This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication.

1

Adaptive Modulation in Spectrum-Sharing Channels

Under Delay Quality of Service Constraints
Leila Musavian, Member, IEEE, Sonia Assa, Senior Member, IEEE, and Sangarapillai Lambotharan, Senior
Member, IEEE

AbstractWe propose a variable-rate variable-power M-level

Quadrature Amplitude Modulation (MQAM) scheme employed
under delay quality-of-service (QoS) constraints over spectrumsharing channels. An underlay cognitive radio system, with
one primary user and a secondary user, with constraints on
interference leakage imposed by the primary receiver is considered. The transmission parameters of the secondary user are
set optimally such that the successful communications for the
primary user in terms of a minimum rate to be supported is
satisfied irrespective of co-existence with the secondary user.
We obtain interference constraints that, when satisfied by the
secondary users, is sufficient for satisfying the service outage
requirement of the primary user. We further study the performance of the secondary users link employing an adaptive
MQAM scheme when, on top of the above-mentioned interference
constraint, the secondary user is also required to satisfy a
statistical delay QoS constraint. Considering two modulation
schemes, namely, continuous MQAM, and discrete MQAM with
restricted constellations, we obtain the effective capacity of the
secondary users link, and derive the optimum power allocation
that maximizes the effective capacity of the secondary user.
Index TermsSpectrum sharing, Adaptive modulation, Delay
QoS constraint, Effective capacity, Service-outage constraint,
Fading channels.

I. I NTRODUCTION
Adaptive resource allocation is considered a powerful tool
for enhancing spectrum efficiency in current and futuregeneration wireless networks. In particular, adaptive power
and rate allocation is known to improve the performance of
wireless fading channels [1]. Several approaches and studies
have taken place to investigate the capacity gains that can be
achieved by these techniques. For instance, it has been shown
in [2] that adaptive power and rate allocation based on Mlevel Quadrature Amplitude Modulation (MQAM) achieves a
Copyright (c) 2010 IEEE. Personal use of this material is permitted.
However, permission to use this material for any other purposes must be
obtained from the IEEE by sending a request to [email protected].
Manuscript received October 01, 2009; revised June 02, 2010 and September 30, 2010; accepted October 12, 2010. The associate editor coordinating
the review of this paper and approving it for publication was Y. Ma. This
work has been supported by the Natural Sciences and Engineering Research
Council (NSERC) of Canada, under research grant RGPIN/22907-2005 and
by the Engineering and Physical Sciences Research Council (EPSRC), UK,
under grant EP/G020442/1. Part of this work is published in the proceedings
of ICC 2009.
Leila Musavian and Sonia Assa are with the INRS-EMT, University of
Quebec, Montreal, QC, Canada, Email:{musavian, aissa}@emt.inrs.ca
Sangarapillai Lambotharan is with the Advanced Signal Processing
Group, Loughborough University, Loughborough, Leicestershire, UK, Email:
[email protected]
Digital Object Identifier ................

20dB gain in spectral efficiency as compared to a nonadaptive scheme. The spectral efficiency can be enhanced further
through dynamic spectrum allocation. Enforced by regulatory
bodies, the spectrum allocation has traditionally followed policies where non-overlapping parts of the spectrum are allocated
to specific applications and users. Nevertheless, while we
witness a huge surge for new wireless applications, recent
spectrum allocation chart suggests that not much spectrum
is left for new applications and for the growing number of
wireless users [3]. Fortunately, recent spectrum measurements
have also shown that significant parts of the spectrum are
inefficiently utilized [3], paving the way for feasible sharing of
the spectrum using the so-called cognitive radio (CR) concept.
One of the major challenges for next-generation wireless
systems in general, and CR systems in particular, is to support
quality-of-service (QoS) requirements for different applications. Indeed, providing QoS measures for secondary users
is even more challenging due to the secondary type of access
to the radio spectrum. One of the critical QoS requirements is
the delay requirement for real-time or delay-sensitive applications. Generally, two different kinds of delay constraints are
considered in communications systems, namely, deterministic
and statistical. Imposing deterministic delay requirements, that
is, the delay should be less than a threshold at all times, is
very challenging or even impossible in fading channels due to
the variations in the capacity as a function of the channel gain
and the availability of channel state information (CSI) at the
transmitter and/or receiver [4]. On the other hand, statistical
delay QoS constraints, where delay is required to be lower
than a specific threshold only for a certain percentage of time,
are considered more pragmatic in various applications [5].
Studying the performance of wireless communication systems using the current physical-layer channel models is very
complicated since these models can not be translated to the
complex link-layer requirements, such as delay bound QoS
requirements [4]. Recently, the concept of effective capacity,
which is a link-layer channel model and aims to model the
wireless channel in terms of functions that can be mapped into
link-layer performance metrics, has been introduced in [4].
Effective capacity is the dual of effective bandwidth [5], and
can be interpreted as the maximum constant arrival-rate that
can be supported by the channel given that the delay QoS
requirement of the system is satisfied [4]. In this regard, an
optimum power and rate allocation strategy that maximizes the
effective capacity in fading channels has been obtained in [6].
It is worth mentioning that providing the QoS constraint in
cognitive radio channels is a further complicated task due to

