Anda di halaman 1dari 4

312 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 1, NO.

4, AUGUST 2012

Achievable Data Rate in Spectrum-Sharing Channels with


Variable-Rate Variable-Power Primary Users
Yuli Yang, Member, IEEE, and Sonia Aissa, Senior Member, IEEE

ST SR
Abstract—In this work, we propose a transmission strategy gs
for secondary users (SUs) within a cognitive radio network
where primary users (PUs) exploit variable-rate variable-power
modulation. By monitoring the PU’s transmissions, the SU
gps gsp
adjusts its transmit power based on the gap between the PU’s
received effective signal-to-noise power ratio (SNR) and the
lower SNR boundary for the modulation mode that is being
used in the primary link. Thus, at the SU’s presence, the PU’s gp
quality of service (QoS) is guaranteed without increasing its PT PR
processing complexity thanks to no interference cancellation
Fig. 1. Spectrum-sharing system model.
required in the PU’s operation. To demonstrate the advantage of
our proposed transmission strategy, we analyze the secondary
user’s achievable data rate by taking into account different the primary link so that the PU’s processing complexity is not
transmission capabilities for the secondary transmitter. The increased while maintaining its QoS.
corresponding numerical results not only prove the validity of
our derivations but also provide a convenient tool for the network The cognitive radio channel capacity has been formulated
design with the proposed transmission strategy. in [5], [6], and [7] by imposing instantaneous or average
interference constraints at the PU. Moreover, the capacity is
Index Terms—Cognitive radio (CR), spectrum sharing, adap-
tive modulation, data rate. analyzed in [8] with imperfect channel information. In this
paper, to illustrate the advantage of the proposed transmission
strategy, we derive the achievable data rate in the secondary
I. I NTRODUCTION
link by taking into account different capabilities of the sec-

T HE concept of cognitive radio has attracted a lot of


interest in both academia [1] and industry [2] as it offers
a promising solution to the problem of spectrum scarcity for
ondary transmitter (ST), namely with or without the peak
transmit power constraint imposed on the ST. Furthermore,
numerical results will substantiate the validity of our deriva-
wireless applications. On the other hand, adaptive modulation tions.
has become a regular setting in current wireless networks to In the following, we detail the design of the proposed
realize robust and spectrally efficient communications [3]. transmission strategy and demonstrate its performance by
As is known, in order for secondary users to increase their presenting the SU’s achievable data rate obtained in closed
transmission opportunities within a cognitive radio network, form as well as in the corresponding numerical results.
any available knowledge on the primary users’ transmissions
has to be fully exploited [4]. Motivated by this philosophy,
we consider a spectrum-sharing channel where variable-rate II. S YSTEM M ODEL
variable-power modulation is adopted in the primary link. Consider a spectrum-sharing system, where a primary
With the adaptive modulation, each modulation mode cor- transmitter-receiver pair uses adaptive rate and power allo-
responds to an interval of the received signal-to-noise power cation to perform communications in between. Meanwhile,
ratio (SNR), given a target bit error rate (BER). According to a secondary transmitter (ST) is trying to access the licensed
the interval where the received SNR at the primary user (PU) spectrum and transmit information to its own receiver by
falls, the corresponding modulation mode is activated. As taking into account the impact of its transmissions on the QoS
such, the gap between the lower boundary of the SNR interval in the primary link.
and the PU’s actually received SNR forms an interference- As shown in Fig. 1, let gp , gps , gs and gsp denote the instan-
tolerable zone and, consequently, this setting may avail the taneous channel power gains from the primary transmitter (PT)
secondary user (SU) more opportunities to access the licensed to the primary receiver (PR), from the PT to the ST, from the
spectrum while maintaining the PU’s quality-of-service (QoS). ST to the secondary receiver (SR), and from the ST to the
Based on said gap, this paper proposes a transmission strategy PR, respectively. All the channel states are assumed to be flat-
for the SU to accommodate its transmit power and achieve fading, following the independent and identically distributed
good performance under the requirements of the PU’s QoS. (i.i.d.) complex Gaussian distribution with zero-mean and unit-
With this strategy, no interference cancellation is involved in variance. Accordingly, all these channel power gains follow
Manuscript received March 18, 2012. The associate editor coordinating the the i.i.d. unit-mean exponential distribution with probability
review of this paper and approving it for publication was S. Sun. density function (PDF)
Y. Yang is with the Division of Physical Sciences and Engineering, King
Abdullah University of Science and Technology (KAUST), Thuwal, KSA (e- fgX (gX ) = e−gX , (1)
mail: yuli.yang@kaust.edu.sa).
S. Aı̈ssa is with the Institut National de la Recherche Scientifique (INRS),
University of Quebec, Montreal, QC, Canada (e-mail: aissa@emt.inrs.ca). where the subscript X refers to the channel power gains of
Digital Object Identifier 10.1109/WCL.2012.042512.120198 the links shown in Fig. 1, i.e., p , ps , s and sp .
2162-2337/12$31.00 
c 2012 IEEE
YANG and AISSA: ACHIEVABLE DATA RATE IN SPECTRUM-SHARING CHANNELS WITH VARIABLE-RATE VARIABLE-POWER PRIMARY USERS 313

