2016 International Conference on Signal Processing and Communications (SPCOM)

Outage Analysis in Cooperative Cognitive Radio

Networks with Simultaneous Wireless Information
and Power Transfer
Sanjay Kumar Yadav∗ , Tamaghna Acharaya†
Department of Electronics and Telecommunication Engineering, IIEST Shibpur
Email: [email protected]∗ , [email protected]†

Abstract—Cognitive radio networks with the provision of RF energy harvesting could liberate SU nodes from the depen-
energy harvesting at secondary user (SU) nodes are expected dence on traditional battery power. In addition, if the PU node
to overcome the traditional limitations of wireless networks, i.e. is green-powered, the green benefit could also be extended
bandwidth and energy. In this paper, we investigate a cooperative
cognitive radio network where the primary user (PU) network to the SU node through RF energy harvesting. In [4], the
consists of a primary transmitter-receiver (PT-PR) pair and the authors investigate throughput performance of SUs in CRN,
SU network consists of a secondary transmitter-receiver (ST-SR) where SU nodes can harvest RF energy only from ongoing
pair. In absence of a direct communication link between PT and nearby PU transmission. In order to leverage the benefits of
PR, the ST is used for relaying PT’s signal to PR. In return of its cooperative communication in cognitive radio networks, the
relaying service to PT, ST is allowed to access the PU spectrum
to enable its own transmission to SR. Besides, the PT is assumed idea of cognitive relay is proposed in [5]. A cognitive relay
to be powered by some green energy source, while the ST is node enjoys access the PU spectrum if agrees to assist the
considered to be driven by energy harvested from the reception of communication between a PU source and destination pair.
PT’s signal. Assuming a time switching-based relaying protocols To improve the spectrum utilization efficiency and resolve
for RF energy harvesting at ST, closed form expressions for PU the problem of energy limitations of nodes in conventional
and SU outage probabilities are obtained. Moreover, simulation
results are presented to validate our analytical results and wireless networks, energy harvesting based cognitive relay
illustrate the impacts for various system parameters on the PU networks have been considered in [6].
and SU outage performances. Two types of energy harvesting relaying protocols have
been reported in [7] named as, time switching-based relaying
Index Terms—Cooperative spectrum sharing, RF energy har-
vesting, amplify-and-forward relay, outage probability. protocol (TSR) where, relay spends predefined time for energy
harvesting and remaining time for information processing
I. I NTRODUCTION whereas, in power splitting-based relaying protocol (PSR) re-
lay uses a part of the received power for energy harvesting and
With the continuous proliferation of smart phone based ap- remaining used for information transmission. In [6], outage
plications and burgeoning wireless subscriber base, judicious analyses for both PU and SU have been carried out considering
utilization of energy and spectral resources have become pri- PSR. However, TSR is reported to show superior performance
mary concerns for wireless network designers across the world than PSR in low signal-to-noise ratio (SNR) regime, high
for sustainable development. Scarcity of these resources acts as transmission rate or low energy harvesting efficiency [7]. To
a barrier for the development of future wireless systems. Cog- the best of our knowledge, performance of a system which
nitive radio is a promising technique to utilize the spectrum supports both RF energy harvesting and cooperative spectrum
efficiently by allowing spectrum sharing between primary user sharing with TSR is not well explored in open literature.
(PU) and secondary user (SU) [1]. Currently, smart cities are In this paper, we study a TSR based SWIPT protocol in a
being developed across India. For the development, a large cooperative cognitive radio network. In the proposed network
number of sensor nodes are required to be deployed. These architecture, an energy harvesting cognitive relay node uses
sensor nodes are expected to help in real-time monitoring of its spectrum access opportunity for forwarding PU’s signal to
public infrastructures like stations, markets, airports, bridges. PU destination node as well as transmitting its own signal
Cognitive radio based wireless sensor networks will be very to SU destination simultaneously. In view of the importance
useful for such applications, especially when the application of characterizing the QoS performance of the PU and SU
has stringent QoS requirement [2]. communications in such systems, closed form expressions of
Recently, the idea of simultaneous wireless information and outage probabilities for both primary and secondary system
power transfer (SWIPT), where a device can harvest energy are obtained through systematic analysis. This helps us to
and process information simultaneously from the received RF explore the impacts of various system parameters e.g., power
signals [3], has spurred great interest both in industry and allocating factor (λ), energy harvesting efficiency (η), energy
academia. Cognitive radio networks with the ability of RF harvesting time factor (α) on the system performance.
978-1-5090-1746-1/16/$31.00
c 2016 IEEE The rest of the paper is organized as follows : Section II