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
2

the secondary type of access to the spectrum for secondary

users. This issue is very challenging and has attracted many
researchers, e.g., [7][9].
In assessing the performance of spectrum-sharing channels in fading environments, we refer to the earlier work
of Gastpar who presented capacity investigations of additive
white Gaussian noise (AWGN) spectrum-sharing channels
under interference power constraint rather than transmit power
constraint [10]. Later, ergodic and outage capacity metrics of
a point-to-point system with constraints on the received-power
at the primarys receiver in fading environment were derived
in [11] and [12]. The underlay spectrum-sharing approach is
considered in [13] wherein optimum power allocation strategies are proposed such that the interference to the primary
user is minimized while a set of minimum signal-to-noise
ratio (SNR) targets is provided at the secondary receivers.
A delay QoS-driven power and rate allocation scheme under
spectrum-sharing constraint was proposed in [14], wherein
the effective capacity of a point-to-point channel in Rayleigh
fading environment was determined.
In this paper, we consider spectrum sharing systems for
which the transmission of the secondary user is subject to
constraints on the interference-power inflicted on the primary
receiver. In general, we assume that there are guidelines and
limitations set by the regulatory bodies on the maximum
interference power, in terms of peak or average values, inflicted on the primary users. In addition, there are certain
service outage constraints for the primary users that should be
satisfied irrespective of the existence of the secondary users
in the network. Specifically, we wish to limit the transmission
parameters of the secondary transmitter such that the primary
user is supported with a minimum-rate for a certain percentage of time. Translating this limitation into an interferencepower constraint, either on peak or average interference limits,
we obtain the maximum throughput of the secondary users
channel under delay QoS constraint by obtaining the effective
capacity of the channel. We determine the maximum arrivalrate that can be supported by the secondary users link subject
to satisfying a statistical delay QoS constraint, by obtaining
the effective capacity of the channel under adaptive MQAM
with interference-power constraint. We further obtain closedform expressions for the effective capacity and its power
allocation in Nakagami-m block-fading channels. The service
outage constraint considered in this work is different from
our previous work on effective capacity of cognitive radio
channel. In addition, in this paper we assume that secondary
users implement MQAM which has not been studied in our
previous publications.
The subsequent sections are organized as follows. In Section
II, we provide the channel and system models. The interference power is studied in Section III wherein the primary
users service outage constraint is translated into an average
or peak interference power constraint. The effective capacity
of the secondary user channel under average interference
power constraint is provided in closed-form in Section IV. The
effective capacity of the channel under peak interference power
constraint is studied in Section V. Numerical results are given
in Section VI, followed by conclusions in Section VII.

II. S YSTEM M ODEL

The transmission parameters of the secondary user are
chosen such that the service outage requirement of the primary
user is satisfied. The effect of the transmission of the primary
user on the secondary receiver is assumed as AWGN. In the
secondary user communication system, the upper layer packets
are organized into frames with duration Tf . The secondary
transmitter employs adaptive MQAM with continuous or discrete constellations. Discrete-time block-fading channels are
assumed for both the secondary and primary users links.
The channel gain between the transmitter and receiver of
the secondary user and the AWGN are denoted by hs [t] and
zs [t], respectively, where t denotes the time index. We define
the channel gain between the secondarys transmitter and the
primarys receiver by hsp [t]. We assume hs [t] and hsp [t] are
statistically independent and identically distributed (i.i.d.) and
also independent from the noise. The channel envelopes are
distributed according to Nakagami-m fading. Channel gains
are stationary and ergodic random processes.
The secondary transmitter is provided with knowledge of
hs [t] and hsp [t]. Information about the latter can be obtained
from a band manager that intervenes between the primary and
secondary users [15], or can be directly fed back from the
primarys receiver to the secondary user, as proposed in [16],
[17] where the protocols allow the primary and secondary
users to collaborate and exchange CSI. The effect of imperfection in the knowledge of the channel gains between the
secondary transmitter and primary receiver at the secondary
transmitter on the ergodic capacity of the secondary users link
has been investigated in [18] for Rayleigh fading channels. The
secondary user knows only the statistical information of the
link between the transmitter and receiver of the primary user,
hp [t]. The instantaneous channel knowledge of hp [t] is known
to the primary users transmitter.
We consider a statistical delay constraint according to
out
,
Pr {D(t) Dmax } Pdelay

