Combining Power Allocation and Superposition Coding for an Underlay Two-way Decode-and-forward Scheme

Nội dung text: Combining Power Allocation and Superposition Coding for an Underlay Two-way Decode-and-forward Scheme

1. VNU Journal of Science: Comp. Science & Com. Eng, Vol. 37, No. 1 (2021) 1-15 Original Article Combining Power Allocation and Superposition Coding for an Underlay Two-way Decode-and-forward Scheme Pham Ngoc Son1,*, Tran Trung Duy2, Phuc Quang Truong1, Son Ngoc Truong1, Pham Viet Tuan3, Van-Ca Phan1, Khuong Ho-Van4,5 1Ho Chi Minh City University of Technology and Education, Vo Van Ngan Street, Thu Duc City, Ho Chi Minh City, Vietnam 2Posts and Telecommunications Institute of Technology, Nguyen Dinh Chieu Street, District 1, Ho Chi Minh City, Vietnam 3University of Education, Hue University, Le Loi Street, Hue City, Vietnam 4Ho Chi Minh City University of Technology (HCMUT), Ly Thuong Kiet Street, District 10, Ho Chi Minh City, Vietnam 5Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam Received 30 May 2020 Revised 15 November 2020; Accepted 15 January 2021 Abstract: In this paper, we analyze an underlay two-way decode-and-forward scheme in which secondary relays use successive interference cancellation (SIC) technology to decode data of two secondary sources sequentially, and then generate a coded signal by superposition coding (SC) technology, denoted as SIC-SC protocol. The SIC-SC protocol is designed to operate in two time slots under effects from an interference constraint of a primary receiver and residual interference of imperfect SIC processes. Transmit powers provided to carry the data are allocated dynamically according to channel powers of interference and transmission, and a secondary relay is selected from considering strongest channel gain subject to increase in decoding capacity of the first data and decrease in collection time of channel state information. Closed-form outage probability expressions are derived from mathematical manipulations and verified by performing Monte Carlo simulations. An identical scheme of underlay two-way decodeand-forward relaying with random relay selection and fixed power allocations is considered to compare with the proposed SIC-SC protocol, denoted as RRS protocol. Simulation and analysis results show that the non-identical outage performances of the secondary sources in the proposed SIC-SC protocol are improved by increasing the number of the secondary relays and the interference constraint as well as decreasing the residual interference powers. Secondly, the performance of the nearer secondary source is worse than that of the farther secondary source. In addition, the proposed SIC-SC protocol outperforms the RRS comparison protocol, and effect of power allocations through channel powers is discovered. Finally, derived theory values are precise to simulation results. Keywords: Successive interference cancellation, superposition coding, power allocation, underlay cognitive radio, non-orthogonal multiple access, outage probability.* ___ * Corresponding author. E-mail address: sonpndtvt@hcmute.edu.vn 1
2. 2 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 1. Introduction constraint of a primary receiver was improved by combining digital network coding and Two-way communication is a protocol for opportunity relay selection in which a selected exchanging data back and forth between users secondary relay created a new data by XOR in which the users are both receiver and operation after decoding received data transmitter points [1, 2]. Energy and spectrum successfully [7]. Effects of multiple primary utilization efficiencies have been enhanced by receivers have been investigated in [8, 9]. The two-way cooperation [1, 2]. Intermediate relays authors in [10] took imperfect channel state operating as amplify-and-forward (AF) and information (CSI) from the SUs to the Pus into decode-and-forward (DF) devices process probability analyses. S. Solanki et. al in [11] received signals before forwarding back to the exploited direct links to build adaptive users. With lower transmit powers, the noise- protocols under average interference added data amplified is always sent to the constraints. The transmit powers of the desired users in the AF operation whereas the secondary sources and the secondary relays cooperative relays in the DF operation drop were set independently to maximum of those data due to unsuccessful decoding. achievable thresholds [7, 10, 11] or minimum However, with higher transmit powers, the of internal powers and mutual interference noises are cleared by the DF relays and then the constraints [8, 9]. In addition, most of these desired users get a high success rate of investigations have been proposed on three- decoding. Transmit frequency spectrum of the phase solutions, and therefore the bandwidth users and relays can be licensed or are shared utilization is divided by three times. by primary users (PUs) [3-5]. Spectrum sharing solution known as cognitive radio is to enhance The authors in [12] employed superposition bandwidth demand for mobile multimedia coding (SC) at each source group and services and the explosive development of next- successive interference cancellation (SIC) at a generation wireless networks such as wireless relay to send lots of broadcast data whereas sensor networks, Internet-of-things (IoT), next- only using two time slots (two phases) in generation mobile networks where secondary two-way relaying networks. The SC and SIC users (SUs) collaborate with the PUs [3-5]. In are core technologies in nonorthogonal multiple the cognitive radio networks, the SUs can access (NOMA) systems [13]. The SC transmit at any time so that the interference technology is used at the transmit source to powers at the PUs are limited under tolerable merge signals with different powers based on thresholds, denoted as underlay operation distances to the destinations. The farther protocol [4, 5]. The tolerable thresholds are destinations are allocated by the higher signal gifts of the primary networks sent to the powers. The nearby destinations apply the SIC secondary networks based on the quality-of- technology to decode the derided signals by service (QoS) contracts. As a result, the canceling the higher-powered signals and transmit power of the SUs must be adjusted treating the lower-powered signals as continuously as a function of tolerable interference [14]. However, decoding interference thresholds and channel gains. operations by the SIC technology can be imperfect because of residual interference Investigations on performance of underlay signals [12]. two-way relaying systems have been considered in [7-11]. System performance of secondary Combination of the SC technology and the two-way networks under an interference power allocation in the NOMA networks has been investigated in [12, 14-19]. Transmit
3. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 3 powers under control of interference constraint selection and the fixed power allocations are [15, 18, 19] and maximum power limitation considered to compare the proposed SIC-SC [12, 14-17] were allocated to fixed values to protocol, denoted as RRS protocol. The share a part of the total power to different users simulation and analysis results show in multiple access operations. In [18], transmit contributions as follows. Firstly, with the SIC power of a base station in multicast networks and SC technologies combining the power was achieved to maximum constraint following allocations, the non-identical outage a min-max rule of overall maximum performances of the secondary sources are permissible interference powers and maximum improved when the number of the secondary transmission power whereas power allocation relays and the interference constraint are coefficient were constant values based on increased as well as the residual interference channel gains from the base station to multi powers are controlled to decrease. Secondly, the users. A similar set-up model as [18] has also performance of the nearer secondary source is been considered in a recent study [19] to against worse than that of the farther secondary source. eavesdroppers. Power allocation coefficients in Thirdly, the proposed SIC-SC protocol the SC technology without constant values outperforms the RRS comparison protocol in should be considered in multiple-access terms of the outage probabilities, and networks. discussions on effect of power allocations In this paper, we analyze an underlay two- through channel powers are presented. Finally, way DF scheme with two secondary sources, derived theory values are precise to simulation multiple secondary relays and a primary results. receiver. In this scheme, the secondary relays This paper is organized as follows. Section use the SIC technology to decode the data of 2 presents a system model of an underlay two- two secondary sources sequentially, and then way DF scheme. Section 3 analyzes outage generate a coded signal by the SC technology, probabilities of the proposed SIC-SC protocol denoted as SIC-SC protocol. By applying the and the RRS comparison protocol. Analysis and uplink NOMA protocol, the SIC technology simulation results are presented in Section 4. and the SC technology, the SICSC protocol Finally, Section 5 summarizes contributions. operates in two time slots, and also suffers an interference constraint of the primary receiver and residual interference of the imperfect SIC 2. System Model processes. Transmit powers provided to carry Figure 1 presents a system model of an the data are allocated dynamically according to underlay two-way DF scheme in which two channel powers of interference and secondary sources SS1 and SS2 send transmission. A secondary relay is selected corresponding data s1 and s2 to each other with from considering strongest channel gain subject the help of a closed group of N intermediate to increase in decoding capacity of the first data secondary relays SRi, where iN 1,2, ,  . The and decrease in collection time of CSIs. System secondary network nodes SS1, SS2 and SRi have performance of the SIC-SC protocol is identical variances of the zero-mean white evaluated by closed-form outage probability Gaussian noises (AWGN) (denoted as N0), and expressions. These outage probability analyses are in an interference constraint of a primary are verified by performing Monte Carlo receiver PR (denoted as I). The secondary simulations. An identical scheme of underlay relays SRi are nearer to the secondary source two-way DF relaying with random relay SS1 than the SS2, and thus the secondary relays
