2016 International Conference on Advanced Technologies for Communications (ATC)

Performance Analysis of Network-Coded Two-Way

Dual-Hop Mixed FSO/RF Systems

Thang V. Nguyen1 , Tu V.M. Pham1 , Thu A. Pham1 , Hien T.T. Pham1 , Ngoc T. Dang1,2 , and Anh T. Pham2
1
Faculty of Telecommunications, Posts and Telecommunications Institute of Technology, Hanoi, Vietnam
2
Computer Communications Lab., The University of Aizu, Aizuwakamatsu, Japan
Email: [email protected]

Abstract—In this paper, a novel two-way relaying technique FSO link RF link
based on network coding is proposed to improve the throughput
of dual-hop mixed free-space optics (FSO)/radio-frequency (RF)
communication systems. We investigate the performance of the Phase 1 Phase 1
proposed systems over Gamma-Gamma atmospheric turbulence
Node A Relay Node B
and Rice/Rayleigh channels. The effect of pointing error on FSO
Phase 2 Phase 2
link is also considered in our analysis. Closed-form expressions for
end-to-end outage probability and bit error rate are analytically
formulated, taking into account the physical layer impairments
Fig. 1. Model of network-coded two-way dual-hop mixed FSO/RF system.
from both FSO and RF links as well as the receiver noise. The
numerical results show the feasibility of using network-coded two-
way relaying in dual-hop mixed FSO/RF systems. In addition, the
comparison between the magnitudes of negative impacts caused done for mixed RF/FSO systems with Nakagami-m fading
by FSO and RF links is discussed. Based on this comparison, channel for RF link [11]. Also, the same system model is
the recommendation for locating the relay node is given so that studied by Samimi et al. in [8] and [11]. However, unlike
better performance could be obtained. previous work [7], the authors assume that FSO link has
Keywords—Free-space optics (FSO), network coding, two-way
M-distributed fading channel instead of Gamma-Gamma one.
relaying, mixed FSO/RF systems. Besides, the effect of pointing error on the outage probability
is also investigated. In [9], the performance of two-hop mixed
RF/FSO is analyzed in underlay cognitive networks, where
I. I NTRODUCTION RF link is subject to an interference constraint. Similarly,
Free-space optics (FSO) and radio-frequency (RF) systems the effect of co-channel interference is considered in mixed
have gained a lot of attention for high-data rate wireless RF/FSO cooperative systems [13]. Last but not least, multi-
connectivity in next-generation broadband access and backhaul user dual-hop relaying system over mixed RF/FSO links has
networks [1]-[3]. Besides the advantage of high-data rate, been studied in [10]. In this system model, multiple users
both FSO and RF systems have the advantage of flexible simultaneously transmit their data to the relay via RF links.
deployment. As FSO and RF channels are not affected in the The relay is based on decode-and-forward (DF) relaying and
same way by weather and fading conditions, there have been V-BLAST techniques to decode each user’s data stream.
many proposals to combine FSO and RF links to form a hybrid
Relaying techniques, which can be categorized into one-
or mixed FSO/RF system that allows for better link reliability
way and two-way relaying, have been proposed to both RF
and availability [3]-[13].
wireless and FSO systems. However, most studies related
There are two approaches to integrate FSO and RF tech- to mixed FSO/RF systems only considered the use of one-
nologies including switch-over hybrid FSO/RF and dual-hop way relaying scheme [7]-[13]. In this paper, we therefore
mixed FSO/RF systems. The first one, which uses parallel propose to apply two-way (or bi-directional) relaying to dual-
transmission of FSO and RF for the same path, inherits the hop mixed FSO/RF systems. At the relay node, DF relaying
complementary behavior of two technologies. This approach, based on network coding (NC) is used to establish a two-
however, requires hardware switching/combining between the way communications with the assumption that all nodes in the
FSO and RF links [4]-[6]. Based on relaying technique, the mixed FSO/RF systems are half-duplex. Thanks to the use of
second approach forms a dual-hop system by using a FSO NC, the number of phases required to exchange two packets
link at one hop and a RF link at the other as shown in Fig. 1. between two nodes via a relay node reduces from 4 phases to
The dual-hop mixed FSO/RF systems could be used to connect 2 phases and thus the throughput improvement is 100% [14].
between the access network and the core network [3]. In the first phase, Node A and Node B transmit their data
(SA and SB ) to the relay node. In the second one, the relay
Many studies have been proposed to analyze the perfor-
node forms a network-coded packet (SR = f (SA , SB )), then
mance of dual-hop mixed FSO/RF systems [7]-[13]. First,
forwards the coded-signal to Node A and Node B.
the performance of an asymmetric RF/FSO link is analyzed
with the assumption that the RF link and FSO link have the We study the performance of proposed network-coded two-
Rayleigh fading channel and Gamma-Gamma fading channel, way dual-hop mixed FSO/RF system over asymmetric chan-
respectively [7]. Amplify-and-forward (AF) fixed-gain relaying nels, which are modeled by Gamma-Gamma/Rice and Gamma-
technique is considered in this study. Similar work has been Gamma/Rayleigh fading channels. The effect of pointing error

