efﬁcient IBFD scheme to improve the security of the system with minimum interference. The interference can be reduced by the addition of orthogonality between the transmitted and received signal of a relay. A component-forward scheme is proposed in this paper to create such orthogonality. For achieving the desired aim, IBFD is used with Device-to-Device (D2D), Artiﬁcial Noise (AN), Modulation based orthogonalization, Radio Frequency Energy Harvesting (RFEH) and proposed Full-Duplex Component Forward (FD-CF) algorithm for multiple relays. We also use non-linear harvested power as one of the sources to reuse the exiting power for evaluating the system performance. The derivation of Secrecy Outage Probability (SOP) and throughput is derived in this paper for the FD-CF cooperative communication and is explored with and without non-linear RFEH. The simulation results show the comparison between the component-forward and decode-and-forward communication with one or more relays.


Introduction
The technological development necessitates the smart way of presentation available natural resources. Recently, several countries have executed 5G with high frequency ranges of the spectrum. Sophisticated use of the remaining electromagnetic spectrum perhaps balances the spectral efficiency [1]. Numerous approaches can be made by directing the high-quality throughput with the prevailing proposed technologies such as In-Band Full Duplex (IBFD). IBFD focus on the efficient utilization of the spectrum for developing a system with better Spectral Efficiency (SE) by the challenging way of spectrum reuse [2]. The existing performance of 5G mostly based on the Orthogonal Frequency Division Multiple Access (OFDMA) with Half Duplex (HD) due to the high co-channel interference. The 4G scheme is likewise based on OFDMA with HD. The replacement of the existing 4G setup with 5G is comparatively easier and inexpensive. Keeping the consideration of spectrum limitation, it would be better to extend the system further by using INFD for the 5G beyond technologies.
significantly. FD can improve the traditional wireless communication to a better extent in terms of loss of data due to high congestion, hidden terminals, time delay and SE. The key idea behind the usage of FD communication is to make heterogeneous dense network with high capacity and flexible relaying modes. The quantitative analysis of theoretical and practical placement shows that, at the cost of enlarged complexity, FD provides diversity, high throughput, low symbol error rate and decreased use of HD. The D2D FD communication becomes better with the increase of in-cell communication ratio that provides to bandwidth efficiency.
The IBFD distribution desires new algorithms for the arrangement of 5G beyond technology. It goals to usage the high frequency bands such as millimeter waves (mmWaves). This is for the low frequency spectrum in fully used with the present system of below 5G technologies. Additionally, several wireless devices are being used in the communication system which cannot be charged using the traditional wired power system. Such devices also contain a limited amount of power that can exhaust rapidly. This situation brings the attention towards renewable energy resources, Wireless Information Transfer (WIT) and Simultaneous Wireless Information and Power Transfer (SWIPT) [14].
To prolong the life of such kind of wireless devices in Full Duplex Relay (FDR) systems, several research articles consider RFEH [9]. There exist several literature works proposing various SWIPT architecture such as ideal receivers, Power Splitting (PS) receivers and Time Switching (TS) receivers [15]. Only PS and TS are applicable practically. In PS, the receiver splits the power for RFEH and transmission of signal, and, in TS, the receive uses two time slots to harvest and retrieve information. Several new approaches and algorithms have been attempted for more accurate RFEH. A realistic measurement and waveform-based energy harvesting model [16] uses the pre-equalization to prevent wireless channel distortion. The result shows that using the phase modulation the power conversion efficiency is increased by a factor of two. FD with bidirectional wireless sensor network (BWSN) and Multiple Input Multiple Output (MIMO) based SWIPT system [17] is a considered approach for harvesting energy and transmitting power constraints with a joint optimization problem for source and relay beamforming. Optimization for MIMO-NOMA [18] with dynamic power allocation and user grouping provides an effective way for SWIPT. For the secure transmission in mind self jamming, SWIPT [19] may provide security along with energy harvesting.