(1)

where D(t) indicates the delay experienced by a packet at

time instant t and Dmax is the maximum delay that can be
out
tolerated for 1 Pdelay
percentage of time. We further assume
that the transmission technique of the secondary user must
satisfy a statistical delay QoS constraint. It is shown that
the probability for the queue length of the transmit buffer
exceeding a certain threshold, x, decays exponentially fast as
a function of x [5], [6]. We now define as the delay QoS
exponent given by
ln (Pr {q() > x})
,
(2)
x
where q(n) denotes the transmit buffer length at time n and
Pr {a > b} denotes the probability that the inequality a > b
holds true. Considering a data source with constant data rate r,
the QoS exponent is related to the delay violation probability
according to
= lim

sup Pr {D(t) Dmax } (r)eDmax ,

(3)

where (r) = Pr{Q(t) 0} is the probability of a non-

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
3

empty buffer. Therefore the maximum constant arrival rate for

providing the delay constraint (1) can be obtained from

a Rayleigh fading environment, so is the service outage

constraint of the primary user.

out
= (r)eDmax .
Pdelay

Eh {Ps hsp } Iavg ,

(4)

Note that 0 corresponds to a system with no delay

constraint, while implies a strict delay constraint.
Considering as the delay QoS exponent, we obtain the
secondary users maximum supported arrival-rate given that
the QoS constraint is satisfied. An interested reader is referred
to [4] for more details. Note that effective capacity relates to
the asymptotic case for the delay and is defined for large value
of Dmax . However, it has been shown in [4] that this model
also provides a good estimate for small values of Dmax .
III. I NTERFERENCE -P OWER C ONSTRAINT
We recall that the transmission power of the secondary
user is limited such that the primary user is guaranteed
with a minimum-rate Rmin for a certain percentage of time
(1 Ppout ). We formulate the interference constraint starting
with the following outage probability:
{
}
Pr Rp Rmin Ppout ,
(5)
where Rp indicates the rate of the primary user link. The
transmission power of the primary user is assumed to be
constrained by an average level, Pp , i.e.,1
Ehp {Pp (hp )} Pp ,

(6)

where Ehp defines the expectation over the probability density

function (PDF) of hp , and Pp (hp ) is the input transmit power
of the primary user as a function of hp .
We consider two different transmission strategies for the
primary user: constant transmit power (cons) and optimum
power and rate allocation (opra) [1]. For opra scheme, the
transmission power at the primary transmitter for each time
instant is chosen using the CSI of hp available at the primary
transmitter. Note that the primary user chooses its transmit
power without taking into consideration the existence of the
secondary user in the network. The secondary transmitter, on
the other hand, should limit its transmit power such that the
communication process of the primary user is not harmed in
the sense defined in (5).
A. Constant Transmit Power Scheme (cons)
For this scheme, we assume that the primary user transmits
with fixed power, Pp , at all time. As such, one can show that
)
(
Pp hp
Rp ,
(7)
ln 1 +
Ps (, hs , hsp )hsp + N0 B
where Ps (, hs , hsp ) denotes the transmit power of the secondary user as a function of , hs , and hsp . The noise power
spectral density and received signal bandwidth are denoted by
N0 and B, respectively. Hereafter, for the ease of notations,
we use Ps to denote the transmit power of the secondary user.
Lemma1: When the following average interference constraint is satisfied by the secondary user, who operates in
1 Hereafter,

we omit the time index t whenever it is clear from the context.

(8)

Rmin
ln(1Ppout )
with k1 = e Pp 1 and Iavg =
N0 B which is
k1
referred to as average interference-limit.
Proof: We start by formulating the service outage constraint
of the primary user according to
{
}
Ppout Pr 0 hp k1 (Ps hsp + N0 B) ,
(9)

holds, then the interference power constraint (5) will be

satisfied. Now, the condition (9) can be expanded as
k1 (Ps hsp +N0 B)
Ppout
fhp (hp )dhp
0
0
0
{z
}
|
(10)
I0

fhsp (hsp )fhs (hs )dhsp dhs

where fx (x) indicates the PDF of the random variable x. The
PDF of hp is given by [19]
m

fhp (hp ) =

m 1

m p p hp p
(mp )

emp hp ,

(11)

where mp indicates the Nakagami-m parameter for the channel

gain between the transmitter and receiver of the primary user,

and () = 0 wz1 ew dw is the Gamma function [20].

The integral function in I0 can be extended using a change of
variable x = mp hp according to
fcons (Ps hsp )
1
I0 =
xmp 1 ex dx,
(12)
(mp ) 0
where fcons (Ps hsp ) = mp k1 (Ps hsp + N0 B). For integer mp ,
the solution for the integration operation (12) can be found as
(
)
mp 1 k
x
fcons (Ps hsp )
x

I0 = e
.
(13)
k! 
k=0

For the special case of Rayleigh fading, i.e., mp = 1, the

interference power constraint in (10) simplifies to
{
}
(14)
Ppout Ehs ,hsp 1 efcons (Ps hsp ) ,
where Ehs ,hsp defines the expectation over the joint PDF of hs
and hsp . Hereafter, we refer to Ehs ,hsp as Eh . The inequality
in (14) can be further simplified by using Jensens inequality
according to (8). This conclude the proof.
It is worth noting that when Iavg 0, no feasible power
allocation satisfying (8) exists, hence, the capacity lower
bound is zero. In the following, we assume Iavg > 0.
For the cases with integer mp > 1, the interference power
simplifies to
{
}
mp 1
(fcons (Ps hsp ))k
Ppout Ehs ,hsp 1 efcons (Ps hsp )
.
k!
k=0
(15)
Satisfying constraint (15) guarantees that the achievable-rate
of the primary user is bigger than Rmin for at least (1 Ppout )