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
978-1-5090-2710-1/16/$31.00 ©2016 IEEE 70
 2016 International Conference on Advanced Technologies for Communications (ATC)

on FSO link is also taken into account. The performance bit duration, it is often assumed that hl is deterministic while
is investigated in terms of bit error rate (BER) and outage ha and hp are random variables.
probability versus various system parameters such as the
transmission distance and the transmitted power. Optical signal is attenuated while propagating over atmo-
spheric channel due to absorption and scattering. The channel
The rest of the paper is organized as follows. Section II loss coefficient can be described accordingly to Beers-Lambert
presents the model of network-coded two-way dual-hop mixed law as
FSO/RF systems. In Section III, the statistical models of FSO hl = exp(−al LF SO ), (4)
and RF channels are introduced. In Section IV, we derive
expressions for outage probability and BER of the proposed where al the attenuation coefficient and LF SO is the distance
systems. The numerical results and discussions are presented of FSO link.
in Section V. Section VI concludes the paper.
In our analysis, Gamma-Gamma distribution is used to
II. N ETWORK -C ODED T WO -WAY D UAL -H OP M IXED study the system performance in moderate-to-strong turbulence
FSO/RF S YSTEMS regime. The probability distribution function (PDF) of the
intensity fluctuation due to atmospheric turbulence thus can
A network-coded two-way dual-hop FSO/RF system that be expressed as [15]
consists of two ending nodes (Node A and Node B) and a relay (α+β)
node (Node R) as illustrated in Fig. 1. In this mixed system, 2(αβ) 2 (α+β)
−1
 p 
the A–R and B–R links are assumed to be FSO and RF links, fha (ha ) = ha 2 Kα−β 2 αβha , (5)
Γ (α) Γ (β)
respectively, where the FSO link is Gamma-Gamma distributed
and the RF link is modeled by the Rayleigh/Rice distribution. where Kv (.) is the modified Bessel function of the second kind
R is equipped with NC scheme and hybrid optical/RF capa- and order v. Γ(.) is the standard gamma function. Parameters
bilities including an optical source, a photodetector, and RF α > 0 and β > 0 can be adjusted to achieve a good agreement
transceiver. between fha (ha) and measurement data corresponding to
In the first phase, Node R receives the signals from Node various turbulence conditions. Alternatively, α and β for the
A and Node B. It is worth noting that signal transmitted from case of spherical wave propagation can be directly linked to
Node A is modulated using on-off keying (OOK) whereas bi- physical parameters [15]
nary phase shift keying (BPSK) is used to modulate the signal    −1
at Node B. At Node R, received signals are converted to digital 2
0.49σR
baseband signals (i.e., SA and SB ) before being encoded using α = exp   7/6  − 1 (6)
   
NC scheme. In general, there are many types of network- 12/5
1+ 1.11σR
coding mappings for f (SA , SB ). For simple illustration, we
assume that f (SA , SB ) is a XOR operation as follows    −1
2
SR = SA ⊕ SB , (1) 0.51σR
β = exp   5/6  − 1 , (7)
   