4. 4 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 SRi use the SIC technology to decode the data SS2 and SRi can know perfectly the CSIs h , SS1 PR s1 firstly. In Figure 1, a direct transmission hSS PR and hSR PR respectively by directly between secondary sources SS1 and SS2 is 2 i skipped by far distance or deep shadow fading feedback channels from the primary receiver [2, 20], and secondary and primary nodes are PR [22, 26] or by indirectly feedback channels installed with a single antenna. from a third party [22, 27], iN 1 ,2 ,. . . ,  . Firstly, the secondary sources SS1 and SS2 PR h SS PR 2 1 broadcast request-to-send messages (RTS) h SS PR h sequentially to the secondary relays SR with SRi PR i small transmit powers and low rates. The RTS messages contain its information such as the links to the PR. From receiving these RTS messages, SR1 the SRi can estimate the fading channels hSS SR h h 1 i SS1 SS1 SRi SS SR 2 i SS2 s1 s2 and h , and then broadcast a helper-ready-to- SRi SS SR2 i s hSR SS h i 1 SRi SS2 send (HTS) message, which contains the SR N and , to the SS1 and SS2. Next, the SS1 and SS2 can estimate the fading channels h and SRi SS 1 h from the received HTS messages, Figure 1. System model. SRi SS 2 respectively. Therefore, the SS1 can know In Figure 1, hXY denotes wireless channels perfectly all fading channels to the secondary of links X − Y which are modeled as complex relays and the information link of the SS2 to normal distributions hCNXYXY 0, with zero allocate the transmit powers and select the cooperative secondary relay. In addition, the means and normalized variances  (  XY XY selected relay can use the detected and estimated are also the normalized powers of the channels) information to cancel interference. Finally, the [21], where X,YSS ,SS,SR12i ,PR  . For the SS1 will send a clear-to-send message (CTS) reason that the secondary relays SRi are located including initial parameters to begin a next data in the closed cluster and are closer to SS1 than transmission phase. SS2, thus the normalized variances are set as In the underlay cognitive radio schemes, the    ,    , interference at the primary receiver PR from the SS11 SRSRii SS 1 SS22 SRii SR SS 2 secondary network are less than or equal the   ,   ,   and SRi PR 3 SSPR1 4 SSPR2 5 constraint I [28, 29]. Inequalities related 2 to transmit powers and channel gains are   [22, 23]. As a result, gh are 12 XY XY obtained as exponentially distributed random variables P g P g I SS1 SS 1 PR SS 2 SS 2 PR (RVs) with the probability density function , (1) x P g I, i 1,2, , N (PDF) as fxe () XY and the SRii SR PR gXYXY where P , P and P are transmit powers cumulative distribution function (CDF) as SS1 SS2 SRi F( x ) 1 e x XY [24] (see the equation of the SS1, the SS2 and the SRi, respectively. gXY To maximize system performance of the (6-68)). secondary network, and coordinate balance Traditionally, a set-up phase is performed manner between the transmit powers and according to the cooperative medium access channel gains from secondary sources SS1 and control protocol [25]. The secondary nodes SS1, SS2 to the PR, we allocate the transmit powers
5. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 5 P , P and P as PIg   , at the SRn in (3) can achieve higher value when SS1 SS2 SRi SSSS PR 545 11 comparing with relay selection randomly PIg   and P I g . (take any secondary relay in a group of N SSSSPR22445 SRSRPRii It is worth noting that this allocation proposal is secondary relays). used in many published literature [7, 23, By the SIC procedure completely or partly, the interference part P s h in (2) can be 30-33]. SSSSSR111 n The operation principle of the SIC-SC canceled after the data s1 is decoded protocol occurs in two time slots and is successfully, and thus, the received signal at the presented by mathematical models as follows. secondary relays SRn after the SIC procedure is In the first time slot, the secondary sources SS1 expressed as and SS send the data s and s , respectively, 2 1 2 yyPs h with the same carrier frequency to the SRsSRSSSSnnn 211 SR 1 (4) secondary relays SRi at the same time, where PshIhnSSSS2 SRnSR  , iN 1 ,2 ,. . . ,  . The received signal at the 22nn where  I is a residual interference secondary relays SRi is stated as component at the SRn due to imperfect SIC yPs hPshn , SRSSSSiiii SRSSSS1122 SRSR12 procedure;  0 and  1 express perfect and (2) imperfect interference cancellation at the SRn, where EsEs 22 1 and n denote respectively; hn is modeled as an identical 12  SRi complex normal distribution hn C N 0,6 the AWGNs at the SRi with the same [12, 16] with zero mean and same normalized normalized variance N0 ( E  denote the 2 variance  , and thus gh are also expectation operator). 6 nn exponentially distributed RVs with the PDF as Since  12 (SRi is closer to the SS1 than fxe()  x 6 and the CDF as the SS2), based on the SIC technology, the SRi gn 6 will decode the data s in (2) firstly by 1 Fxe()1 x 6 [24] (see the equations gn considering the signal PshSSSSSR2 as 22i (6-68)). interference. From (2), the received signal-to- The received SINR at the selected interference-plus-noise ratios (SINRs) at SRi to secondary relay SRn to decode s2 is obtained decode s1 is expressed as and manipulated as Pg SSSS11 SR i  Qg 4 SS2 SRn  SRs  . (5) i 1 PgN SRsn 2 gQg   1  SSSS22 SR i 0 SS2 PRn 45 (3)  Qgg If the selected secondary relay SRn decodes 5 SS12 SRSSi PR , successfully both data s1 and s2, a coded data s  Qgggg   445 SS2112 SRSSi PRSS PRSS PR is created by the SC technology as where Q=I/N . 0 s s    s    . In From acquisitions of perfect CSIs, the 1 1 1 2 2 2 1 2 secondary source SS1 only select one secondary this case, because the SRn is nearer to the secondary source SS1 than to the secondary relay (denoted as SRn, nN 1,2, ,  ) so that source SS2, thus, the SRn sends the data s2 to the the decoding capacity is increased and the SS1 with the smaller power parameter number of the pilot channels is decreased. The 2  1  2 and vice versus. secondary relay SRn is obtained as SRarg max g . As a result, the SINR In the second time slot, the SRn broadcasts nSS SR 1 i iN 1,2, , back the coded data s to two secondary sources