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
4

percentage of time. Analyzing the throughput of the secondary

user link under the constraint in (15) is, however, complicated.
Hence, we proceed by simplifying the inequality in order to
obtain a simple peak interference power constraint, which will
be a sufficient condition to satisfy (5).
We proceed by simplifying (15) assuming that the amount of
the interference-power Ps hsp should satisfy the inequality (15)
at all times. Hence, we have Ps hsp Ipeak , where Ipeak is
the solution for x from

is sufficient for (20) to be satisfied. By following similar

approach as in Subsection A, we can obtain the interferencelimit in Lemma 2. This concludes the proof.
In order to obtain the peak power constraint, we first
set fopra (Ps hsp ) = mp (k1 (Ps hsp + N0 B) + k2 ). Now, peak
power constraints for integer mp and non-integer mp can be
found by replacing fcons (Ps hsp ) with fopra (Ps hsp ) in (15)
and (17), respectively, and assuming that the inequalities are
satisfied at all time.

mp 1

efcons (x)

(fcons (x))k
= 1 Ppout .
k!

(16)

k=0

For non-integer mp , the interference constraint can be found

Ehs ,hsp { (mp , fcons (Ps hsp ))} ,
(17)
t s1 x
where (s, t) = 0 x e dx is the lower incomplete
Gamma function [20]. A peak constraint can be obtained when
the above inequality is satisfied at all the time.
To summarize, we translated the service-outage constraint
of the primary user into a peak or average interference power
constraint.
Ppout

IV. AVERAGE I NTERFERENCE -P OWER C ONSTRAINT

A. Continuous MQAM
In this section, we consider the case when the service-outage
constraint of the primary user is translated into an average
interference-power constraint. The secondary transmitter employs MQAM with no restriction on the constellation size.
The constellation set is chosen adaptively while satisfying the
interference-power constraint.
Defining M (, hs , hsp ) as the number of points in the signal
constellation, i.e., modulation order, one can obtain an upper
bound on the required bit-error-rate (BER) of the system
according to [2]
1.5Ps hs

B. Optimum Power and Rate Allocation Scheme (opra)

Recall that the transmission parameters of the primary user
is chosen without any consideration to the presence of the
secondary user in the frequency band.
Lemma2: Satisfying average interference power constraint
Eh {Ps hsp } Iavg , with
)
(
ln 1 Ppout + k2
Iavg =
N0 B,
(18)
k1
where k1 = e 1 and k2 = N0 B , by a secondary user
who operates in a Rayleigh fading environment is sufficient to
satisfy the service outage constraint of the primary user who
employs opra technique at its transmitter.
Proof:The primary transmitter employs adaptive power and
rate allocation technique (opra) [1]. Hence, the power allocation of the primary user can be found as
[
]+
N0 B
Pp (hp ) =
,
(19)
hp
Rmin

where [x]+ indicates max{x, 0}. The cutoff threshold is

found so that the power constraint (6) is satisfied with equality.
The left-hand-side of the constraint in (5) can now be
expanded according to
}
{
}
{
N0 B
N0 B
out
+ Pr Rp Rmin , hp
.
PP Pr hp <

Using the inequality

)
}
{ (
Pp (hp )hp
Rmin Pr {Rp Rmin } ,
Pr ln 1 +
Ps hsp + N0 B
(20)
where Pp (hp ) is given in (19). Hence satisfying the inequality
{
}
Ppout Pr 0 hp (k1 (Ps hsp + N0 B) + k2 ) , (21)

BER 0.2e N0 B(M (,hs ,hsp )1) ,

(22)

M (, hs , hsp ) = 1 + KPs hs ,

(23)

which leads to
where K = N0 B 1.5
ln(5BER) . We further define {R[t], t =
1, 2, . . . } as the stochastic service rate which is assumed to
be stationary and ergodic. We can now obtain the service rate
of the MQAM scheme according to
(
)
R[t] = Tf B ln 1 + KPs [t]hs [t] .
We now introduce the concept of effective capacity and
obtain the effective capacity of the secondary user link when
employing adaptive MQAM under the interference-power constraint (8). Effective capacity was originally defined in [4] as
the dual concept of effective bandwidth. Assuming that the
function
})
1 ( { N
t=1 R[t]
() = lim
ln E e
,
(24)
N N
exists, the effective capacity is outlined as [4]
})
( {
N
1
()
= lim
ln E e t=1 R[t] .
Ec () =
N N

It is worth noting that the effective capacity quantifies the

maximum arrival-rate that can be supported by the channel
under the constraint of QoS exponent , interpreted as the
delay constraint. Moreover, in block-fading channels, where
the sequence R[t], t = 1, 2, . . . , is uncorrelated, the effective
capacity can be simplified to
})
1 ( {
(25)
Ec () = ln E eR[t] .