978-1-5090-1746-1/16/$31.00 ©2016 IEEE

 2016 International Conference on Signal Processing and Communications (SPCOM)

presents the related work. In section III the system model from the received PU signal for a time duration of αT.
and protocol description are presented. Outage probability for Here T is the duration of total time frame in which entire
both PU and SU communications are discussed in section IV. signal transmission is done from PT to PR and ST to SR
In section V the simulation results are shown and finally, in respectively. α is a fraction of the time frame in which ST
section VI conclusions are drawn. harvests energy from the received PU signal. In phase 2, PT
II. S YSTEM M ODEL AND P ROTOCOL D ESCRIPTION transmits information Xpt to both ST and SR for a duration
of (1 − α)T /2. Finally, in phase 3, ST uses harvested energy
A. System Model to broadcast both Xpt and as well as its own information Xst
We consider a cognitive radio network with a PU in remaining duration of (1 − α)T /2. SR considers Xpt as
transmitter-receiver pair PT-PR and a SU transmitter-receiver an interference in phase 3. As it has prior knowledge of Xpt
pair ST-SR as shown in Fig.1.(a). Here the primary transmitter from phase 2, SR cancels Xpt and extracts only the required
PT intends to communicate with PR, but can not establish di- signal i.e. Xst .
rect communication as the latter is not within the transmission PT uses constant transmission power i.e. Pt . Each of PT,
range of the former. Therefore, PT looks for a neighbour node ST, PR and SR is equipped with a single antenna, where ST is
which can help it to achieve its target rate of communication following amplify-and-forward relaying protocol. The channel
Rpt . We assume that ST agrees to relay PT’s signal to PR coefficients of links PT-ST, PT-SR, ST-SR and ST-PR are given
and at the same time transmits its own message to SR. Here, as g1 , g2 , g3 , g4 respectively. Here all channels are considered
the secondary user pair does not have any dedicated channel as quasi-static Rayleigh fading channel with gi ∼ CN (0, θi2 ), i
between them for communication. Further, ST is not battery = 1,2,3,4 which shows zero mean and θi2 is variance, i=1,2,3,4.
powered, although PT and SR are assumed to be driven by Xpt and Xst are assumed to independent of each other. Both
independent green energy sources. ST harvests energy from of them have zero mean and unit variance; i.e. E{ | Xpt 2
| }=1,
the received PU signal and ST uses this harvested energy for 2
E{ | Xst | }=1.
forwarding PT’s signal and transmission of its own message
signal simultaneously. B. Energy Harvesting and Spectrum Sharing Protocol
As described in the system model, in phase 1 the primary
user PT sends its signal to the secondary transmitter ST. Thus
signal received at the ST node can be stated as
s
Pt
Yst = g1 Xpt + n1 (1)
du1

where d1 is the distance between PT and ST and Pt is the

transmitted power from the source PT, u is the path loss
exponent. Therefore using (1), energy harvested at the ST node
is given by
ηPt |g1 |2 αT
EhST = (2)
Fig. 1. a) System Model b) Frame structure for simultaneous wireless du1
information and power transfer
where 0 < η < 1 is the energy conversion efficiency.
In the next phase, signal received at the information receiver
TABLE I circuit of ST and SR are expressed as
KEY PARAMETERS AND ABBREVIATIONS USED s s
0 Pt Pt
Symbols Definitions Yst = g1 Xpt + n1 + nc1 = g1 Xpt + nr1 (3)
η Energy conversion efficiency du1 du1
α Fraction of the time slot in which ST-SR pair
harvest energy from PT and
Pt , Ph Transmitted power from PT and the relay ST s s
λ Fraction of the ST transmitted power i.e. Ph , which is 0 Pt Pt
used to transmit PT information to PR Ysr = g2 Xpt + n2 + nc2 = g2 Xpt + nr2 (4)
β Power normalization factor du2 du2
EhST Energy harvested at ST node
u Path loss exponent respectively, Here n1 ∼ CN (0, σ12 ) and n2 ∼ CN (0, σ22 ) are
Xpt , Xst
p
Primary and secondary user information signal the narrowband Gaussian noise introduced in the receiving
Pout s
, Pout Outage probability for the primary antenna of ST and SR respectively, whereas nc1 ∼ CN (0,
system and the secondary system respectively
K1 (.) Modified Bessel function of second kind of order 1 σc21 ) and nc2 ∼ CN (0, σc22 ) are the additive white Gaussian
noise (AWGN) introduced due to pass band to baseband signal
The entire communication process is divided into three conversion at ST and SR node [6]. Here nr1 ≈ n1 +nc1 and
phases as shown in Fig.1.(b); In phase 1, ST harvest energy nr2 ≈ n2 +nc2 are the overall AWGNs at the ST and SR.
 2016 International Conference on Signal Processing and Communications (SPCOM)

