Impact of Secondary User Interference

on Primary Network in Cognitive Radio Systems

Amit Kachroo and Sabit Ekin
School of Electrical and Computer Engineering
Oklahoma State University, Stillwater, OK, USA
Email: amit.kachroo, sabit.ekin{@okstate.edu}

Abstract—Most of the research in cognitive radio field is • Probability distribution function (PDF) and cumulative
primarily focused on finding and improving secondary user distribution function (CDF) expressions for noise plus
arXiv:1809.00446v1 [eess.SP] 3 Sep 2018

(SU) performance parameters such as bit error rate, outage interference, signal to noise and interference ratio (SINR)
probability and capacity etc. Less attention is being paid towards
the other side of the network that is the primary network which are derived for both cases of interference from a single
is under interference from SU. Also, it is the primary user SU and multiple SUs on PU network.
(PU) that decides upon the interference temperature constraint • Closed form mean SINR expression, mean capacity and
for power adaptation to maintain a certain level of quality of outage probability expressions are derived.
service while providing access to SUs. However, given the random • The expression generated above are validated with sim-
nature of wireless communication, interference temperature can
be regulated dynamically to overcome the bottlenecks in entire ulation results to show the accuracy of the theoretical
network performance. In order to do so, we need to analyze expressions.
the primary network carefully. This study tries to fill this gap The rest of the paper is organized as follows. In Section II,
by analytically finding the closed form theoretical expressions
for signal to interference and noise ratio (SINR), mean SINR, system model with underlying assumptions is described in de-
instantaneous capacity, mean capacity and outage probability of tail. Section III presents the extensive theoretical analysis and
PU, while taking peak transmit power adaptation at SU into comparisons with simulation results and finally, conclusions
picture. Furthermore, the expressions generated are validated are given in Section IV.
with the simulation results and it is found that our theoretical
derivations are in perfect accord with the simulation outcomes.
II. S YSTEM M ODEL
Index Terms—Cognitive Radio Network, Interference Temper-
ature, Mean Capacity, SINR, Outage Probability The cognitive radio network that is under consideration in
this work is shown in Fig. 1. It consists of n SUs (SU network)
and k PUs with corresponding SBS and PBS. Since we are
I. I NTRODUCTION analyzing the effect of SU on PU network, we don’t need to
consider the channel gain between PU and SBS, and also we
In cognitive radio network, a secondary user (SU) is allowed don’t need to consider the interference among PUs because of
to access the primary user (PU) spectrum completely if the orthogonal resource allocation between them.
available spectrum is not used by PU (interweave) or con-
currently (underlay) with PU. The concurrent transmission is
allowed if and only if the SU maintains a certain power thresh-
old constraint known as interference temperature [1], [2]. Most
of the studies [1]–[6] that involve this interference temperature
model utilize either peak or average transmit power adaptation
for the purpose of analyzing or improving the performance of
secondary network. However, the impact of SU interference on
primary network is let off completely. Therefore, to asses the
performance and other quality of service (QoS) parameters,
closed form expression need to be derived and validated. This
study focuses on the mathematical foundation to derive these
necessary performance expressions. This theoretical analysis Fig. 1. Underlay cognitive network with n-SUs sharing the spectrum with
PU network of k-PUs. The channel power gain between any SU-i (ith user)
is done by first considering interference from a single SU and and PBS is denoted by αi , between any SU-i (ith user) and SBS by βi and
then extended to the case of interference from multiple SUs on between ith -PU and PBS by γi .
primary network. As per our knowledge, this is the first paper
to analytically analyze the effect of SU interference on primary In addition, since the channel fading is assumed to be
network. The contribution of this paper can be summarized as Rayleigh distributed, the channel power gains follow an ex-
follows: ponential distribution. Considering the peak power adaptation
[4], [6] for the case of interference from single SU on primary and rate parameter. The PDF and CDF of Gamma distribution
network, the secondary transmit power is given by: is given as
n qo n  
P tx = min p, (1) X λ2