Having introduced the formulation for effective capacity, we

now obtain the optimum power and rate allocation that maximizes the effective capacity of the channel. This maximization

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
5

problem can be formulated as

integral in (32), J0 , can be obtained using [14] according to
{
(
(
)
{
(
) } )}

1
msp + 1+
Tf B ln 1+KPs hs
opt
Ec () = max ln Ehs ,hsp e
)
J0 = (
Ps 0

msp + 1+2
(33)
1+
s.t. Ehs ,hsp {Ps hsp } Iavg ,
(26)
(
)
1 + 2
K
2 F1 ms + msp , 1; msp +
;
,
where Ecopt () indicates the maximum of the effective capac1 + K +
ity. Using a similar approach as in [14], the solution for the
where 2 F1 (a, b; c; z) denotes the Gausss hypergeometric funcmaximization problem in (26) can be obtained as
tion [20]. A closed-from expression for J1 can also be ob[
]+
1
tained by following a similar approach. Now, by inserting (33)
1+
1
,
(27) into (32), and using the equality (1 + z) = z(z), we obtain
Ps =

Khs
1+
hsp
(Khs ) 1+
a closed-form expression for (32) according to
[(
)1
where = Tf B, = 0 , [x]+ denotes max{0, x}, and

ms (K)msp +1
1
0 = 0 , 0 being the Lagrangian multiplier chosen to satisfy Iavg =
m
+
sp
K(ms , msp )(K + )ms +msp
1+
the interference-power constraint in (8) with equality. The
(
)
power allocation policy can be expressed as
1 + 2
K
2 F1 ms + msp , 1; msp +
;

1
1 + K +
1+
1

(
) ]

if hsp Khs ,
1

1
K
Kh
1+
Ps = hsp (Khs ) 1+
(28)
s

,
2 F1 ms + msp , 1; msp + 2;

msp + 1
K +
0
otherwise.
from which 0 can be obtained. We now derive a closed
In order to obtain a solution for 0 =
, we need to evaluate from expression for the effective capacity of the channel by
the integration in
evaluating the integration in (26) as follows:

(
) 1+
)
Khs (
})
( {(
1
h
h
]+ )
[
sp
sp
1
1
1
Iavg =
1+

opt
Ec () = ln Ev
1 + (K) 1+ v 1+ 1
(29)
Khs
Khs
0
0

fhsp (hsp )fhs (hs )dhsp dhs .

( ms
1
1
(K) 1+ K v msp 1+
= ln
dv
m +m
Noting that (29) depends on the channel gains only through

(ms , msp ) 0
(v + ) s sp
hsp
)

ratio values, we define a new random variable v = hs . Using

ms
v msp 1
+
dv . (34)
the fact that the distribution of the ratio between two Gamma
(ms , msp ) K (v + )ms +msp
distributed random variables with parameters 1 and 2 is a
vmsp 1
ms
beta prime distribution with parameters 1 and 2 [11], [21], By using
(ms ,msp ) 0 (v+)ms +msp dv = 1, we get (37).
we can determine the distribution of the random variable v as
ms
v msp 1
fv (v) =
,
(30) B. Restricted MQAM
(ms , msp ) (v + )ms +msp
We now consider the case when the number of signal
(m )(m )

where = ms and (ms , msp ) = (mss +mspsp) with (z) =

z1 tmsp
t e dt defining the Gamma function [20]. We now
0
obtain the solution for 0 by evaluating the integration in (29)
as follows:
K (
)
1

ms
Iavg =
(K) 1+ v 1+ v
K(ms , msp ) 0
(31)
v msp 1

dv
m +m
(v + ) s sp
ms (K)msp +1
=
(32)
K(ms , msp )(K + )ms +msp
[
(
)(ms +msp )
1

K
msp 1+ 1+
1
x

(1 x)
dx
K +
{z
}
|0
J0
(
)(ms +msp ) ]
1
K
msp

(1 x)
1
x
dx ,
K +
0
{z
}
|
J1

where x = 1

v
K .