12/5
where ⊕ denotes the pairwise application of bit-by-bit XOR 1 + 0.69σR
over SA and SB .
where σR is the Rytov variance, which is defined as
In the second phase, the relay node broadcasts SR to the
Node A and Node B. When receiving SR , the two nodes obtain  7/6
the information from SR using the self-information as follows 2 2π 11/6
σR = 1.23 Cn2 LF SO , (8)
λ
SA ⊕ SR = SA ⊕ (SA ⊕ SB ) = SB
and SB ⊕ SR = SB ⊕ (SA ⊕ SB ) = SA . (2) where λ is the wavelength. The unitless Rytov variance rep-
2
resents the strength of the turbulence. σR is ruled by Cn2 , the
index of refraction structure parameter, and the distance LF SO .
III. C HANNEL M ODELS
By using the assumptions and methodology described in
A. FSO channel model [16], i.e., the detection aperture is circular with the radius of
Three major parameters of FSO channel including channel r and optical beam is Gaussian, the PDF of hp can be derived
loss, atmospheric turbulence-induced fading, and pointing er- as
rors are considered in our analysis. The mathematical model γp 2 γ 2 −1
fhp (hp ) = γ 2 (hp ) p , (9)
of channel coefficient can be expressed as A0p
h = hl ha hp , (3)
where γp = ωzeq /2σs is the ratio between the equivalent beam
where hl and ha are the channel loss coefficient and the in- radius and the pointing error displacement standard deviation
tensity fluctuation due to atmospheric turbulence, respectively. (jitter) at the receiver (σs ). The parameter
√ ωzeq can be calcu-
√ 2
hp is the fraction of power collected by a photo-detector (PD). lated using the √ relations v = πr/ 2ωz , A0 = [erf (v)]
2 2 2
This parameter depends on the relative distance between the and ωzeq = ωz πerf (v) /2v exp(−v ), where erf(.) is the
location of PD and the center of received optical beam thus error function and ωz is the beam waist (radius calculated at
it represents the effect of pointing error. During a considering e−2 ) at the distance z (see [16] for more detail).

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
71
 2016 International Conference on Advanced Technologies for Communications (ATC)

As mentioned in (3), h = hl ha hp . By expressing the Kv (.) γB is the instantaneous SNR of the RF link. Furthermore, the
in terms of Meijers G-function and making simplification, the CDF of Rician distribution can be expressed as
PDF of the channel state is given as [17] s !
√ 2(1 + K)
αβγp 2 FγB −Rice (γB ) = 1 − Q1 2K, γB , (17)
 
αβ γp 2
fh (h)= G3,0 h| . (10) γ̄B
A0 hl Γ(α)Γ(β) 1,3 A0 hl γp 2−1, α−1, β−1
where Q1 (., .) is the Marcum Q1 -function.
The instantaneous signal-to-noise ratio (SNR) of the FSO
link between node A and the relay node is defined as follows In the case of urban environment, the RF transmission link
spanning from Node B to Node R is subjected to multipath
PA 2 ℜ2 h2 fading due to non-line-of-sight (NLOS) links. In this case,
γA = 2 , (11) RF channel can be modeled by the Rayleigh distribution [18].
σA
Consequently, the PDF of instantaneous SNR of the RF link,
where γA is instantaneous SNR. PA is the transmitted op- which obeys an exponential distribution, can be expressed as
2
tical power, σA is variance of additive white Gaussian noise  
(AWGN). ℜ is the responsivity of the photodetector. The PDF 1 γB
fγB −Rayleigh (γB ) = exp − , (18)
of instantaneous γA is given as γ̄B γ̄B
γp 2
 r 
γA γp 2 + 1 where γB is the average SNR of the RF link. In this case, the
fγA (γA ) = G3,0 αβ | , (12)
2γA Γ(α)Γ(β) 1,3 γ̄A γp 2 , α, β CDF is given as
 