Here, the variable-rate variable-power M QAM in [9] is strategy, where no peak transmit power constraint is imposed
adopted by the PU. With a pilot protocol, for a constant pilot on the ST, and the other is with the peak power constraint
power S̄, the PR’s instantaneous received SNR in the absence considered at the ST, which is upper bounded by the first
of a SU can be expressed as case.
It should be noted that the noise plus interference at the SR
γp = gp S̄/Np , (2)
is assumed to be AWGN, as the ST may get the waveforms of
where Np is the additive white Gaussian noise (AWGN) power the PT’s transmissions by overhearing the pilot via the channel
received at the PR. Assuming S̄ = 1, the PDF of γp is given gps , thereby exploiting precoding techniques to alleviate the
by impact of the PT’s transmissions on the SR [10].
fγp (γp ) = Np e−Np γp . (3)
For each symbol to be transmitted in the primary link, an A. ST without Peak Transmit Power Constraint
M QAM constellation from N candidates is selected according As above-mentioned, the modulation mode M0 is activated
to γp . The N candidates are arranged in an ascending order (i.e., no data transmission) in the primary link once a deep
of the data rate and the j th constellation is activated if γp ∈ channel fading emerges with γp ∈ [0, M1 γ ∗ ), where the ST
[Mj γ ∗ , Mj+1 γ ∗ ), j = 0, 1, · · · , N , where Mj is the number may transmit with a power as high as possible within its power
of points in the j th M QAM constellation with M0 = 0 and budget. In the scenario without transmit power constraint at
MN +1 = ∞. The parameter γ ∗ > 0 can be optimized using the ST, the SU’s achievable data rate, with M0 constellation
numerical search to maximize the spectral efficiency. activated by PU, can be expressed in [nats/sec/Hz] as
To avoid deep channel fading in the primary link, choosing  M1 γ ∗  ∞
the M0 constellation corresponds to no data transmission once C0 = fγp (γp ) dγp ln (1 + gs PM /Ns ) fgs (gs ) dgs
γp < M1 γ ∗ . When γp ∈ [Mj γ ∗ , Mj+1 γ ∗ ), j = 1, · · · , N , and 0
 M1 γ ∗
0
 ∞
the j th constellation is activated, the data rate is log2 Mj bits
= Np e−Np γp dγp ln (1 + gs PM /Ns ) e−gs dgs
and the PT’s transmit power Sj is adapted relative to γp , as
0  0
specified by ∗
= e−Np M1 γ − 1 eNs /PM Ei (−Ns /PM ) ,
Sj = (Mj − 1)/(Kγp ), (4)
(8)
where K = −1.5/ ln(5BER) is a parameter pertaining to the
target BER of the adaptive modulation scheme. where PM represents the ST’s transmit power with an arbi-
Thus, with the variable-rate variable-power modulation, trarily high value once there is no data transmission in the
for each M QAM constellation, the PR’s received SNR is a primary link, and Ns denotes the SR’s received AWGN power.
constant, given by The PDFs of γp and gs , fγp (γp ) and fgs (g xs ), are given in (3)
and (1), respectively. Besides, Ei(x) = −∞ (et /t) dt is the
γj = (Mj − 1)/K, j = 1, · · · , N. (5) exponential integral function.
When it comes to the modulation modes other than M0
Since the pilot protocol is used in the primary link, the ST is
activated by the PU, the ST is required to control its transmit
supposed to have the knowledge of the pilot via the broadcast
power based on (7) so as to guarantee the QoS of the primary
channel in the system that it is trying to access. Hence, the
communications is unaffected. Without loss of generality, if
channel state gps may be estimated at the ST by overhearing
the j th constellation is being used in the primary link, j =
the pilot transmitted by the PT through this channel [10]
1, · · · , N , the ST’s transmit power should be specified as a
and, consequently, the ST can obtain the PT’s instantaneous
function of the channel power gain from the ST to the PR,
transmit power Sj as well as deduce the constellation that is
gsp , according to
being used in the primary link.  
In the presence of the ST’s transmission, the PR’s effective Np (Mj − 1)/K Np
Tj = ∗
−1 = μj , (9)
received SNR with each modulation mode is found at gsp Mj γ gsp
γj
γe,j = , j = 1, · · · , N, (6) (Mj − 1)/K
1 + gsp Tj /Np where the parameter μj  − 1 is defined for a
Mj γ ∗
where Tj is the ST’s transmit power as the j th constellation simple expression.
is active in the primary link. Hence, in order to guarantee the Subsequently, based on the ST’s transmit power given
QoS of primary communications, it is obligatory upon the ST by (9), in the case of no peak transmit power constraint
to control its transmit power Tj to satisfy imposed on the ST, the SU’s achievable data rate in the unit
of [nats/sec/Hz] can be calculated using (10) (see at the top
γe,j  Mj γ ∗ (7) of next page), where C0 is given by (8) and the parameter
when the PU is employing the j th
constellation. λj  μj Np /Ns is introduced for simplicity, with μj defined
in (9).
III. S ECONDARY U SER ’ S ACHIEVABLE DATA R ATE
In this section, we derive the achievable data rate in the B. ST with Peak Transmit Power Constraint
secondary link by considering two scenarios with respect to To better protect the PU’s communications, the peak trans-
the ST’s transmission capabilities. The first is a general one mit power of a ST is constrained when the ST accesses to a
to demonstrate the full benefit of the proposed transmission spectrum-sharing channel. Assuming that the peak transmit
314 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 1, NO. 4, AUGUST 2012