Finally, in phase 3, ST, which is acting as a relay for PT, where n4 ∼ CN (0, σ42 ) is the AWGN at SR. As mentioned
0
forwards the received information Yst together with its own earlier, SR decodes Xpt successfully in phase 2. Therefore,
00
information Xst . For this, ST splits its total transmitted power SR cancels interference component present in Ysr (10). Thus
Ph in a ratio of λ : (1 − λ). λPh is used to forward the PT’s after interference cancellation, signal available at SR is given
information to PR and remaining (1−λ)Ph is used to transmit by
its own information to SR. Thus the signal transmitted by ST s s
is given by 000 (1 − λ)Ph λPh g3
Ysr = .g3 .Xst + . .nr2 + n4 (11)
0 du4 du4 β
p Y p
XST = λPh . st + (1 − λ)Ph .Xst (5)
β Hence, resultant SNR at SR is given by
β is the power normalization factor which is used to normalize 2ηPt α(1−λ) 2 2
(1−α)σ02 du u |g1 | |g3 |
transmitted signal power of PT to λPh . So β is given by 1 d4
γsr = 2ηαλ
(12)
2
du (1−α) |g3 | + 1
s s
Pt 2 2
Pt 4
β= |g1 | + σr1 ≈ |g1 |2 (6)
u
d1 du1 where σr22
= σ42
= σ02 .
Thus the data rates achieved at PT →
PR with ST as a relay and ST → SR link are given by
where equation (6) is approximated for high average SNR and
nr1 ∼ CN (0, σr21 ) and σr21 ∼
= σ12 + σc21 . Ph is used for time (1 − α) (1 − α)
Rpr = log2 (1 + γpr ), Rsr = log2 (1 + γsr )
interval (1−α)T /2 and is related to the EhST as shown below. 2 2
(13)
EhST 2ηPt |g12 |α
Ph = = (7) III. O UTAGE P ROBABILITY
(1−α)T du1 (1 − α)
2
Therefore, the signal received at the primary receiver PR A. Outage Probability Analysis of Primary System
can be expressed as An outage event for the primary system occurs when
g4 achieved data rate Rpr is less than the target rate Rpt . Thus
Ypr = p u .XST + n3 the outage probability for the primary system is given by
d
s 3 s p
λPt Ph g1 g4 (1 − λ)Ph Pout (Rpr < Rpt ) = P(γpr < γpt ) (14)
= u u . .Xpt + .g4 .Xst
d1 d3 β du3 2Rpt

| {z } | {z } where γpt =2 − 1, is the target SINR corresponding to

(1−α)

s
required signal interference Rpt for PT → PR link. Plugging the equation (9) in (14), we
λPh g4 have
+ . .nr1 + n3 (8) !
du3 β p a|g1 |2 |g4 |2
Pout = P < γpt (15)
b|g1 |2 |g4 |2 + c|g4 |2 + 1
| {z }
overall noise

where n3 ∼ CN (0, σ32 ) is the AWGN at PR. PR tries to decode 2ηPt αλ

where, a = (1−α)σ 2ηPt α(1−λ) 2ηαλ
2 u u , b = (1−α)σ 2 du du and c = du (1−α) .
the required information i.e. Xpt from the received signal 0 d1 d3 0 1 3 3
Let X1 = |g1 |2 and X2 = |g4 |2 be the two random variables,
and consider Xst as an interference signal. For simplicity in
which are exponentially distributed having mean µ1 and µ2
analysis we consider σr21 = σ32 = σ02 . Hence the signal-to-
respectively. Therefore equation (15) can be rewritten as
interference-plus-noise ratio (SINR) i.e. γpr at PR is given !
by p aX1 X2
2ηPt αλ 2 2
Pout = P < γpt (16)
(1−α)σ02 du u |g1 | |g4 | bX1 X2 + cX2 + 1
1 d3
γpr = 2ηPt α(1−λ) 2ηαλ
(9)
2 2 2 p
(1−α)σ02 du u |g1 | |g4 | + du (1−α) |g4 | +1 Here Pout can be conditionally expressed as
1 d3 3

Similarly, signal received at secondary receiver SR is given (i) if γpt ≥ ab

by γpt !
p bγpt −a
00 g3 Pout = P X2 > − cγpt =1 (17)
Ysr = p u .XST + n4 X1 + bγpt −a
d
s 4 s Equation (17) is mainly because of probability of exponential
(1 − λ)Ph λPt Ph g1 g3
= .g3 .Xst + . .Xpt distribution greater than negative number is always 1.
du4 du1 du4 β (ii) if γpt < ab
| {z } | {z }
required signal interference

γpt
!
s  a−bγpt cγpt
P X2 < X − cγpt , X1 > a−bγ .