where γ̄A is the average SNR between two FSO nodes. In −γB
FγB −Rayleigh (γB ) = 1 − exp . (19)
addition, the cumulative distribution function (CDF) is can be γ̄B
expressed as
IV. P ERFORMANCE A NALYSIS
γp 2
 r 
γA 2, γp 2 + 1
FγA (γA ) = G3,1 αβ | . (13)
2Γ(α)Γ(β) 2,4 γ̄A γp 2 , α, β, 1 A. Outage probability
Outage probability is an important performance metric for
B. RF channel models evaluating wireless communication systems in general and
RF channel can be modeled by either Rician distribution FSO systems in particular. For a dual-hop FSO/RF system,
or Rayleigh distribution. Rician channel model is used to outage happens when any of the links i.e., (A-R) or (B-R) fail
characterize the fading in RF propagation environments where and is defined as
there is a strong direct light-of-sight (LOS) path between the Pout = Pr [min(γA , γB ) ≤ γth ], (20)
transmitter and the receiver. The average power gain of the RF
link is given by [4] where γth is the threshold SNR. γA and γB are instantaneous
  SNRs of the FSO and RF links, respectively. Since, γA and
4πL γB are independent, (20) can be re-written as
gB [dB] = GT +GR − 20log10
λB
−αoxy LRF − αrain LRF , (14) Pout = 1 − Pr (γA > γth )Pr (γB > γth ). (21)

where GT and GR denote the transmit and receive antenna As P r(γA > γth ) = 1 − P r(γA < γth ) = 1 − FγA (γth ) and
gains, respectively. λB is the wavelength of the RF system. P r(γB > γth ) = 1 − P r(γB < γth ) = 1 − FγB (γth ), the
αoxy and αrain are the attenuation caused by oxygen absorp- outage probability in (21) can be expressed as
tion and rain, respectively. LRF is the distance of RF link. Pout = FγA (γth ) + FγB (γth ) − FγA (γth )FγB (γth ), (22)
2
The noise variance in the RF link is given by [4] σB = where FγA (γth ) and FγB (γth ) can be derived from (13) and
BN0 NF , where B is the RF bandwidth, N0 is the noise power (17) or (19), respectively.
spectral density, and NF is the noise figure of the receiver. The
average SNR of the Rician channel is defined as
B. Bit error rate
PB g B
γ̄B = 2 , (15) Bit error rate of the dual-hop FSO/RF system can be
σB expressed as
where PB is the transmitted RF power.
Pe = PF SO (e)[1−PRF (e)]+PRF (e)[1−PF SO (e)], (23)
The PDF of Rician distribution with instantaneous SNR
where PF SO (e) and PRF (e) are the average BER of FSO
per bit is given by [18]
  link and RF link, respectively.
K +1 γB
fγB −Rice (γB ) = exp − (K + 1) −K Denoting p(1) and p(0) are the probabilities of sending bit
γ̄B γ̄
 r B “1” and bit “0”, the BER of a FSO link using OOK is given
γB by PF SO (e) = p (1) p (e|1) + p (0) p (e|0), where p(e|1) and
×I0 2 K(K + 1) , (16)
γ̄B p(e|0) are the conditional bit error probabilities. Assuming that
p(1) = p(0) = 1/2 and p(e|1) = p(e|0), the conditional bit
where K is the Rician factor that depends on various link pa- error probabilities can be computed as
rameter, such as antenna height, link distance and environment. √
I0 is the 0-th order modified Bessel function of the first kind. Pe,F SO (e|h) = p (e|1, h) = p (e|0, h) = Q ( γA ) , (24)

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
72
 2016 International Conference on Advanced Technologies for Communications (ATC)

−1
TABLE I. S YSTEM PARAMETERS AND C ONSTANTS .
10

Name Symbol Value

−2 RF link
10
Carrier frequency fc 60 GHz
Bandwidth B 250 MHz
C2 −14
= 5×10 Transmit antenna gain Gt 44 dBi
−3 n
10
Outage Probability