α fγ̄ (x) = αi p = γ̄ x, n, = γ̄ x, n, λ̄ ,
i=1
p
In the following sections, for theoretical analysis pur-
pose, we assume that there are n SUs that form the un- λ̄n xn−1 −λ̄x 
(6)
= e , ∀ x ≥ 0, n > 0, λ̄ > 0 ,
derlay cognitive network with primary user, where n = Γ(n, 0)


{1, 2, 3 . . . , n}. Furthermore, the thermal additive white Gaus- Γ n, λ̄x 
sian noise (AWGN) in the network is assumed to have circu- Fγ̄ (x) = 1 − , ∀ x ≥ 0, n > 0, λ̄ > 0 ,
Γ(n, 0)
larly symmetric complex Gaussian distribution with zero mean
where n is the total number of SUs in the underlay network,
and variance as σ 2 , i.e., CN (0, σ 2 ).
λ̄ = λ2 /p is the scaled rate parameter between SU and PBS
III. T HEORETICAL A NALYSIS and Γ(a, x) is an incomplete gamma function defined as:
Z ∞
In this section, theoretical expressions for PU performance
parameters with interference from a single SU will be derived Γ(a, x) = ta−1 e−t dt, ∀ a > 0, x ≥ 0.
a
first and then extended to the case of interference from Following the same mathematical approach that was used
multiple SUs. The interference observed at primary receiver in single SU case, the distribution of noise plus interference
because of a single SU and multiple SUs with peak power in multiple SUs case is then derived as:
adaptation will be given as:

m Γ n, λ̄(x − σ 2 )
Isingle = αPsec = min{αp, q}, FN I (x) = 1 − + H(x − σ 2 − q)
( n ) Γ(n, 0)
X (2) Γ n, λ̄(x − σ 2 )


Imulti = min αi p, q . × ,
i=1 Γ(n, 0)

(7)
Eq. (2) represents a minimum of a random variable and a m Γ n, λ̄(x − σ 2 ) 2 n
fN I (x) = δ(x − σ − q) + λ̄
constant1 . From the theory of mixed random variables [7]–[9], Γ(n, 0)
a constant c can be modelled as a random variable with PDF (x − σ 2 )n−1 −λ̄(x−σ2 )
equal to δ(x − c) and CDF equal to H(x − c), where H(x) × e [1 − H(x − σ 2 − q)],
Γ(n, 0)
is a Heaviside function and δ(x) is a Dirac Delta function.
So with interference from a single SU, the CDF and PDF of where σ 2 is the CN (0, σ 2 ). Fig. 2 and Fig. 3 plots the CDF
minimum of two independent random variables is then given and PDF for the theoretical expression (Eq. (7) ) with the
by: simulation result of p > q and q < p for different SU densities
of n = 1, 2, 3.
FI (x) = Fαp0 (x) + Fq (x) − Fαp0 (x)Fq (x),
λx (3)
FI (x) = 1 − e− p (1 − H(x − q)).
On including noise CN (0, σ 2 ), the CDF of noise plus
interference will be then,
λ(x−σ 2 )
FIN (x) = 1 − e− p (1 − H(x − σ 2 − q)). (4)
Correspondingly, the PDF of interference and noise is
given by differentiating the CDF with respect to noise and
interference variable x, i.e.
λ λ(x−σ2 ) 
fIN (x) = e− p 1 − H(x − σ 2 − q)
p (5)
p 
+ δ(x − σ 2 − q) , ∀ σ 2 ≤ x ≤ ∞.
λ
For the case of interference from multiple SUs, the distri-
Fig. 2. PDF and CDF of noise and interference for different number of SUs
bution of of interference given in Eq. (2) will follow Gamma (n = 1, 2, 3), when p > q, where p = 4, q = 2 and σ 2 = 1 with support
distribution2 , fγ̄ (x, κ, θ), where κ and θ represent the shape region from σ 2 ≤ x ≤ ∞.