λPh g3

1 pt
+ . .nr2 + n4 (10)
 a−bγpt
p
du4 β Pout = γpt
! (18)
 a−bγpt cγpt
P X2 > X1 − a−bγ = 1, X1 < a−bγpt .
| {z } 
cγpt


overall noise pt
 2016 International Conference on Signal Processing and Communications (SPCOM)

2ηPt α(1−λ) 2ηαλ 2

Here secondequality of equation (18) is true as the term where, A = (1−α)σ 2 u u and B = du (1−α) . Let X3 = |g3 |
γpt 0 d1 d4 4

a−bγpt
cγpt
cγpt
becomes negative for X1 < a−bγ and proba- be the exponentially distributed random variable having mean
X − s
1 a−bγpt
pt
µ3 . Thus Pout can be stated as
bility of exponential distribution greater than negative number !
p
is 1. Thus, following (18) Pout is expressed by s 1
Pout = P X3 < (24)
X1 γAst − B
cγpt  
a−bypt
Z γpt ! Here the term X1 γAst − B can be positive or negative.
p a−bγpt s
Pout = P X2 > cγpt fX1 (z)dz Therefore Pout is given by
z − a−bγ pt
0
Z∞
  
γpt !
1 γst B
P X3 < , X1 > A .

a−bγpt 
+ P X2 < cγpt fX1 (z)dz (19) s
X1 γA −B
z − a−bγ pt
Pout =  st  (25)
cγpt 1 γst B
P X3 > = 1, X1 < A .


a−bypt
X1 γA −B
st

Plugging the equation (18) in (19), It is mentioned earlier for the primary system that the proba-
cγpt bility of exponential distribution greater than negative number
a−bypt
Z is always 1. Here the second equality in equation (25) follows
p
Pout = fX1 (z)dz the same distribution as mentioned in (18). Due to space
0 constraint, we skip the intermediate steps which is similar
Z∞ γpt ! to the primary outage probability calculation. Therefore SU
a−bγpt
+ P X2 < cγpt fX1 (z)dz (20) outage is given by
z − a−bγ pt
r r 
cγpt
s
Bγ
− Aµst 4γst 4γst
a−bypt
Pout =1 − e 1 K1 (26)
z
Aµ1 µ3 Aµ1 µ3
−
Here, FX2 (z) ∼= P (X2 < z) ∼ = (1 − e µ2 ) is the cumulative where K1 (.) represents the modified Bessel function of second
distribution function (CDF) of exponential random variable kind of order 1.
1 − µz
|g4 |2 and fX1 (z) = e 1 is the probability density
µ1 IV. S IMULATION R ESULTS AND D ISCUSSIONS
function (PDF) of exponential random variable |g1 |2 and z In this section, the influences of various system parameters
is the integration variable. Due to space constraint, we skip i.e. energy conversion efficiency (η), energy harvesting time
the intermediate steps and write down the final expression of factor (α), power splitting factor (λ) and relay distance (d) on
PU outage as follows. primary and secondary outage probabilities are investigated
cγ
s through simulation study. Following [9], we also assume that
p − (a−bγpt )µ 4γpt
Pout =1 − e pt 1 PT, ST and SR are in the same line as shown in Fig. 4.
(a − bγpt )µ1 µ2 Distance between PT and PR is assumed to be 150 meters
s !
4γpt and other distances PT-ST, ST-PR, PT-SR and ST-SR are d,
× K1 (21) (150-d), d/2 and d/2 respectively. Transmitted power from the
(a − bγpt )µ1 µ2
source Pt is 30 dBm and noise variance at all receivers i.e.
where K1 (.) represents the modified Bessel function of second σ02 is assumed to be −110 dBm. Target rate for the primary
kind of order 1, which is defined in [8]. and secondary system i.e. Rpt = Rst is 1, whereas path loss
exponent u is 3.
B. Outage Probability Analysis of Secondary System Fig. 2 illustrates the impact of d on primary and secondary
outage probabilities for the four different set of η and α values
It is assumed that the target rate for ST → SR link is Rst .
i.e. η = 0.5, 1 and α = 0.35, 0.70 for a given λ = 0.8.
An outage event is defined for the secondary system when
We notice in Fig.2.(a) that initially primary outage probability
achieved data rate i.e. Rsr for ST → SR link is less than
increases when d increases and as d crosses 75 meter primary
the target rate i.e. Rst . Therefore outage probability for the
outage probability starts decreasing. The reason behind that
secondary system is given by
can be explained as follows: initially when d increases distance
s
Pout = P(Rsr < Rst ) = P(γsr < γst ) (22) between PT and ST increases and as a result less energy is
harvested (2). Therefore initially primary outage probability
2Rst
where γst = 2 (1−α) − 1, is the target SNR corresponding to increases. Further when d >75m, link quality between ST
Rst for ST → SR link. Substituting the equation (12) in (22), and PR becomes better which leads to decrease the primary
we have outage probability. Moreover, primary outage probability also
decreases when both η and α increases. This is just because of
A|g1 |2 |g3 |2
 