N 
 Mj+1 γ ∗  ∞  ∞

gs Tj
Cs = C0 + fγp (γp ) dγp ln 1 + fgs (gs ) fgsp (gsp ) dgs dgsp
j=1 Mj γ
∗ 0 0 Ns
N  Mj+1 γ ∗  ∞ ∞

−Np γp gs Tj
= C0 + Np e dγp ln 1 + e−gs e−gsp dgs dgsp
j=1 M j γ ∗ 0 0 N s
(10)
N   ∞  g

−Np Mj γ ∗ −Np Mj+1 γ ∗ gsp /λj sp


= C0 + e −e −e Ei − e−gsp dgsp
j=1 0 λ j

N 
 
∗ ∗ λj
= C0 + e−Np Mj γ − e−Np Mj+1 γ ln λj
j=1
λj − 1

power constraint imposed on the ST is PC , when the j th 4.5


constellation is selected by the PU (j = 0), the ST’s transmit no Pc, N = −10dBw

SU’s Achievable Data Rate [nats/sec/Hz]


4 p
power will be adapted from (9) to Pc = 10dBw, N = −10dBw
3.5
p
no Pc, N = −20dBw
C Np μj /gsp , gsp > Np μj /PC p
Tj = 3 Pc = 10dBw, N = −20dBw
PC , 0  gsp  Np μj /PC (11) p

2.5
j = 1, · · · , N,
2 Np = −10dBw
where the superscript C is used to represent the case that a
1.5
peak transmit power limit is imposed on the ST. N = −20dBw
p
As a result, if the ST’s peak transmit power is limited by PC , 1
the SU’s achievable data rate with the proposed transmission 0.5
strategy is calculated in [nats/sec/Hz] as shown in (12) on the
0
top of next page, where C0C is the SU’s achievable data rate, −20 −15 −10 −5 0
SR’s Received AWGN N [dBw]
with the ST’s peak transmit power constrained, once there is s

no data transmission in the primary link. Fig. 2. Achievable data rate in the secondary link versus the received AWGN
In detail, C0C can be found as follows: power at the SR, Ns , for the PR’s received AWGN Np = −10dBw and
−20dBw.
 M1 γ ∗  ∞
C −Np γp
C0 = Np e dγp ln (1 + gs PC /Ns ) e−gs dgs
0  0