6. 6 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 SS1 and SS2, and then the received signals at the sssRRS 2314, where 34 1 and secondary sources SSk are obtained as 34 (the SS1 is the nearer user to receive the yPshn SSSRSRknnkk SSSS data s2) [14].  Ps h  SR1121SRnnk SS (6) 3. Outage Probability Analyses  Pshn  , SR2122SRnnkk SSSS Outage probability at a node is defined where n denote AWGNs at the SSk with SSk as probability that the node cannot decode the same normalized variance N0 , and successfully the desired data or the received k 1,2. data rate at the node is less than a threshold data rate Rth [29,37] RR zth , We assume that the SSk can perfectly cancel the known components including its data sk, i.e., SSSSSRz1, 2 ,n , ss 1 , 2 , n 1,2, , N . P   s h in (6). The The selected secondary relay SRn applies SRnkk 1 2 SR n SS k the SIC procedure which was assigned in the received SINRs at the secondary sources SS k set up phase to decode successively the data are obtained to take sl, where l 1,2 and  from s1 to s2, the SS1 cannot get the data s2 in lk , as the three cases as: 1) the SRn cannot decode the 2 first-ordered data s1 (denoted as RR ), Ph Qg  SRsthn 1 SRlSRnnk SS SRnk SSl  . (7) 2) the SRn gets the data s1 successfully but does SSskl Ng   01212 SRn PR not decode the second-ordered data s2 (denoted Here, we received data rates at the SS1, SS2 and as RRRR  ), or 3) the SRsthSRsthnn 12 SRn as 1 SRn gets both s1 and s2 successfully but the SS1 RZzZz log1bits/s/Hz2  , (8) cannot decode the desired data s2 in the second 2 time slot (denoted as where ½ shows that the proposed SIC-SC RRRRRRSRsthSRsthSSsth   . protocol operates in the two time slots, nn1212 By summing the above cases, the outage ZSSSSSR 12,,n, z s12, s  and nN 1,2, ,  . probability of the secondary source SS1 is For comparison purpose, we also consider a expressed mathematically as random relay selection (RRS) protocol with OP Pr R R SS11 SRn s th fixed power allocations where a collaborative 1 SRi is randomly selected in a group of N Pr RRRR  (9) secondary relays for the two-way relaying, SRnn s12 th SR s th iN 1,2, ,  [17, 34-36]. In particular, from (1), 2 the transmit powers of the SS1 and SS2 are RRRRSR s th  SR s th Pr nn12 maximally set as PRRS  I g and SS111 SS PR  RR SS12 s th PIgRRS  where  1[15]. In SSSS21 PR 2 12 3 addition, the secondary relay SRi applies the SC where Pr mean probability operations technology with power allocation parameters 3 of events  . and 4 to create a coded data as Lemma 1: The probability 1 is solved as
7. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 7 N p SIC (ε = 1) operations as formula at the top of 5 p 1 1 1  next page.  45324p 1  N  pp (10) In the formulas (11),  ab, is an 2453 pp incomplete Gamma function [38] ( see the 1ln,   ppp equation (8.350.2)),  212Rth  Q , 532424 5454 2Rth    Q and a function where: 145  21 ; 6255556 2Rth 7354564 pppQp   . 2 21  4  5  5Q ; functions Proof of the Lemma 2 is provided in 3 p and 4 p are defined respectively as Appendix B. 3 pp  1  4 2  1  4 and The event RR happens SSsth12 p independently with the intersection event of 411 pp ; denotes the binomial N RR and RR , thus we SRn s1 th SRsthn 2 p N! have an equivalent representation of the coefficient . probability in (9) as formulas (12). N pNp!! 3 Proof: Proven in Appendix A. Lemma 2: The probability 2 is given in two cases of the perfect (ε = 0) and imperfect H 2 55 p N 2  5 5 55 1 p 1  ,0 2  5 5  4p 0 N  5  3 pp  2  4  p    p 2 5 4 ln 2 5 5 3 5 3 pp  2 4 5 2  3 pp  4 p (11) 6 N 1 p   Q    p  N e 0,6  2 1 p 5 3 2 5 5 6 4 5 p 2 1  2  5  6 6 5  4Q p 0 N 4 p 0 N  5 3 pp  2 4 p  1 24 p 2  5  5 3 p  p 1 ln ep7  0, , 1 p   p   p   p  pp   7 5 3 2 4  5 3 2 4 2 3 5 4 J 1 3  Pr RRRRSR sthSR sth PrPr RRR log 1  nn12 SS1 sthSS 21 2 sth 2 2 Pr RR (12) SS12 sth QgSR SS 2 Pr21 n 1 2Rth g   1Pr RR SRn PR 12 12 SS12 sth (13) 2Rth g 12  21 The probability Pr RR is solved SRn SS1 SS12 s th Pr gQ with referring the formula (A.2) as SRn PR 2 2Rth 12  3  21 . 2Rth 1  212Q 3   21
8. 8 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 Substituting (12) into (9) and using (13), the Finally, the outage probabilities of the SS1 outage probability of the secondary source SS1 and SS2 in the RRS protocol (denoted as is analyzed by a closed-form expression as OPRRS and OPRRS ) are expressed and solved in SS1 SS2   212Rth  Q 1231212 the similar approach as in the proposed SIC-SC OPSS (14) 1 2Rth 12123   Q  21 protocol. Hence, these outage probabilities are obtained as where 1 and 2 are provided by the RRS Lemmas 1 and 2, respectively. OPRR Pr SRsth SSi1 1 The SS2 cannot get the desired data s1 in 4 only the two cases as: 1) the SRn cannot also  Pr RRRR SRsthSRsthii 12 decode the first-ordered data s1, 2) the SRn 5 transmits the signals containing the decoded RRRR  (16) data s but the SS cannot get the s (denoted as SRsthSRsthii 12 1 2 1 Pr RRRR  . Therefore, the  RR SRsthSSsthn 121 SSsth12 outage probability of the secondary source SS2 2Rth 21   33145 Q is presented and solved as 2Rth 21   331  Q OPRRSSSRsth Pr 21 n RRS OP Pr RSR s R th  Pr RRRR SS2 i 1 SRsthSSsthn 121 (15) Pr RRRR  (17) PrPr RRRR SRi s1 th SS 2 s 1 th 1 SRsthSSsthn 121 2Rth 2Rth 123  21 21 3  4  2Q 4 1 11 2R   Q   21th 2Rth 12123 21 3  4  2Q 2Rth 123121   21  Q where 4 and 5 are inferred from 2Rth 12123  Q   21 Lemmas 1 and 2, and are presented by the below formula (18) and the formula (19) at the where 1 is provided by the Lemma 1. top of this page. From (13-14), we notice that by allocating 5 the transmit power ratios for the data s1 and s2 4 1 in the SC-coded signal s, the decoding outage  43542  (18) probabilities at the secondary sources SS1 and     1ln 42 35 SS2 from the selected secondary relay SRn are 354242  identical, i.e. Pr RRRR Pr . SS1 s 2 th SS 2 s 1 th In the formulas (18-19), 2Rth , 1 21  2  1 In addition, from (14-15), we have a result  Q ,     , , as OPOP due to additional data decoding 212 312 41 4411 SSSS12     ,        Q and of at the selected secondary relay in the 5 5 1 2 6 2 5 5 5 5 6     Q . OP . 734 54 56 SS1 U
9. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 9        5 5 5 42 ln3 2 5 5 , 0  43542     2553542       3452    6  e 0, 6  1  3 5  4  2  5  5  6Q   2 5 5   Q 2      5 6  5 5 65 4Q  3  5  4  2 4 3 5 4 2 (19)     42 3 2 5 5 7 1 ln e 0,7  , 1        3 5 4 2 3 4 5 2 K T 4. Results and Discussions relays N and interference constraint Q increase in two cases of the perfect SICs and imperfect This section presents analysis and SICs. Secondly, these probabilities move to simulation results of the proposed SIC-SC saturation values at the high Q regions, e.g. protocol and the RRC comparison protocol with Q 30 (dB) when N = 8. Thirdly, the outage two cases of perfect SICs (  0 ) and imperfect performance of the secondary source SS is SICs ( 1). The Monte Carlo simulation 2 better than that of the secondary source SS1. results are made to verify the theoretical Fourthly, the outage probability of the derivations in the section 3. The values for secondary source SS2 is not affected by the  m ,1,2,3,4,5,6m  are referenced in [12, 18, cases of the SIC operations and the secondary 29, 39, 40]. In all the subsequent results, the source SS1 achieves smaller outage threshold data rate Rth is fixed to 3 (bps/Hz) and probabilities in the perfect SICs. Fifthly, the the normalized variance of the additive white proposed SIC-SC protocol outperforms the Gaussian noises (N0) is set to 1. In addition, RRS comparison protocol where the random blue and red markers denote simulated values relay selection and fixed power allocations are of the secondary sources SS1 and SS2, considered. Finally, the derived theory values respectively, and black solid lines present (black solid lines) are precise to the simulation theoretical analyses. ones (marker symbols). These conclusions are Figure 2 presents the outage probabilities of explained by increasing the diversity amount the secondary sources SS1 and SS2 in the and the transmit powers. In addition, the protocols SIC-SC and RRS versus Q (dB) when received SINRs at the secondary source SS1 are 1234 20  dB ,10  dB ,1 dB ,1 dB , weakened by combining effects of the interference component from the data-carried   0.5dB,5dB and number of the 56 signal, the residual interference by imperfect secondary relays is examined at 5 and 8 SICs and interference constraints of the primary N 5,8 . Power allocation parameters for network whereas the perfect and imperfect SIC events existed in the received SINR  as the RRS comparison protocol are set to fixed SRsn 2 values as 1213 0.3,10.7,0.4 and in (5) and (8) are not considered in taking the own data s2 of the secondary source SS2 (see the 43 10.6 . As shown in the Figure 2, some observations for the proposed SIC-SC formula (15)). One more thing, the RRS protocol are listed as follows. Firstly, the outage protocol does not depend on the number of the secondary relays and can be used if there is at probabilities of the secondary source SS1 and least one cooperative secondary relay. SS2 decrease when the number of the secondary
10. 10 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 Figure 2. Outage probabilities of the SIC-SC and Figure 3. Outage probabilities of the SIC-SC and RRS protocols versus Q (dB) when Ψ1 = 20 (dB), Ψ2 RRS protocols versus (dB) when Ψ1 = 20 (dB), = 10 (dB), Ψ3 = 1 (dB), Ψ4 = 1 (dB), Ψ5 = 0.5 (dB), Ψ2 = 10 (dB), Ψ3 = 1 (dB), Ψ5 = 0.5 (dB), Ψ6 = −5 (dB), , N = 8 Ψ6 = −5 (dB), 1 0 . 3, 21 1 0 . 7 , 3 0 . 4 , 1234 0.3,0.7,0.4,0.6 and Q = 10 (dB). 43 10.6 and N 5,8 . Figure 4 shows the outage probabilities of Figure 3 presents the outage probabilities the SIC-SC and RRS protocols versus 1 (dB) versus 4 (dB) when   1220dB,10dB, in situations as   2310dB,1 dB, 3561 1 dB  ,0.5 dB ,5 dB ,0.3,  45123 1 dB ,0.5 dB ,0.3,0.7,0.4,   0.7,0.4,0.6 , N = 8 and Q = 10 234  0.6 , N = 8, Q = 10 (dB) and (dB). In Figure 3, in the proposed SIC-SC 4 protocol, the outage probability of the  6 10(dB),5(dB)  . As shown in Figure 4, secondary source SS1 has a little reduction the system performance of the protocols SIC- before is increased in the two cases of the SC and RRS is enhanced according the perfect and imperfect SICs whereas the outage increasing of the channel power 1 because of probability of the secondary source SS2 is high-success decoding of the first data in the always increased because of variability between SIC technology. In addition, the lower residual the interference channels and transmit powers. interference parameters lead to the higher Furthermore, the system performance of the performances of the secondary source SS1. The RRS comparison protocol in terms of the case of the perfect SICs is viewed as outage probabilities of the SS1 and SS2 is  6 (dB). In both protocols SIC-SC and declined when the interference channel power RRS, the SIC and SC technologies improve the from the SS1 to the PR occurs in the spectrum utilization efficiency by decreasing increasing sense. A reason for these results is the number of the time slots from the secondary that the transmit power allocation in the SIC-SC sources to the secondary relays and vice versus, protocol is adjusted according to the but the SIC-SC protocol performs better by the interference channel powers 4 , e.g. the adaptive power allocation parameters transmit power of the SS1 (denoted as P ) will SS1  112  and  212  in the coded be decreased when the 4 increase. s as a function of the channel powers 1 .
11. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 11 performing the Monte Carlo simulations. The simulation and analysis results shown that 1) the system performance of the proposed SIC-SC protocol in terms of the outage probabilities was enhanced by increasing the number of the secondary relays and the interference constraint as well as decreasing the residual interference powers, and 2) the non-identical outage performances of the secondary sources depended on both interference channel powers to the primary receiver and desired channel powers to the selected secondary relay from the secondary sources, and 3) the performance of the nearer secondary source is worse than that of the Figure 4. Outage probabilities of the SIC-SC and distant secondary source, and finally, the RRS protocols versus  (dB) when Ψ2 = 10 (dB), 1 system performance of the proposed SIC-SC Ψ3 = 1 (dB), Ψ4 = 1 (dB), Ψ5 = 0.5 (dB), 1 0 . 3, protocol outperforms that of the RRS protocol 234 0.7,0.4,0.6 , N = 8, Q = 10 (dB) and in terms of the outage probabilities.  6 10(dB),5(dB)  . Appendix A: Proof of Lemma 1 Substituting (7) into the probability of as 5. Conclusions 1 in (9), the 1 is expressed as In this paper, we analyzed the underlay 1 Pr log 1  R 12 SRn sth1 two-way DF scheme in which the secondary 2 relays use the SIC technology to decode the 2R Pr21  th data of two secondary sources sequentially, and SRn s1  Qgg then make a coded signal by the SC technology, 5 SS12 SRn SS PR  Qgggg   known as the SIC-SC protocol. The SIC-SC Pr 445 SS2112 SRn SS PRSS PR SS PR protocol was designed to operate in two time 2Rth 21 slots under effects from the interference 2Rth ggSS SRSS SR 21  4 constraint of the primary receiver and residual 12nn gg interference of the imperfect SIC processes. SS12 PRSS PR 5  Transmit powers provided to carry the data Pr 1 2Rth were allocated dynamically according to 21 45  channel powers of interference and 5Q transmission. The secondary relay was selected (A.1) 2 from considering strongest channel gain subject to increase decoding capacity of the first data fxgggg Fxdx 12 , SS2211 SRnn SS PRSS SR SS PR and decrease collection time of the CSIs. The 0 where fx are Fx are the identical underlay two-way DF operation with ggXY ggXY the random relay selection and the fixed power PDF and CDF of the RVs gg, allocations (called the RRS protocol) was also XY investigated to compare the proposed SIC-SC XY,SS ,SS 12 ,SR ,PR n  , nN 1,2, ,  . protocol. The closed-form outage probability By referring from [29] (see equations expressions were derived from mathematical (24–25)), the fxgg is obtained as manipulations and verified exactly by SS22 SRSSn PR
12. 12 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 Fxgg 411 pp , the Lemma 1 is proven SS22 SRSSn PR fxgg SS22 SRSSn PR x completely. (A.2) 25 2 . Appendix B: Proof of Lemma 2  25  x From (8) and (A.1), the probability 2 as in The Fxgg is expressed as SSSRSSPR11n (9) is expressed as g 5QggSS SRSS PR Fxx Pr SS1 SRn 12n ggSS SRSS PR  Qgggg   11n g 445 SS2112 SRSSn PRSS PR SS PR SS1 PR Pr 2Rth 2 21 Pr gxgSS SRSS PR (A.3) 1 11n  log1  R 2 SRsthn 2 2 fyFxydygg , ggSS SRSS SR1 SS11 PRSS SR n 12nn  0 gg 2 SS12 PRSS PR Pr where Fxg is the CDF of the RV SSSR1 n PgSSSS SR 2R  22n 21th 2 g and is expressed as (see the equation  I hNn 0 S S1 S R n gg (7–14) in [24]) SS12 SRSSnn SR 1  2 N gg N p p SS12 PRSS PR  xpx 11(A.4) Fxeeg 11. SS1 SRn  g p 0 N Pr. SS2 SRn 2R    21th Q g 454 SS2 PR p  In (A.4), denotes the binomial 5 (B.1) N    212Rth g 454 n p N! To solve the in (B.1) by closed-form coefficient . 2 N pNp!! expressions, we consider two cases of perfect Then, the Fxgg is solved as SICs  0 and imperfect SICs  1 as SS11 SRSSn PR y N e 4 p p follows. Fxedy 1 py 1 ggSS SRSS PR  11n  N - Perfect SICs ( 0 ): By referring from 0 4 p 0 (A.5) (A.2) and using (A.5), the 2 is obtained as N p 1 p 5 1  . N 212 fxFxdxgggg 1  p 0 14  px SS2211 SRnn SS PRSS SR SS PR 0 Substituting (A.2) and (A.5) into (A.1), the Fgg 5 is manipulated equivalently as SS22 SRn SS PR (B.2) 1 5 p N fxgggg Fxdx 12 25 p 1 SS2211 SRnn SS PRSS SR SS PR 11  dx 0 2  N   px 0 25  x p 0 14 12 N (A.6) 55 p p    1 2 5  1 N p N p 1 dx 25 5   p 0 1  2  5 .   N 2 5 dx p 0 0 14 24ppxx  125   . 2 0  ppxx    By performing variable transformations as 14 24 12 55 2 Also performing as (A.6), 2 is solved as in 1  p  p y yx   and 5 3 2 4 25 xp 4 Lemma 2 with the case  0 . 25 - Imperfect SICs ( 1): The 2 in this , where 3 pp  1  4 2  1  4 and case ( 1) is presented as
13. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 13 5 1 Qx [4] T.M.C. Chu, H. Zepernick, Performance 212 fxFyggg 1  Optimization for Hybrid Two-Way Cognitive nSS SRSS PR 11n 00 Cooperative Radio Networks With Imperfect fydydxgg Spectrum Sensing, IEEE Access 6 (2018) 70582- SS22 SRSSn PR 70596. x 6 e  55 1 Qx [5] K. Ho-Van, T. Do-Dac, Security Analysis for    1 Qx 0 6255 Underlay Cognitive Network with Energy- p 1 N p 1 Scavenging Capable Relay over Nakagami-m  Fading Channels, Wireless Communications and  p 0  N  pp 45324 Mobile Computing 2019 1-16.  1 Qxp 2  10.1155/2019/5080952. 55254  2555324    1 Qxpp [6] X. Zhang, Z. Zhang, J. Xing, R. Yu, P. Zhang, W. dx. Wang, Exact Outage Analysis in Cognitive Two-  2553   1 Qxp ln Way Relay Networks With Opportunistic Relay  pQxp 1 2354  (B.3) Selection Under Primary User’s Interference, IEEE Transactions on Vehicular Technology 64(6) (2015) ln  By extending , changing varibables 2502-2511. and then solving the integrals in (B.3), the TVT.2014.2346615. [7] T.T. Duy, H.Y. Kong, Exact outage probability of probability 2 is answered as in Lemma 2 with cognitive two-way relaying scheme with the remaining case  1. Hence, the Lemma 2 opportunistic relay selection under interference is verified commpletely. constraint, IET Communications 6(16) (2012), 2750- 2759. 10.1049/iet-com. 2012.0235. Acknowledgments [8] H.V. Toan, V.N.Q. Bao, Opportunistic relaying for cognitive two-way network with multiple primary This research is funded by Vietnam receivers over Nakagami-m fading, presented at National Foundation for Science and 2016 International Conference on Advanced Technology Development (NAFOSTED) under Technologies for Communications (ATC), Hanoi city, 2016, 141-146. grant number 102.04-2019.13. Khuong Ho-Van acknowledges the support of time and facilities [9] H.V. Toan, V.N.Q. Bao, H. Nguyen-Le, Cognitive from Ho Chi Minh City University of two-way relay systems with multiple primary Technology (HCMUT), VNU-HCM, for receivers: exact and asymptotic outage formulation, this study. IET Communications 11(16) (2017), 2490-2497. 0400. [10] H.V.Toan, V.N.Q. Bao, K.N. Le, Performance References analysis of cognitive underlay two-way relay networks with interference and imperfect channel [1] P. Popovski, H. Yomo, Physical Network Coding in state information, EURASIP Journal on Wireless Two-Way Wireless Relay Channels, presented at Communications and Networking 2018 53 (2018). 2007 IEEE International Conference on Communications (ICC), Glasgow, 2007, 707-712. [11] S. Solanki, P.K. Sharma, P.K. Upadhyay, Adaptive Link Utilization in Two-Way Spectrum Sharing [2] Z. Cao, X. Ji, J. Wang, S. Zhang, Y. Ji, J. Wang, Relay Systems Under Average Interference Security-Reliability Tradeoff Analysis for Underlay Constraints, IEEE Systems Journal 12(4) (2018), Cognitive Two-Way Relay Networks, IEEE 3461-3472. Transactions on Wireless Communications 18(12) JSYST.2017.2713887. (2019) 6030-6042. [12] X. Yue, Y. Liu, S. Kang, A. Nallanathan, Y. Chen, TWC.2019.2941944. Modeling and Analysis of Two-Way Relay Non- [3] J. Mitola, G.Q. Maguire, Cognitive radio: making Orthogonal Multiple Access Systems, IEEE software radios more personal, IEEE Personal Transactions on Communications 66(9) (2018), Communications 6(4) (1999) 13-18. org 3784-3796. /10.1109/98.788210. 2018.2816063.
14. 14 P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2021) 1-15 [13] X. Zou, B. He, H. Jafarkhani, An Analysis of E98-B(4) (2015), 661-672. TwoUser Uplink Asynchronous Non-orthogonal Multiple Access Systems, IEEE Transactions on [23] K. Tourki, K.A. Qaraqe, M. Alouini, Outage Wireless Communications 18(2) (2019), 1404-1418. Analysis for Underlay Cognitive Networks Using Incremental Regenerative Relaying, IEEE [14] Z. Yang, Z. Ding, P. Fan, N. Al-Dhahir, The Impact Transactions on Vehicular Technology 62(2) (2013) of Power Allocation on Cooperative Nonorthogonal 721-734. Multiple Access Networks With SWIPT, IEEE 2012.2222947. Transactions on Wireless Communications 16(7) [24] A. Papoulis, S.U. Pillai, Probability, random (2017), 4332-4343. variables and stochastic processes, 4th ed., McGrawHill, New York, 2002. [15] P.N. Son, T.T. Duy, K. Ho-Van, SIC-Coding [25] L. Pei, T. Zhifeng, L. Zinan, E. Erkip, S. Panwar, Schemes for Underlay Two-Way Relaying Cooperative wireless communications: a cross-layer Cognitive Networks, Wireless Communications and approach, IEEE Wireless Communications 13(4) Mobile Computing 2020 1-24. (2006) 84-92. 