Receive antenna gain Gr 44 dBi

Oxygen attenuation αoxy 15.1 dB/km
C2 = 10−14 Rain attenuation (clear air) αrain 0 dB/km
n
−4 Rician factor K 6 dB
10
Noise power spectral density N0 -114 dBm/MHz
Receiver noise figure NF 5 dB
−5 FSO link
10
Bit rate Rb 1 Gbps
Wavelength λ 1550 nm
2
−6 Refractive index structure coeff. Cn 10−15 m−2/3
10 Responsivity ℜ 0.8 W/A
2
Gamma−Gamma/Rice channel Noise variance σA 10−16 A2 /Hz
−7
Gamma−Gamma/Rayleigh channel Attenuation coefficient al 0.1 km−1
10 Receiver diameter 2a 20 cm
1000 1500 2000 2500 3000 Beam radius at 1 km ωz 2.0 m
Total distance (m) Jitter standard deviation σs 10 cm

Fig. 2. Outage probability versus the total distance with PA = PB = 10

dBm, LRF /LF SO = 1/2, and γth = 0 dB. V. N UMERICAL R ESULTS
In this section, we show numerical results of the network-
where the Gaussian Q function denoted as Q(.) is related√
to the coded two-way dual-hop mixed FSO/RF systems. The FSO
complementary error function erfc(.) by erfc (x) = 2Q( 2x). link is modeled as the Gamma-Gamma fading channel with
γA is the SNR of FSO link, which is derived from (11). The pointing error while the RF link is modeled as either Rayleigh
average BER, PF SO (e), can be obtained by averaging (24) or Rice fading channel. The RF link operates at millimeter-
over the PDF of h as follows wave band (i.e., 60 GHz). Other system parameters and
constants are shown in Table I.
∞
PF SO (e) = ∫ fh (h)Pe,f so (e|h) dh (25) Figure 2 shows the outage probability of the dual-hop
0
mixed FSO/RF system versus the total distance between Node
A and Node B. The numerical result affirms that Rician fading
By substituting (10), (24) in (25) and expressing
 erfc(.)
 channel has better performance than Rayleigh fading channel,
1 2,0 1 which is more affected by multipath propagation. The figure
as Meijer’s G–function, i.e., erfc (x) = π G1,2 x|
√
also shows that, when the distance of FSO link is large enough,
0, 1/2
[[19], Eq. (06.27.26.0006.01)], the extract-form BER of FSO the impact of atmospheric turbulence on FSO link is dominant
link is given by (26). Next, using [[20], Eq. (21)] and [[21], compared to the impact of RF fading caused by RF link.
Eq. (9.31.1)] the closed-form of BER can be expressed as (27). As a result, there is no difference in terms of performance
between Rician fading channel and Rayleigh fading channel.
To compute the average BER of RF link with BPSK mod- The similar phenomenon also happens for the case of strong
ulation, the integrand consisting of the product of a Gaussian turbulence, i.e., Cn2 = 5 × 10−14 .
Q-function and a fading PDF need to be integrated as follows
It can be seen from Fig. 3 that the location of the relay
∞ p  node, which is represented by the ratio between LRF /LF SO ,
PRF (e) = ∫ Q 2γB fγB (γB )dγB . (28) also affects on the performance of dual-hop mixed FSO/RF
0
systems. When the total distance is shorter than a specific
value, the impairment is dominated by RF fading and attenua-
Using the PDF for the Rician distribution with Laplace tion, thus LRF /LF SO ratio with small value should be chosen.
transform, the average BER of RF link can be computed as On the contrary, high value of LRF /LF SO ratio should be
[18] selected since it helps to shorten the FSO link hence the impact
of atmospheric turbulence is reduced.
1 π/2 (1 + K)sin2 θ
PRF −Rice (e)= ∫ In order to quantify the effect of the location of the relay
π 0 (1 + K)sin2 θ + γ̄B
  node on the system performance, in Fig. 4, BER is investigated
K γ̄B versus the LRF /LF SO ratio when the total transmission length
×exp − dθ. (29)
(1+K)sin2 θ+γ̄B (L) is fixed to 2000 m. In this figure, LRF /LF SO = 1 means
that the relay node is placed at the midpoint of the link A–
B. The best location of relay node is corresponding to the
In addition, the average BER of RF link for the case of optimum value of LRF /LF SO at which BER reaches the
Rayleigh fading channel can be evaluated as [18] minimum value. It is clearly that, when the impairment induced
 r  by RF link is dominant compared to that caused by FSO link,
1 γ̄B
PRF −Rayleigh (e) = 1− . (30) the best location of the relay node should be near Node B
2 1 + γ̄B (i.e., LRF /LF SO < 1). When turbulence strength increases,