s
Pout =P < γst (23) increasing η and α result in availability of more transmission
B|g3 |2 + 1 power at ST.
 2016 International Conference on Signal Processing and Communications (SPCOM)

Fig.2.(b)
Fig.2.(a)

Fig. 2. Outage probability versus distance, where Fig.2.(a) Primary outage probability versus distance, Fig.2.(b) Secondary outage probability versus distance.

increasing λ, primary outage probability decreases and sec-

ondary outage probability increases. This is just because of the
fact that when λ increases transmitting power for the primary
user information also increases whereas transmitting power for
the secondary user information decreases. Furthermore it is
interesting to see that a proper choice of λ helps the proposed
system to operate with a acceptable values of both primary
and secondary outage probability.
V. C ONCLUSION
In this paper, we have considered a framework for joint
spectrum sharing and RF energy harvesting in a cooperative
cognitive radio network. We have analysed the performance of
Fig. 3. Outage probability versus λ for (η = 1, α = 0.35 ) the system for TSR based SWIPT protocol. Closed form out-
age expression are derived for PU and SU. Further simulation
results are presented to validate our analytical results.
R EFERENCES
[1] S. Haykin, “Cognitive radio: brain-empowered wireless communications,”
IEEE J. Sel. Areas Commun., vol. 23, no. 2, pp. 201–220, Feb. 2005.
[2] G. P. Joshi, S. Y. Nam, and S. W. Kim, “Cognitive radio wireless sensor
networks: applications, challenges and research trends,” Sensors, vol. 13,
Fig. 4. Simulation Model no. 9, pp. 11196–11228, 2013.
[3] L. R. Varshney, “Transporting information and energy simultaneously,”
in Proc. IEEE ISIT, July 2008.
[4] S. Lee, R. Zhang, and K. Huang, “Opportunistic wireless energy harvest-
On the other hand Fig.2.(b) presents secondary outage prob- ing in cognitive radio networks,” IEEE Trans. Wireless Commun., vol. 12,
no. 9, pp. 4788–4799, Sept. 2013.
ability versus distance (d). We notice that secondary outage [5] J. Jia, J. Zhang, and Q. Zhang, “Cooperative relay for cognitive radio
probability increases as d increases. Here as the distance networks,” in Proc. INFOCOM 2009, pp. 2304–2312, Apr. 2009.
increases, relay ST moves further away from PT and SR, [6] Z. Wang, Z. Chen, L. Luo, Z. Hu, B. Xia, and H. Liu, “Outage analysis of
cognitive relay networks with energy harvesting and information transfer,”
resulting in degradation of link PT-ST and ST-SR. Therefore in Proc. ICC, pp. 4348–4353, IEEE, 2014.
secondary outage probability increases. Further, performance [7] A. Nasir, X. Zhou, S. Durrani, R. Kennedy, et al., “Relaying protocols
for α = 0.70 and η = 1 is better compared to α = 0.35 and η for wireless energy harvesting and information processing,” IEEE Trans.
Wireless Commun., vol. 12, no. 7, pp. 3622–3636, 2013.
= 0.5. It is mainly because larger α and η causes more energy [8] I. Gradshteyn and I. Ryzhik, “Table of integrals, series, and Products,7th
harvesting, which creates more transmission power at ST. ed.,” 2007.
In Fig. 3, the impact of power allocation factor (λ) on [9] Y. Han, A. Pandharipande, and S. H. Ting, “Cooperative decode-and-
forward relaying for secondary spectrum access,” IEEE Trans. Wireless
both primary and secondary outage probability is shown. As Commun., vol. 8, no. 10, pp. 4945–4950, 2009.
mention in Section II, λ is the fraction of the transmitting
power which is used to transmit the primary user information
whereas remaining (1 − λ) fraction is used to transmit the
secondary user information. Therefore, we notice that with