1 For illustration purposes, the value’s of peak power p and interference

temperature q in this paper are chosen to be in linear scale. However, the A. Instantaneous SINR
expressions derived in this paper hold for any value of p and q for any scale. The instantaneous SINR at PBS considering the system
2 The distribution of sum of independent exponential random variables with
the same rate parameters follows Gamma distribution. Also, to distinguish model (Fig. 1) is given by:
between the channel γ between PU and PBS, the Gamma distribution is γp
denoted as γ̄ in this study. SINR = 2 , (8)
σ +I
 1, with AWGN as CN (0, σ 2 = 1), the PDF can be further
simplified to
  
1 −z p z+q(z+1)
fz (z) = e p 1+ + e− p
p(z + 1) z+1
  (10)
p
× (1 + q)z − .
z+1
Following the same analytical framework used for the single
SU case, the distribution with interference from multiple SUs
will be:
Z ∞ n
¯
fzm (z) = y λ¯1 e−λ1 yz fγ̄ (y − σ 2 , n, λ̄) + δ(y − σ 2 − q)
σ2
× (1 − Fγ̄ (y − σ 2 , n, λ̄)) − fγ̄ (y − σ 2 , n, λ̄)
o
Fig. 3. PDF and CDF of noise and interference for different number of SUs × H(y − σ 2 − q) dy.
(n = 1, 2, 3) when p < q, where p = 2 q = 4 and σ 2 = 1 with support
region from σ 2 ≤ x ≤ ∞. which on further evaluation and simplification reduces to
" 
2 ¯
f (z) = λ¯1 λ̄ e
m
z
n −σ λ 1 z
Θ−1−n
n + σ2 Θ
where I is the interference from SUs given by Eq. (2). The   #
distribution of numerator is a scaled exponential distribution Γ(n, qΘ) Γ(n + 1, qΘ) (11)
× 1− −
and the distribution of denominator is already derived in the Γ(n, 0) Γ(n, 0)
previous section (Eq. (5) and Eq. (7)). Therefore, the PDF of
Γ(n, q λ̄) −λ¯1 (σ2 +q)z
ratio of two independent random variables [10] i.e. z = x/y, + λ¯1 (σ 2 + q) e ,
where x = γp and y = σ 2 + I will be given as Γ(n, 0)
where Θ is the scaled and shifted random variable version 4 of
∞ ∞
z given by Θ = λ̄ + λ¯1 z. Fig. 4 shows the plot of the derived
Z Z
fz (z) = y · fx,y (yz, y)dy = y · fx (yz)fy (y)dy
σ2 σ2 theoretical expression with simulation data for the two cases
∀ y ≥ 0. of p < q and p > q with different SU densities (n = 1, 2, 3).
(9)
1.4 2.5
Simulation Simulation
For the interference from a single SU user, the SINR Theoretical Theoretical
1.2 Simulation Simulation
n=2
distribution will be as follows: n=3
Theoretical 2 Theoretical
Simulation Simulation
1 Theoretical Theoretical
−λ1 yz n=3
Z ∞ n=2
λ1 e λ2 − λ2 (y−σ
p 2)
1.5
fz (z) = y· e p , 0.8
fz(z)

p p
fz(z)

2
nσ p o 0.6
× 1 − H(y − σ 2 − q) + δ(y − σ 2 − q) dy n=1 1

Z ∞ λ2 0.4
n=1
Z ∞
λ1 λ2 2 λ σ 2 −y(λ 1 z+λ2 ) 0.5
= 2 e p y·e p dy − y 0.2
p σ2 σ 2 +q
 0
−y(λ1 z+λ2 ) p −(σ 2 +q)(λ1 z+λ2 ) 0
×e p dy + (σ 2 + q)e p . 0.5 1
z
1.5 2 2.5 0.1 0.5 0.9 1.3 1.7
z
2.1