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
73
 2016 International Conference on Advanced Technologies for Communications (ATC)

αβγp2 γp2 Pt2 2

   
∞ αβ 1 1
PF SO (e) = ∫ G3,0 h| 2 √ G2,0 2 h | 0, 1/2 dh. (26)
A0 hl Γ (α) Γ (β) 0 1,3 A0 hl γp − 1, α − 1, β − 1 2 π 1,2 σA
2
!
2α+β−3 γp2 αβ 16Pt2 A20 h2l − γp −2 , −α+1 , −α+2 , −β+1 , −β+2 , 1
PF SO (e) = √ G2,5
6,3 2 α2 β 2 |
2 2 2 2 2 . (27)
π 3 A0 hl Γ (α) Γ (β) σA 0, 1/2, −γp2 /2

10
−2 Gamma/Rice and Gamma-Gamma/Rayleigh fading. We de-
rived the closed-form expressions for end-to-end outage prob-
−3 LRF/LFSO = 1/4 ability and BER considering the impact of pointing error
10
of FSO link and receiver noise. The numerical results show
acceptable values of outage probability and BER, which prove
10
−4
the feasibility of using network-coded two-way relaying in
Outage Probability

dual-hop mixed FSO/RF systems. Based on the numerical

−5 results, it is also possible to determine the best location of
10
the relay, at which the proposed system achieves the minimum
LRF/LFSO = 1/2
BER. The best location of the relay depends on the strength
−6
10 of impairments cased by both FSO and RF link.

−7
10 ACKNOWLEDGMENT
Gamma−Gamma/Rice channel

−8
Gamma−Gamma/Rayleigh channel This research is funded by Vietnam National Foundation
10 for Science and Technology Development (NAFOSTED) under
1000 1500 2000 2500 3000
Total distance (m) the grant number 102.02-2015.06.

Fig. 3. Outage probability versus the total distance with PA = PB = 10 R EFERENCES

dBm, Cn 2 = 10−14 , and γ
th = 0 dB.
[1] D. Kedar and S. Arnon, “Urban optical wireless communication net-
works: The main challenges and possible solutions,” IEEE Commun.
−2
10 Mag., vol. 42, no. 5, pp. S2–S7, Jan. 2004.
Node B Relay Node A [2] C. Dehos, J. L. Gonzlez, A. D. Domenico, D. Ktnas and L. Dussopt,
“Millimeter-wave access and backhauling: the solution to the exponential
10
−3 data traffic increase in 5G mobile communications systems?,” IEEE
Commun. Mag., vol. 52, no. 9, pp. 88–95, Sept. 2014.
[3] A.T. Pham, P.V. Trinh, V.V. Mai, N.T. Dang, and C.T. Truong, “Hybrid
−4 free-Space Optics/millimeter-Wave architecture for 5G cellular backhaul
10
networks,” Proc. the 20th OptoElectronics and Communications Confer-
BER

ence (OECC 2015), Shanghai, China, June–July, 2015.