Motivation and Contribution
The requirement for the innovative spectrum consumption in a diligent way with least interference and high security in 5G leads towards the motivation to develop a system for IBFD with least interference. After reviewing the literature based on IBFD, a space of interference management and security issues is observed for 5G. To target the complexity of interference and security issues of a FD system, we propose a new component and forward FD scheme.
For achieving the desired objectives, the contributions of this paper are: • A component-forward cooperative communication is proposed to minimize the possibility of interference in IBFD with D2D, AN and modulation-based interference reduction [1]. • Artificial Noise (AN) is added to prevent the increasing security threats for the proposed FD component-forward (FD-CF) cooperative communication system. The AN increases the interference at the Eve's node and declines the possible material decoding. It also drains Eve by providing high computational complexity and energy. • The aforesaid approaches are used for refining the system's secrecy capacity, outage probability and throughput. • Alongside the basic system, RFEH is also used to further elaborate the proposed scheme with secrecy capacity, outage probability and throughput in the presence of a TS based RFEH circuit.
The idea is elaborated with the difference between the FD-DF and FD-CF in Table 1. In this paper, we use one of the baseline FD-DF techniques with a single relay selection from K relays [20] for the comparison with FD-CF in the simulation result. The proposed idea can be implemented with any of the existing FD-DF techniques to obtain the same differences.

Modulation on the Alice
Actual message signal is modulated on both real and imaginary component of the modulation technique.
Actual message signal is modulated either on the real or imaginary component of the modulation technique to prevent interference in the FD mode.

Modulation on the Relay
Actual message signal is modulated on both real and imaginary component of the modulation before the relay forwards the signal.
The received message signal is modulated on the opposite component of the received signal to prevent interference with the same component in the FD mode.

Co-channel Interference with AN
Yes.
Only with the similar (real or imaginary) component of AN.
Very less as compared to FD-DF.

Co-channel Interference without AN
Yes. Ideally No.

Signal to interference plus noise ratio
Low due to interference. High due to less or no interference.

Channel Capacity
Low due to interference. High due to less or no interference.

Secrecy Capacity
Low.
High due to high interference on Eve's node.

Structure
The structure for the rest of the paper is as follows. Table 2 lists the abbreviations used in the paper. Section 2 provides the proposed system model details with the description of proposed FD-CF and addition of AN. Section 3 contains the computation of secrecy capacity, secrecy outage probability, harvested power and system throughput. Section 4 shows the numerical results for the comparison of FD-CF and FD-DF schemes followed by conclusion in Section 5 and Appendix A.  Figure 1 represents the system model for a D2D component-forward supported IBFD (FD-CF) for a Downlink (DL) wireless cooperative communication system. The system can be enabled in any other scenario as well. However, this paper focuses on the basic system model of FD-CF. The offered system model includes a source (Alice), receiver (Bob), IBFD enabled relays R k for k = {1, · · · , K} and the passive eavesdropper (Eve). Bob and Eve are considered to have similar channel conditions. Alice, Bob and Eve contain a single antenna excluding relays. P T is the total power for each transmission, which is divided equally for Alice-relay and relay-Bob transmission. Alice-relay and relay-Bob distances are considered to be equal, d ar = d rb . The distance can be varied according to the system model requirements. As per assumption, relay and Bob are facing a Rayleigh fading channel and their Channel State Information (CSI) is considered to be known by Alice, relay and Bob. Eve is considered not to have the CSI information of the legitimate user. Eve's channel is considered to be the same as Bob, √ g e ≈ √ g rb . Therefore, Eve is capable of detecting the signal of Bob. √ g ar and √ g rr are the other channels for the component-forward cooperative transmission in the IBFD mode from Alice to relay and the SI channel from relay to itself.