λ2
Fig. 4. PDF of SINR for two cases of p < q and p > q for different number
By using integration by parts and on further simplification, of SUs (n = 1, 2, 3).
the PDF is reduced to:
 In the following sections, we will look into the crucial per-
λ1 λ2 −σ 2 λ1 z
 p formance metrics (outage probability and capacity) of underlay
fz (z) = e p σ2 +
Λp Λ cognitive network. The mentioned approach can be extended
to the case of interference from multiple SUs given that
 2 
σ 2 λ1 z+qΛ (σ + q)λ 1z p
+ e− p − , the important SINR expression Eq. (11) for multiple SUs is
λ2 Λ
already been derived. However, given the space limitations, the
where Λ is the scaled and shifted random variable version 3 of derivations considering multiple SUs are not detailed herein
z given by Λ = λ1 + λ2 z . Under the scenario of λ1 = λ2 = in coming sections. Nonetheless, the fundamental case of
interference from single SU case has been presented in detail.
3 λ is the channel rate parameter between PU and PBS, whereas λ is the 4 Here λ̄ = λ /p is the scaled rate parameter of SUs and λ¯ = λ /p is
1 2 2 1 1
channel rate parameter between SU and PBS. the scaled rate parameter for PU.
B. Mean SINR nothing but the CDF of SINR. Therefore,
R∞ Z ψ
The mean SINR is given as µ = 0 zfz (z)dz where the Fz (ψ) = fz (z)dz,
PDF of SINR fz (z) was derived in Eq. (10). Thus, 0
−z −z
Z ψ ! Z ψ !
ep ep
= dz + dz
∞ n −z  p(z + 1) (z + 1)2
Z 
1 n p 0 0
µ= z e p 1+
0 p(z + 1) z+1 Z ψ z+q(z+1) !
z(1 + q)e−( p )
+ dz
 o
−(
z+q(z+1)
) p p(z + 1)
+e p (1 + q)z − dz, 0
z+1 z+q(z+1) !
e −( p )
Z ∞ −z
! Z ∞ −z
! Z ψ
ze p ze p − dz.
= dz + dz (z + 1)2
0 p(z + 1) 0 (z + 1)2 0

Z ∞ z+q(z+1) !
z 2 (1 + q)e−( p ) which on further integration and simplification reduces to
+ dz −ψ
0 p(z + 1) e p  −q(ψ+1)

z+q(z+1) !
Fz (ψ) = 1 − 1 + ψe p
Z ∞
ze−( p ) ψ+1
− dz.
0 (z + 1)2

1
which on further simplification reduces to p<q

0.8
     −q
1 1 1+q pe p p>q
µ=e Γ 0,
p − Γ 0, + . (12) 0.6
p p 1+q
Fz ( )

0.4
Fig. 5 shows that the change in mean SINR while varying the
0.2
Simulation
4.5 Theoretical
Simulation 0
0 1 2 3 4 5 6 7 8 9 10
4 Theoretical

3.5
Fig. 6. Outage probability of PU for p < q, where p = 2 and q = 4 and for
3
p > q, where p = 4 and q = 2.
p=4
2.5

2
It can be directly inferred from Fig. 6 that if q > p, the
p=2 outage probability is higher than in the case of p > q. In
1.5
addition to this inference, it can be also observed that the
1 theoretical expressions derived are in sync with the simulation
0.5 results, i.e., increase the spectral efficiency of the network.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
q
D. Instantaneous Capacity of Primary Network
Fig. 5. Mean SINR vs Interference Temperature, q, for p = 2 and p = 4. The PDF of instantaneous capacity can be readily found
from the PDF of instantaneous SINR by using transformation
of random variables method [7], [8]. This can be obtained by
interference temperature q for a constant peak transmit power
using:
p. The higher transmit power (for both PU and SU) with lower
 dz 
interference temperature gives better mean SINR than lower
fx (x) = fz (z)  ,

transmit power (for both PU and SU) with high IT constraint.
C. Outage Probability of Primary Network
The outage probability is defined as the probability when
the instantaneous SINR drops below a given threshold. Math-
ematically, this is given as: P r(γ ≤ ψ) = Fz (ψ), which is
dx z=ex −1
where fz (z) is derived in Eq. (10) for the case of interference
from single SU on primary network. It can be seen from
Fig. 7 that there is a point where the instantaneous capacity
for p < q goes below p > q. It proves the point that the
interference temperature should not be kept constant rather
should be dynamic in nature to exploit full potential of the
network.
 1.4 1.4
Simulation Simulation
Theoretical Theoretical
1.2
p>q 1.2
1

Mean Capacity
p=4
1
f x(x)

0.8

0.6
0.8
p<q
p=2
0.4
0.6
0.2

0 0.4
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7
x
Interference Temperature (q)

Fig. 7. Instantaneous PDF of Capacity for p < q, where p = 2 and q = 4 Fig. 8. Theoretical and simulation result plots for capacity at p = 2 and
and for p > q, where p = 4 and q = 2. p = 4 with varying interference temperature: q.