A closed-form expression for the first

points in the MQAM is not continuous but restricted to a set

{Mn : n = 0, ..., N }, where Mn = 2n . The spectral efficiency
related to each constellation is given by n(bits/s/Hz). As such,
the service rate can be found according to rn = Tf Bn [6],
with rn denoting the service rate of the n_th mode. At
each time, the secondary transmitter chooses an appropriate
constellation size based on its own channel gain, hs , the
channel gain between its transmitter and the primary receiver,
hsp , and the delay QoS exponent, . In addition, the secondary
transmitter should determine the transmission power that satisfies the BER requirement of the system, the interference-power
restriction (8), and the delay QoS constraint.
As stated earlier, the effective capacity of the channel in the
continuous constellation case depends on the channel gains hs
and hsp only through the ratio of these two parameters. Using
this fact, we partition the entire range for the random variable
w , hhsps into N non-overlapping intervals, and denote the set
pertaining to the boundaries of these intervals as {Wn : n =
0, ..., N + 1}, with W0 = 0 and WN +1 = . We associate
the constellation Mn to the n-th boundary which refers to the
case when Wn w < Wn+1 . The constellation employed in

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
6

Ecopt ()

(
1
(K)msp ms
(
= ln 1+

(ms , msp )( + K)ms +msp msp +

(
)
1 + 2
K
) 2 F1 ms + msp , 1; msp +
;

1 + + K
1+
(
))
(K)msp ms
K

.
(37)
2 F1 ms + msp , 1; msp + 1;
(ms , msp )msp ( + K)ms +msp
+ K

the 0-th interval is M0 = 0, meaning that the transmission is

cut off when v < W1 , or equivalently, when the secondary
users channel gain is weak compared to hsp .
We now need to find the boundary points and the transmission power for each interval that maximizes the effective
capacity of the secondary user while satisfying the interference
power constraint and the BER requirement of the system. For
this purpose, we first obtain the optimal boundary points by
inserting the power allocation (27) into (23), yielding
(
) 1
wK 1+
M (, hs , hsp ) =
.
(38)
0
Using (38) as a guideline, we obtain the boundary points as
0
,
(39)
K
where 0 should be found such that the interference-power
constraint in (8) is satisfied with equality. Once the boundary
points and their associated constellations are found, we need
to obtain the transmission power level at each boundary. A
fixed BER means that the received SNR is fixed. As such, the
power allocation can be obtained using (23), according to

Mn 1 W w < W
n
n+1 , n = 1, ..., N,
Khs
Ps =
(40)
0
0w<W .
Wn = Mn1+

can be obtained by inserting (40) into the

The parameter
interference-power constraint (8), and replacing the inequality
with equality, thus yielding
N
1 M +1 0

n+1 K

Iavg =

n=1

+1 0
Mn
K

+1 0
MN
K

Mn 1
1

fw (w)dw
K
w

MN 1
1

fw (w)dw,
K
w

(41)

where
fw (w) =

msp
wms 1
.
(msp , ms ) (w + 1/)msp +ms

(42)

Finally, the effective capacity in this case can be found as

( N
1 M +1 0
n+1 K
1
dis
Ec () = ln
Mn fw (w)dw

+1 0

K
n=1 Mn
(43)
)

MN fw (w)dw .
+

+1 0
MN
K

V. P EAK I NTERFERENCE -P OWER C ONSTRAINT

Here, we consider the case when the service-outage constraint of the primary user is translated into peak interference-

power constraint, and obtain the maximum arrival rate for the
secondary user under delay QoS constraint.
A. Continuous MQAM
In this case, the power of the secondary user can be found
Ipeak
as Ps =
. Therefore, the service rate is given by
hsp
(
)
hs
R[t] = Tf B ln 1 + Ipeak K
, which leads to the effective
hsp
capacity
(
{(
) })
hs
1
1 + Ipeak K
. (44)
Ec () = ln Ehs ,hsp

hsp
A closed-from expression for the effective capacity can
be obtained according to (45), see Appendix A, where
F1 (a; ; ; ; x, y) is the appell hypergeometric function of
the first kind defined in [22] as
F1 (a; ; ; ; x, y) =

(a)m+n ()m ( )n m n
x y ,
m!n!()m+n
m=0 n=0

with (x)n = x(x+1) . . . (x+n1) indicating the Pochhammer

symbol [20].
B. Restricted MQAM
Here, we study the effective capacity of the secondary
users link under peak interference power constraint when
the secondary transmitter emblements discrete MQAM. We
partition the entire range for the random variable W into
N + 1 non-overlapping regions. In order to satisfy the peak
interference power constraint , the secondary users transmit
Ipeak
power should be limited to
. Now, using (23), we get
hsp
M (, hs , hsp ) = wIpeak , where can be used as a guideline to
Mn
. Therefore,
obtain boundary points according to Wn =
Ipeak
the effective capacity can be obtained according to
( N
1 Mn+1
Ipeak
1
dis
Mn fw (w)dw
Ec () = ln
Mn

n=1 Ipeak
) (46)

+

MN
Ipeak

MN
fw (w)dw .