System Model
Alice uses the algorithm used in [1,21], which is basically a technique for reduction of interference in Non-Orthogonal Multiple Access (NOMA), for the superposition of the signals, as it prevents the signal from interference as compared to conventional techniques due to the orthogonal nature of modulated signals. The interference in [1,21] is half as compared to conventional NOMA because half of the receivers are modulated on the real component and the remaining are modulated on the quadrature components of the modulation. Ideally, there will be no interference between the receivers' signal modulated on the perpendicular component. This paper does not include the NOMA system. However, the basic idea of [1] is used to balance the offered interference of the proposed FD-CF algorithm. FD-CF can also be extended for NOMA in future work.
FD-CF is an algorithm in which the source (Alice) encodes the transmission message on the real component of the modulation only and add a complex AN with the signal which is a null space of the receiver's channel (relay(s)). The relay selection takes place according to [20]. The selected relay from the K relays receives the real component of the signal only, as AN has been nullified already from its channel. After receiving the signal, relay modulates the signal on the quadrature component of the modulation, adds AN which is null space of Bob and forwards it without decoding. Relay receives the real component signal and forwards the quadrature component of the signal. Therefore, ideally, there is no co-channel interference, which is the main problem in IBFD transmission. There is a self-interference at the FD mode between the real component of the received signal by the relay and the complex A 2 added in the signal forward by the relay. For RFEH circuit in the given system model, TS is used for harvesting energy. Relay decodes the signal from Alice in time αT and harvests energy. Now, relay carries the signal and harvested energy which was received already in the previous transmission. In the second time slot (1 − α)T, relay receives the signal from Alice and forwards the previously decoded signal by using the harvested power P r to Bob within the same time and frequency band.

Addition of Artificial Noise for Better Security
Artificial Noise (AN) is a sufficient way for the protection of the transmission signals from an Eve and other users. AN system design depends on the receiver's channel but not the Eve's channel. AN is generated, before transmission of signal by the Alice and the relay to degrade the Eve's channel. The AN A n = a n + a nĵ is complex Gaussian in nature. In the case of the fixed AN, the value of ||g e A n || might be smaller. To avoid this situation, the value of AN is considered as the Gaussian random variable in the null space of √ g n of the relay's and Bob's channels, respectively, such that √ g n A n = 0 [22]. The superposed signal on the Alice's node for the broadcast can be given as In the above equation, the signal does not have the complex nature. However, it is combined with the AN A 1 to make the total complex signal. For more than one users, some users' signals can be modulated on imaginary and some on the quadrature component of the modulation to make a complete complex signal for transmission. In this paper, the use of AN makes the total complex transmission signal. Additionally, A 1 is added by Alice as a null space of the relay and relay added A 2 as the null space of Bob. The superposed signal for broadcast on the relay's node is given as (2)

System Analysis
The relay and Bob decodes the signal normally as in a normal communication system. However, Eve cannot decode the signal received from Alice or the relay due to the addition of AN, which is the null space of the channel g ar and g rb of the relay and Bob only. The signal received by Bob also includes the self-interference for its co-channel transmission as [20]: (1); where w n ∼ CN(0, σ 2 ) is the Additive White Gaussian noise (AWGN) and s rb [t] is the signal transmitted by relay (received along with the previous transmission) to the Bob in the FD mode, which causes self-interference. The FD cooperative communication is used for Bob due to no direct link between Alice and Bob. The total DL received signal by Bob from the relay is given in [20]: (5), In the above equation, the signal received from the relay contains only the imaginary part, as √ g rb A 2 = 0.
Eve's received signal can be determined by adding AN [22] A 2 in Equation (4). Eve receives signal with interference due to lack of information about AN.

Performance Evaluation
For the evaluation of a system, capacity, secrecy capacity, secrecy outage probability and throughput are some of the important parameters to prove authenticity and usefulness of any system. These parameters were computed along with the elaboration of RFEH effect on the system.

Computation of Secrecy Capacity
Each node receives a certain level of SINR or SNR depending on its channel condition and the interference. In this paper, SINR and SNR are used interchangeably, denoted as ζ. The respective received ζ of Bob and Eve are given as: and ζ e = ζ re , where ζ ar = P 1 G ar σ 2 , ζ rr = a 2 G rr σ 2 , ζ rb = P 2 G rb σ 2 and ζ re = P 2 G re A 2 G re +σ 2 which follows the exponential distribution with parameter λ ar = P 1 G ar σ 2 , λ rr = a 2 G rr σ 2 , λ rb = P 2 G rb σ 2 , λ re = P 2 G re A 2 +σ 2 and G ni = K ni d 0 γ d ni −γ g ni . For λ re ; it is assumed that σ 2 = σ 2 /G re .
The achievable data rate for Bob and Eve is given as: and R re = log 2 1 + ζre The possible secrecy capacity of the system for FD-CF system is given as C sec = max 0, log 2 1 + min ζ ar ζ rr +1 , ζ rb 1 + ζre (11)