= e−Np M1 γ − 1 eNs /PC Ei (−Ns /PC ) . imposed on the ST is investigated as well. From this figure,
(13) we observe that the SU’s achievable data rate is reduced with
the increase in Ns and, however, improved with the increase in
IV. N UMERICAL R ESULTS AND D ISCUSSIONS Np . The former phenomenon is intuition-compliant. The main
reason behind the latter is that the impact of the interference
Based on the analysis and derivations presented in the from the ST to the PR’s effective received SNR becomes
above, the performance of the proposed transmission strategy more and more ignorable as the PR’s received AWGN power
is illustrated numerically in this section. Variable-rate variable- increases. Therefore, the ST is allowed to transmit with a
power M QAM [9] is exploited in the primary link with relatively higher power in the case of higher Np , which can be
N = 4 constellations as the candidates to be selected. The perceived from the ST’s transmit power adaptation as shown
number of constellation points for each modulation mode in (9) and (11).
Mj = 0, 2, 4, 16, 64 as the mode order j varies from 0 to
N = 4. The target BER is set to 10−3 . For the pilot power Furthermore, to demonstrate the effect of the ST’s peak
S̄ = 1, the parameter γ ∗ is optimized as 1.22 and 1.685 transmit power constraint, PC , on the performance of the
when the PR’s received AWGN power Np = −10dBw and secondary link, we depict the SU’s achievable data rate versus
−20dBw, respectively. In addition, for the case that no peak PC in Fig. 3 for Np = −10dBw and −20dBw, where the SR’s
transmit power constraint is imposed on the ST, the latter sets received AWGN power Ns = −10dBw and the case with no
its transmit power to PM = 15dBw when constellation M0 is peak transmit power constraint on the ST is also provided
activated in the primary link. for the sake of comparison. As is shown in this figure, the
To begin with, the SU’s achievable data rate is plotted as case with no peak transmit power constraint at the ST is an
a function of the SR’s received AWGN power Ns in Fig. 2. upper bound on that with the peak transmit power constraint
Two examples are considered for the PR’s received AWGN at the ST. As PC is enhanced, the achievable data rate in
power: Np = −10dBw and −20dBw. The peak transmit the secondary link with the peak transmit power constraint
power constraint imposed on the ST is set to PC = 10dBw increases and approaches that without the peak transmit power
and, moreover, the case with no peak transmit power constraint constraint, which substantiates the validity of our derivations.
YANG and AISSA: ACHIEVABLE DATA RATE IN SPECTRUM-SHARING CHANNELS WITH VARIABLE-RATE VARIABLE-POWER PRIMARY USERS 315

N 
 

 Mj+1 γ ∗ ∞ ∞
gs
CsC = C0C + Np e −Np γp
dγp ln 1 + λj e−gs e−gsp dgs dgsp
j=1 Mj γ ∗
Np
PC jμ 0 g sp
 ∞  ∞

g s
+ N e−gsp dgsp ln 1 + PC e−gs dgs
p
PC μ j 0 N s

N 
 ∗ ∗
  ∞   g

sp
= C0C + e−Np Mj γ − e−Np Mj+1 γ −e gsp /λj
Ei − e−gsp dgsp
j=1
Np
PC μj
λ j
(12)
 ∞

−gsp
Ns Ns
+ N e dgsp · e Ei −
PC
p
PC μj
PC
N 
   


∗ ∗ λj Np −(λ −1) PNs Ns
= C0C + e−Np Mj γ − e−Np Mj+1 γ Ei − μj − e j C Ei −
j=1
λj − 1 PC PC
N


− pμ Ns Ns
+ e PC j − 1 e PC Ei −
PC

1.5
of cognitive broadcast and multi-access channels presented
in [11] and [12], future work includes the extension of our
SU’s Achievable Data Rate [nats/sec/Hz]