10.1155/2020/8860551. MWC.2006.1678169. [16] M.F. Kader, M.B. Shahab, S.Y. Shin, Exploiting [26] A. Ghasemi, E.S. Sousa, Fundamental limits of Non-Orthogonal Multiple Access in Cooperative spectrum-sharing in fading environments, IEEE Relay Sharing, IEEE Communications Letters 21(5) Transactions on Wireless Communications 6(2) (2017) 1159-1162. (2007) 649-658. 10.1109/LCOMM.2017.2653777. 2007.05447. [17] X. Yue, Y. Liu, S. Kang, A. Nallanathan, Z. Ding, [27] J. M. Peha, Approaches to spectrum sharing, IEEE Spatially Random Relay Selection for Full/Half- Communications Magazine 43(2) (2005) 10-12. Duplex Cooperative NOMA Networks, IEEE 1391490. Transactions on Communications 66(8) (2018) [28] H. Kim, S. Lim, H. Wang, D. Hong, Optimal Power 3294-3308. Allocation and Outage Analysis for Cognitive Full 2018.2809740. Duplex Relay Systems, IEEE Transactions on [18] Y. Liu, Z. Ding, M. Elkashlan, J. Yuan, Wireless Communications 11(10) (2012) 3754-3765. Nonorthogonal Multiple Access in Large-Scale 2012.083112.120127. Underlay Cognitive Radio Networks, IEEE [29] P.N. Son, T.T. Duy, Performance analysis of Transactions on Vehicular Technology 65(12) underlay cooperative cognitive full-duplex networks (2016) 10152-10157. with energy-harvesting relay, Computer TVT.2016.2524694. Communications 122 (2018) 9-19. [19] Y. Song, W. Yang, Z. Xiang, N. Sha, H. Wang, Y. 10.1016/j.comcom.2018.03.003. Yang, An Analysis on Secure Millimeter Wave [30] T.V. Nguyen, T. Do, V.N.Q. Bao, D.B.d. Costa, B. NOMA Communications in Cognitive Radio An, On the Performance of Multihop Cognitive Networks, IEEE Access 8 (2020), 78965-78978. Wireless Powered D2D Communications in WSNs, IEEE Transactions on Vehicular Technology 69(3) [20] X. Ding, T. Song, Y. Zou, X. Chen, L. Hanzo, (2020) 2684-2699. Security-Reliability Tradeoff Analysis of Artificial Noise Aided Two-Way Opportunistic Relay [31] Y. Ruan, Y. Li, C. Wang, R. Zhang, H. Zhang, Selection, IEEE Transactions on Vehicular Energy Efficient Power Allocation for Delay Technology 66(5) (2017) 3930-3941. Constrained Cognitive Satellite Terrestrial Networks Under Interference Constraints, IEEE Transactions [21] B. Zheng, M. Wen, F. Chen, J. Tang, F. Ji, Secure on Wireless Communications 18(10) (2019) 4957- NOMA Based Full-Duplex Two-Way Relay 4969. 2019.2931321. Networks with Artificial Noise against [32] H. Gao, S. Zhang, Y. Su, M. Diao, M. Jo, Joint Eavesdropping, presented at 2018 IEEE International Multiple Relay Selection and Time Slot Allocation Conference on Communications (ICC), Kansas City, Algorithm for the EH-Abled Cognitive Multi-User 2018, 1-6. Relay Networks, IEEE Access 7 (2019) 111993- 10.1109/ICC.2018.8422946. 112007. [22] P.N. Son, H.Y. Kong, Exact Outage Analysis of ACCESS.2019.2932955. Energy Harvesting Underlay Cooperative Cognitive Networks, IEICE Transactions on Communications
15. P.N. Son et al. / VNU Journal of Science: Comp. Science & Com. Eng., Vol. 37, No. 1 (2020) 1-15 15 [33] H. Arezumand, H. Zamiri-Jafarian, E. Soleimani- Networks, Wireless Communications and Mobile Nasab, Exact and Asymptotic Analysis of Partial Computing 2019 1-7. Relay Selection for Cognitive RF-FSO Systems With Non-Zero Boresight Pointing Errors, IEEE [37] A.A. Nasir, Z. Xiangyun, S. Durrani, R.A. Kennedy, Access 7 (2019) 58611-58625. Relaying Protocols for Wireless Energy Harvesting 10.1109/ACCESS.2019.2914480. and Information Processing, IEEE Transactions on [34] P.N. Son, H.Y. Kong, Energy-Harvesting Relay Wireless Communications 12(7) (2013) 3622-3636. Selection Schemes for Decode-and-Forward Dual- 2013.062413.122042. Hop Networks, IEICE TRANSACTIONS on [38] R.I. Gradshteyn, I.M. Ryzhik, A.Jeffrey, D. Communications E98-B(12) (2015) 2485-2495. Zwillinger, Table of integral, series and products, 7th ed., Elsevier, Amsterdam, 2007. [35] T.N. Nguyen, T.H. Quang Minh, P.T. Tran, M. [39] H. Haiyan, L. Zan, S. Jiangbo, G. Lei, Underlay Voznak, T.T. Duy, T.-L. Nguyen, P.T. Tin, cognitive relay networks with imperfect channel Performance enhancement for energy harvesting state information and multiple primary receivers, based two-way relay protocols in wireless ad-hoc IET Communications 9(4) (2015) 460-467. networks with partial and full relay selection methods, Ad Hoc Networks 84 (2019) 178-187. [40] B. Zhong, Z. Zhang, Opportunistic Two-Way Full- Duplex Relay Selection in Underlay Cognitive [36] L. Pan, Z. Li, Z. Wang, F. Zhang, Joint Relay Networks, IEEE Systems Journal 12(1) (2018) 725- Selection and Power Allocation for the Physical 734. 2016.2514601. Layer Security of Two-Way Cooperative Relaying I