Relay Selection
For the better secrecy performance in the presence of Eve, an opportunistic relay selection scheme is used [20]. The scheme is based on the selection of the relay among K relays that maximizes the secrecy capacity of the system.
where R s is the selected relay. Whil selecting the relay, the relay selection scheme is considering the channel between relays and Eve. A centralized approach is used in this paper, where the source or destination keeps record of the K relays and their CSI. Using the criteria of Equation (12), the best relay for the transmission is decided.

Computation of Secrecy Outage Probability
For the derivation of proposed system's Secrecy Outage Probability (SOP), the min-max approached is used. The SOP for FD-CF cooperative communication for the relay selection scheme is given as: where Pr[.], f X (.) and F X (.) are the notation for probability, Probability Density Function (PDF) and Cumulative Distributive Function (CDF). a = 2 C th − 1, b = 2 C th , y = γ re ≥ 0 and f ζ re = e − y λre /λ re . The CDF F Z (z) of the random variable Z is derived in the Appendix A and is given as Substituting the required parameter and considering ζ sr = ζ rd , the SOP is given as ζsr bζ re ζ rr be 2a ζsr ζ re ζ rr + e L Γζ sr + e L ζ sr log 1 The above expression is the conditional expression with Re[p] > 0 and Re[1/ζ re + 2b/ζ sr ] ≥ 0 where L = (2bζ re +ζ sr )(aζ rr +ζ sr ) bζ re ζ rr ζ sr , Γ = EulerGamma and 1F1[1, 0, 0] is the Kummer confluent Hypergeometric function.

Radio Frequency Energy Harvesting (RFEH)
The RFEH at relay for time αT is given as (16) where 0 < η < 1 shows the energy conversion efficiency and its magnitude varies with the quality of electronic circuit. It is considered that, if the relay needs to use the RFEH for assisting Bob, then the harvested power by the relay is given as Practically, the output harvested power of a RFEH circuit may depends on the input harvested power up to a saturated threshold power P th . If the input power exceeds the threshold level, then the output power remains unchanged. The non-linear characteristics of RFEH are caused by several factors including saturation non-linearity and diode. Diode has non-linearity as intrinsic properties. Experimentally, non-linear characteristics of RFEH circuits are confirmed. Consequently, the power harvested by the relay using a non-linear circuit [16] is given as In the FD mode of time duration (1 − α)T, the relay receives and forward the signal to Bob. When the signal is forwarded by the relay to Bob using the harvested power P r , the secrecy outage probability can be obtained from Equation (15), after replacing P 2 with P r .

Secrecy Throughput Evaluation
Throughput is another significant system parameter that can clarify the authenticity of a system. The throughput in a FD-CF system, when the relay uses its internal power for the transmission of Bob's signal, is given as The throughput for RFEH circuit [9] of the proposed FD-CF system is given as where α is the time switching ratio.