VI. N UMERICAL R ESULTS

In this section, we numerically evaluate the effective capacity of the secondary users link in Nakagami-m block fading
under peak or average interference-power constraints when the

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
7

( (
)
)
1
1 msp (KIpeak ) (ms )(msp + )
1

ln
F
m
+
,
,
m
+
m
;
m
+
m
+
;
1

;
1

1
sp
s
sp
s
sp

KIpeak

(msp , ms )(ms + msp + )

for 0.5 KIpeak and 0.5 ,

(
)
Ec () =
1
ms (ms )(msp + )

ln F1 (ms , , ms + msp ; ms + msp + ; 1 KIpeak ; 1 )

,

(msp , ms )(ms + msp + )

for KIpeak 2 and 2,

(45)

0.7

10

out

ms=msp=1

Pp =1%

Normalized Effective Capacity (nats/s/Hz)

Pout=2%

10

Pout=3%
p

avg

(watts)

10

10

10

Iavg>0

Iavg>0
Iavg>0

Iavg>0

Iavg>0
Iavg>0

10

10

0.1

0.2

0.3

0.4

0.5
0.6
0.7
Rmin(nats/s/Hz)

0.8

0.9

1.1

Fig. 1: Average Interference-limit versus Rmin for various outage

probabilities for cons (solid lines) and opra (dashed lines).

0.6

ms=1, msp=1.5
ms=1, msp=2

0.5

0.4

0.3

0.2

0.1
5

3
2
Interference limit (dBW)

Fig. 3: Normalized effective capacity of the secondary link versus

interference-limit, average (solid lines) or peak (dashed lines).

10

0.45

mP=4
10

mP=2
0

Ipeak (watts)

10

10

Ipeak>0

10

Ipeak>0

Ipeak>0

10

Ipeak>0
I

10

>0

peak

0.6

0.8

1.2
1.4
1.6
Rmin(nats/s/Hz)

1.8

2.2

m =m =1
s

Normalized Effective Capacity (nats/s/Hz)

mP=3

sp

m =2, m =1

0.4

sp

m =3, m =1
s

sp

m =1, m =2

0.35

sp

ms=1, msp=3
0.3
0.25
0.2
0.15
0.1
0.05
3
10

10

10
(1/nats)

10

10

Fig. 2: Peak Interference-limit versus Rmin for various outage

probabilities for cons (solid lines) and opra (dashed lines).

Fig. 4: Normalized effective capacity of the secondary user versus

QoS exponent for various Nakagami parameters, ms and msp .

secondary transmitter employs MQAM adaptive modulation

scheme. Hereafter, we assume Tf B = 1.
We start by examining the effect of different transmission techniques, namely, opra and cons adopted by the primary user, on the interference constraints obtained in this
paper. Fig. 1 depicts the average interference-limit versus
the minimum-rate required by the primary user with Pp =
15dBW. The solid and dashed lines represent opra and cons
techniques, respectively. The arrows indicate the regions for
which Iavg 0 holds true. The figure shows that after certain
thresholds for Rmin , the interference-limit decreases rapidly as
the minimum rate, Rmin , increases or as the outage probability

decreases. The figure also reveals that the interference-power

constraint obtained when the primary user employs cons
techniques is much tighter than those with opra case.
Fig. 2, on the other hand, shows the results for the peak
interference power limit, Ipeak , obtained in Section III for
Nakagami fading parameters mp = 1. The plots depict the
peak interference-limit values versus the required minimumrate for the primary user with Pp = 15dBW for different Nakagami fading parameters mp . The figure shows that when mp
increases the peak interference-limit increases significantly.
We continue by examining the effective capacity of the
secondary users when the secondary transmitter employs

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
8

0.4
Rayleigh, BER=10(3)
Normalized Effective Capacity (nats/sHz)

Normalized Effective Capacity (nats/a/Hz)

(5)

Rayleigh, BER=10

0.35

m =m =2, BER=10(3)
s

sp

ms=msp=2, BER=10(5)

0.3

0.25

0.2

0.15

2.8

2.6

2.4

2.2

1.8

mp=2

0.1

mp=3
1.6

0.05
3
10

10

10
(1/nats)

1.2

1.4

1.6

1.8

10

Fig. 5: Normalized effective capacity of the secondary users link

versus QoS exponent, , for various Nakagami parameters, ms and
msp , and BER requirements.

2.2

2.4

2.6

2.8

Pout
(%)
p

10

Fig. 7: Normalized effective capacity of the secondary users link

versus Ppout with Pp = 15dBW, Rmin = 0.1nats/s/Hz, = 0.1,
BER=103 , and Nakagami parameters ms = msp = 1.
3
out

Pp =1%

0.9

Normalized Effective Capacity (nats/s/Hz)

Normalized Effective Capacity (nats/s/Hz)

1
Optimum case
Continuous MQAM
Discrete MQAM

0.8
0.7
0.6
0.5
0.4

Pout
=2%
p

2.5

Pout=3%
p

2
PP=15dBW
1.5

0.5
Pp=12dBW
0
0.1

0.3
0.2
5

2
Iavg (dBW)

Fig. 6: Normalized effective capacity of the secondary users link

versus Iavg , with = 0.1, BER=103 , and ms = msp = 2.