−5
[4] B. He and R. Schober, “Bit-interleaved coded modulation for hybrid
10 RF/FSO systems,” IEEE Trans. Commun, vol.57, no.12, pp.3753–3763,
Dec. 2009.
[5] H. Moradi, M. Falahpour, H. H. Refai, P. G. LoPresti and M. Atiquzza-
−6
10 man, “On the capacity of hybrid FSO/RF links,” IEEE Global Telecom-
C2 = 2×10−14
n
munications Conference (GLOBECOM 2010), Miami, FL, 2010, pp. 1-5.
2 −14
Cn = 1×10 [6] T. Rakia, H. C. Yang, M. S. Alouini and F. Gebali, “Outage analysis
−7 of practical FSO/RF hybrid system with adaptive combining,” IEEE
10
−1 0 1
10 10 10 Communications Letters, vol. 19, no. 8, pp. 1366–1369, Aug. 2015.
L /L ratio [7] E. Lee, J. Park, D. Han and G. Yoon, “Performance analysis of the
RF FSO
asymmetric dual-Hop relay transmission with mixed RF/FSO links,”
IEEE Photonics Technology Letters, vol. 23, no. 21, pp. 1642–1644,
Fig. 4. BER versus the LRF /LF SO ratio for the case of Gamma- Nov 2011.
Gamma/Rayleigh fading with PA = PB = 10 dBm and L = 2000 m. [8] H. Samimi and M. Uysal, “End-to-end performance of mixed RF/FSO
transmission systems,” IEEE/OSA J. of Optical Communications and
Networking, vol. 5, no. 11, pp. 1139–1144, Nov. 2013.
e.g. Cn2 increases from 10−14 to 2 × 10−14 , FSO link should [9] I. S. Ansari, M. M. Abdallah, M. S. Alouini and K. A. Qaraqe, “A
be shortened thus the best location of the relay node should performance study of two hop transmission in mixed underlay RF and
be shifted to Node A. FSO fading channels,” IEEE Wireless Communications and Networking
Conference (WCNC), Istanbul, 2014, pp. 388–393.
[10] N. I. Miridakis, M. Matthaiou and G. K. Karagiannidis, “Multiuser
VI. C ONCLUSION relaying over mixed RF/FSO links,” IEEE Trans. on Commun., vol. 62,
no. 5, pp. 1634–1645, May 2014.
In this paper, we have proposed to a dual-hop mixed
[11] S. Anees and M. R. Bhatnagar, “Performance of an amplify-and-forward
FSO/RF systems using network-coded two-way relaying tech- dual-hop asymmetric RF/FSO communication system,” IEEE/OSA Jour-
nique and analyzed the performance of the proposed sys- nal of Optical Communications and Networking, vol. 7, no. 2, pp. 124–
tems. The dual-hop mixed channel is modeled by Gamma- 135, Feb. 2015.

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
74
 2016 International Conference on Advanced Technologies for Communications (ATC)

[12] P. V. Trinh and A. T. Pham, “Outage performance of dual-Hop AF [17] H. T. T. Pham, N. T. Dang, and A. T. Pham, “Effects of Atmospheric
relaying systems with mixed MMW RF and FSO links,” IEEE Vehicular Turbulence and Misalignment Fading on Performance of Serial Relaying
Technology Conference (VTC Fall), Boston, MA, 2015, pp. 1–5. M-ary PPM FSO Systems with Partially Coherent Gaussian Coherent
Beam,” IET Communications, vol. 8, issue 10, pp. 1762–1768, July 2014.
[13] E. Soleimani-Nasab and M. Uysal, “Generalized performance analysis
of mixed RF/FSO cooperative systems,” IEEE Trans. on Wireless Com- [18] Marvin K. Simon and Mohamed-Slim Alouini, “Digital communication
munications, vol. 15, no. 1, pp. 714–727, Jan. 2016. over fading channels”, John Wiley & Sons, Inc. 2000.
[19] Wolfram: “The Wolfram functions site”, 2002. [Online] Available
[14] S. C. Liew, L. Lu, and S. Zhang, “A primer on physical-layer network at:http://functions.wolfram.com.
coding”, 2015.
[20] V. S. Adamchik and O. I. Marichev, “The algorithm for calculating
[15] I. Djordjevic, W. Ryan, B. Vasic, “Coding for optical channels”, integrals of hypergeometric type functions and its realization in REDUCE
(Springer, 2010). system”, Proc. International Conference on Symbolic and Algebraic
Computation, 1990, pp. 212–224.
[16] A. A. Farid, S. Hranilovic, “Outage capacity optimization for free-space
optical links with pointing errors”, IEEE J. Lightw. Technol, vol. 25, no. [21] Gradshteyn, I.S., Ryzhik, I.M.: “Table of integrals, series, and prod-
7, pp. 1702–1710, 2007. ucts”, Academic, New York, 2007.

Authorized licensed use limited to: Aizu University. Downloaded on October 03,2024 at 07:09:31 UTC from IEEE Xplore. Restrictions apply.
75