Numerical Results
In this section, we discuss the simulated results comparison between the proposed FD-CF and FD-DF cooperative communication system. For the simulation of the systems' comparison, Rayleigh flat fading channel and 16-QAM was considered. Other numerical values that were used for the simulation are given as: d ar = 0.5 m, d rb = 0.5 m, γ = 2, d 4 = 1 m, P T = 1 W, P 1 = 0.5 W and P 2 = 0.5 W. MATLAB was used as a simulation tool for the comparison between the proposed and the baseline scheme.
For the fair comparison between FD-CF and FD-DF, all selected parameters were same including AN. The only difference was the interference level during the IBFD mode in both techniques. Figure 2 shows the simulation result of the achievable data rate of the FD-CF and FD-DF K relay(s) system. For this particular simulation, the distance was changed from 0.5 to 2, 4 and 6 m. However, the same power was used for a fair comparison. It shows that the achievable data rate decreases with the increase of distance. However, the performance can be maintained by increasing the amount of power respectively. The proposed FD-CF system outperforms FD-DF system. This is due to the comparatively less interference in the proposed FD-CF system. The opposite components reduces the chance of interference. Figures 3 and 4 show the secrecy capacity comparison with respect to the respective SNR. In Figure 3, the transmitted powers of Alice P 1 and the relay P 2 are same. In Figure 4, the relay uses harvested power P r for forwarding the signal of Bob.It is clear from both figures that the secrecy capacity of both FD-CF and FD-DF is increasing with time. However, the secrecy capacity for FD-CF is higher as compared to FD-DF. This is due to the less interference at the relay by the FD-CF scheme and the high introduced interference at the Eve's node.Therefore, the proposed FD-CF outperforms FD-FD. Figures 5 and 6 show the comparison for the secrecy outage probability offered by FD-DF and FD-CF. In Figure 5, the transmitted powers of Alice P 1 and the relay P 2 are the same. In Figure 6, the relay uses harvested power P r for forwarding the signal of Bob. To show the different response with respect to the number of relays K, we simulated the results for K = 1, 2, 4. SOP result for both algorithms decreases with the increase of SNR. However, FD-CF outperforms FD-DF, due to the same reason discussed of minimum interference as discussed for the simulation results above. For the proposed technique the ζ re is approximately equal to 0 due to high interference at Eve. According to the derived equation of outage probability ζ re must be greater than zero. Therefore, for simulation purposes, we considered ζ re = 0.1 for FD-CF. Figures 7 and 8 show the increasing throughput with SNR for FD-DF and FD-CF schemes. To show the different response with respect to the number of relays K, we simulated the results for K = 1, 2, 4. Throughput of FD-CF is higher due to its low SOP as compared to FD-DF algorithm. It can be seen in figures that with K = 4 the throughput is maximum. The level of throughput decreases with the reduction in the number of relay(s) selection. In addition, the proposed FD-CF with K = 2 outperforms FD-DF with K = 4 Minimum interference in FD-CF makes it reliable comparatively. Figure 9 shows the throughput change with respect to α. FD-CF throughput is better than FD-DF. Clearly, the throughput for both techniques first increases and then decreases with the increase of time switching ratio. When the time switching ratio is maximum, the throughput for both schemes is zero. It means that the entire available time is used for harvesting power and not for any transmission. It can be concluded with this result that the throughput can be maximize by allocating suitable α for harvesting power.

Conclusions
Security and spectral efficiency requirement is obligatory in the upcoming 5G systems. In this paper, aiming at such issues, a FD-CF algorithm is proposed to reduce interference and improve the security of a spectral efficient IBFD system. The proposed system improved the SOP, secrecy capacity and throughput. For the proposed system, IBFD algorithm is used to save time and bandwidth with the D2D communication. FD-CF reduced the interference of the IBFD mode of the system, hence improved the capacity, secrecy capacity, SOP and throughput. The comparison of FD-CF and FD-DF is given in Section 4 which shows that the proposed FD-CF outperforms FD-DF scheme with the choice of K relay(s). In addition, the non-linear harvested power is used to observe the difference between the performance of the proposed and baseline system's SOP and throughput.
The security management is the challenging concern in the modern system due the connection of secure devices with the cloud. The main advantage of the proposed FD-CF system is the improvement of the system security. It is a basic system that can be extended with other algorithms which are used in the existing literature with FD-DF. The proposed system can also be combined with NOMA to further enhance the system security and spectral efficiency.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
Derivation for the CDF of F Z (z) For the CDF of the random variable Z, recall that Z = min ζ ar ζ rr +1 , ζ rb . For simplicity of the derivation, consider L = ζ ar ζ rr +1 and M = ζ rb . As we know that The above equation shows the requirement of CDF of F L and F M for obtaining the CDF of Z. The CDF of F L is calculated as F ζ ar (z(r + 1)) f ζ rr (r)dr