N = −10dBw
p proposed scheme into the scenario with multiple secondary
users.
1

no Pc, N = −10dBw ACKNOWLEDGEMENT


p
Np = −10dBw The authors would like to thank the anonymous reviewers
no Pc, N = −20dBw
p
for bringing [4] and [12] to their attention.
0.5 N = −20dBw
p
Np = −20dBw
R EFERENCES
[1] A. Goldsmith, S. A. Jafar, I. Maric, and S. Srinivasa, “Breaking spectrum
gridlock with cognitive radios: an information theoretic perspective,”
0 Proc. IEEE, vol. 97, no. 5, pp. 894–914, May 2009.
0 5 10 15
Peak Transmit Power Constraint on the ST [dBw] [2] C. Stevenson, G. Chouinard, Z. Lei, W. Hu, S. Shellhammer, and
W. Caldwell, “IEEE 802.22: the first cognitive radio wireless regional
Fig. 3. Achievable data rate in the secondary link versus the peak transmit area network standard,” IEEE Commun. Mag., vol. 47, no. 1, pp. 130–
power constraint imposed on the ST, Pc , for the SR’s received AWGN Ns = 138, Jan. 2009.
−5dBw. Also shown is the case with no peak transmit power constraint at [3] J. S. Blogh and L. Hanzo, Third-Generation Systems and Intelligent
the ST. Wireless Networking: Smart Antennas and Adaptive Modulation. John
Wiley & Sons, Inc., 2002.
V. C ONCLUSIONS [4] Z. Zhang, H. Luo, J. Zhang, W. Wang, and G. Yu, “Cognitive radio
transmission strategies exploiting the primary link adaptivity,” IEEE
In general, secondary users (SUs) should stop their com- Trans. Veh. Technol., vol. 60, no. 8, pp. 3805–3813, Oct. 2011.
munications in a cognitive radio network once primary [5] A. Ghasemi and E. S. Sousa, “Capacity of fading channels under
users (PUs) emerge to occupy the licensed spectrum. For the spectrum-sharing constraints,” in Proc. 2006 IEEE Int. Conf. Commun.,
pp. 4373–4378.
purpose of increasing the SUs’ transmission opportunities, [6] L. Musavian and S. Aissa, “Ergodic and outage capacities of spectrum-
in this work we proposed a transmission strategy for SUs sharing systems in fading channels,” in Proc. 2007 IEEE Global
within a spectrum-sharing context where PUs adopt variable- Commun. Conf.
[7] L. Musavian and S. Aissa, “Capacity and power allocation for spectrum-
rate variable-power M QAM modulation. Even when a PU is sharing communications in fading channels,” IEEE Trans. Wireless
communicating, the SU can adjust its transmit power based Commun., vol. 8, no. 1, pp. 148–156, Jan. 2009.
on the gap between the PU’s received SNR and the lower [8] L. Musavian and S. Aissa, “Fundamental capacity limits of cognitive
radio in fading environments with imperfect channel information,” IEEE
SNR boundary for the modulation mode of the moment in the Trans. Commun., vol. 57, no. 11, pp. 3472–3480, Nov. 2009.
primary link, so as to promise the PU’s QoS. Considering dif- [9] A. J. Goldsmith and S.-G. Chua, “Variable-rate variable-power MQAM
ferent transmission abilities of the secondary transmitter (ST), for fading channels,” IEEE Trans. Commun., vol. 45, no. 10, pp. 1218–
1230, Oct. 1997.
we derived achievable data rates in the secondary link and [10] A. Jovicic and P. Viswanath, “Cognitive radio: an information-theoretic
obtained corresponding closed-form expressions. From numer- perspective,” IEEE Trans. Inf. Theory, vol. 55, no. 9, pp. 3945–3958,
ical results provided here, the validity of our derivations is Sep. 2009.
[11] V. Asghari and S. Aissa, “Resource management in spectrum-sharing
substantiated and we may get a conclusion that, with the pro- cognitive radio broadcast channels: adaptive time and power allocation,”
posed transmission strategy, the SU’s performance improves IEEE Trans. Commun., vol. 59, no. 5, pp. 1446–1457, May 2011.
with the decrease in the SR’s received AWGN power, while [12] B. Maham, P. Popovski, X. Zhou, and A. Hjorungnes, “Cognitive
multiple access network with outage margin in the primary system,”
with the increase in the PR’s received AWGN power or the IEEE Trans. Wireless Commun., vol. 10, no. 10, pp. 3343–3353, Oct.
peak transmit power constraint value at the ST. Finally, in light 2011.

Anda mungkin juga menyukai