E. Mean Capacity for the PDF and CDF of interference and noise, SINR for
The average capacity from the PDF of instantaneous SINR interference from single and multiple SUs are derived. Further-
(fz (z)) is given as: more, instantaneous capacity with theoretical expressions for
Z ∞ mean SINR, mean capacity and outage probability are deduced
C̄ = log(1 + z)fz (z)dz. for simplistic network consisting of interference from single
0 SU. Finally, the theoretical expressions are validated with the
Substituting Eq. (10) in the above expression and on further simulation results.
evaluation.
1      ACKNOWLEDGMENT
ep z+1 (q + 1)(z + 1) This work was supported by NASA Oklahoma Space Grant
C̄ = Γ 0, − Γ 0,
p p p Consortium (EPSCoR) Research Initiation Grant. Also, the
1−(q+1)(z+1)

e p n q(z+1) authors would like to thank the TPC and reviewers for their
× (p + q + 1) − e p valuable feedback and suggestions.
z+1
o∞

  R EFERENCES
× 1 + log(z + 1) + zlog(z + 1) − 1  .

 [1] S. Ekin, M. M. Abdallah, K. A. Qaraqe, and E. Serpedin, “Random
0 subcarrier allocation in ofdm-based cognitive radio networks,” IEEE
Transactions on Signal Processing, vol. 60, no. 9, pp. 4758–4774, 2012.
[2] L. Musavian and S. Aı̈ssa, “Capacity and power allocation for spectrum-
1−(q+1)(z+1)
p sharing communications in fading channels,” IEEE Transactions on
At z = ∞, Γ(0, z) → 0 and also, e z+1 → 0. Wireless Communications, vol. 8, no. 1, pp. 148–156, 2009.
Therefore, the final mean capacity expression will be evaluated [3] H. Tran, M. A. Hagos, M. Mohamed, and H.-J. Zepernick, “Impact
at z = 0: of primary networks on the performance of secondary networks,” in
" Computing, Management and Telecommunications (ComManTel), 2013
1    # International Conference on. IEEE, 2013, pp. 43–48.
−pq ep q+1 1
C̄ = 1 − e + (p + q + 1)Γ 0, − Γ 0, [4] T. W. Ban, W. Choi, B. C. Jung, and D. K. Sung, “Multi-user diversity
p p p in a spectrum sharing system,” IEEE Transactions on Wireless Commu-
nications, vol. 8, no. 1, pp. 102–106, 2009.
Fig. 8 shows the plot of this theoretical expression with [5] S. Ekin, F. Yilmaz, H. Celebi, K. A. Qaraqe, M.-S. Alouini, and
E. Serpedin, “Capacity limits of spectrum-sharing systems over hyper-
simulation results for two cases of transmit power p = 2 and fading channels,” Wiley Wireless Communications and Mobile Comput-
p = 4. Intuitively, high transmit power p = 4 will result in ing, vol. 12, no. 16, pp. 1471–1480, 2012.
high capacity for the network than the low transmit power of [6] R. Zhang, “On peak versus average interference power constraints for
protecting primary users in cognitive radio networks,” IEEE Transac-
p = 2 but when the interference temperature is relaxed, the tions on Wireless Communications, vol. 8, no. 4, 2009.
interference caused due to secondary user on primary will also [7] A. Papoulis and S. U. Pillai, Probability, random variables, and stochas-
increase that in turn will reduce the overall capacity. tic processes. Tata McGraw-Hill Education, 2002.
[8] S. Miller and D. Childers, Probability and random processes: With
IV. C ONCLUSION applications to signal processing and communications. Academic Press,
2012.
In this paper, the performance of primary network is studied [9] H. Pishro-Nik, Introduction to probability, statistics, and random pro-
cesses. Kappa Research LLC, 2014.
considering interference from the SU network. The analysis [10] J. H. Curtiss, “On the distribution of the quotient of two chance
is done under peak power adaptation method at secondary variables,” The Annals of Mathematical Statistics, no. 4, pp. 409–421,
transmitter. Given the importance of dynamic interference 1941.
temperature for network performance, closed form expressions