continuous MQAM for different Nakagami fading parameters. Fig. 3 depicts the normalized effective capacity versus
average (solid lines) and peak (dashed lines) interferencelimit values with = 0.1(1/nats) and BER = 103 . This
figure includes the plots for the expectation equations of the
effective capacity, i.e., (34) and (44), and their corresponding
closed-from expressions, i.e., (37) and (45). The plots from
the expectation equations are shown by different markers with
no lines. The closed-from expressions are shown with lines,
steady and dashed lines, with no markers. As the figure shows
the closed-from expressions and the expectation equations
match perfectly. We further observe that when the Nakagami
parameter of the interference link, msp , increases the effective
capacity decreases. The figure also reveals that the capacity
under average interference constraint is considerably higher
than that under peak interference power constraint.
On the other hand, in Fig. 4, we keep the fading parameter
of one of the links, either hs or hsp , fixed and change the
parameter on the other link. The figure includes plots for
the effective capacity versus , with Iavg = 5dBW and
BER = 103 . The figure reveals that the changes in the
fading parameter of the secondary users link have negligible

0.2

0.3

0.4

0.5
0.6
Rmin(nats/s/Hz)

0.7

0.8

0.9

Fig. 8: Normalized effective capacity of the secondary users link

versus Rmin under opra technique with mp = 3, = 0.1,
BER=103 , and ms = msp = 1.

effects on the effective capacity as long as the fading parameter

pertaining to hsp is fixed. On the other hand, increasing the
Nakagami parameter of hsp degrades the effective capacity of
the secondary users link significantly.
Plots for the normalized effective capacity versus the delay
QoS exponent under average interference-power constraint
at Iavg = 5dBW, are provided in Fig. 5. We observe that
the capacity increases as decreases, however, the gain in the
effective capacity decreases for lower values of .
Fig. 6 depicts the effect of different modulation techniques
on the effective capacity of the secondary users link. The
figure includes plots for three different cases, namely, continuous MQAM, discrete MQAM, and the case when there
is no restriction on the coding employed by the secondary
transmitter, referred to as the optimum case. In this figure,
has been set to = 0.1 (1/nats), BER=103 , and N = 5.
The figure shows that the capacity with discrete MQAM is
smaller than that with continuous MQAM. The loss in the
capacity, however, is small when compared to the one between
the optimum case and continuous MQAM.
We further examine the effect of the service-outage probability of the primary user, Ppout , on the achievable effective
capacity of the secondary users link in Fig. 7 and Fig. 8. In

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
9

particular, Fig. 7 depicts the plots for the effective capacity

of the secondary user versus Ppout for various Nakagami
parameters for the primary users link, mp , under opra (solid
lines) and cons (dashed lines) schemes with Pp = 15dBW,
Rmin = 0.1nats/s/Hz, = 0.1, BER=103 , and ms = msp =
1. The figure reveals that under the same fading parameters
and service-outage constraints the effective capacity of the
secondary user link is higher when primary user employs cons
scheme compared to opra technique.
Fig. 8 includes the plots for the effective capacity versus
the minimum-rate required by the primary user for various
primary service-outage probabilities under opra transmission
technique with = 0.1, BER=103 , and Nakagami parameters
mp = 3 and ms = msp = 1. The solid and dashed lines
refer to Pp = 15dBW and Pp = 12dBW, respectively. The
figure shows that the capacity decreases significantly when
the minimum-rate required by the primary user increases.
VII. C ONCLUSIONS
We considered spectrum-sharing channels in Nakagamim fading environments and studied the effects of adaptive
MQAM modulation on the capacity gain of the secondary
users channel under delay QoS constraints. We assumed
that the spectrum band occupied by a primary user may be
accessed and utilized by a secondary user, as long as the
latter adheres to interference limitations set by the primary
user. Specifically, the successful communication process of the
primary user requires a minimum-rate to be supported by its
channel for a certain percentage of time. We obtained average
or peak interference-power constraints as a sufficient condition
for satisfying the service-outage requirement of the primary
user. Under average or peak interference-power constraint, we
obtained the effective capacity of the secondary users channel
for two different modulation schemes, namely, continuous
MQAM and discrete MQAM with limited constellations. For
these schemes, we determined the optimal power and rate
allocation strategies that maximize the effective capacity. Also,
we obtained closed-form expressions for the capacity and
the corresponding power allocation policy under Nakagamim block-fading for continuous MQAM. Considering the Nakagami parameter m as a measure of fading severity, it has been
observed that the effective capacity of the secondary user is
more sensitive to the fading severity of the interference link
between secondary transmitter and primary receiver compared
to the one between the secondary transmitter and receiver of
the secondary user.
A PPENDIX A
The integration in the effective capacity formula in (45) can
be expanded as follows
(
msp
1
Ec () = ln

(msp , ms )
)

wms 1

(1 + KIpeak w) (
)ms +msp dw ,
0
w + 1
|
{z
}
I

where w = v1 and I can be simplified by using the change

1
of variable x = 1+w
according to
(
(
) )
1
1

+msp 1
I = (KIpeak )
x
1 1
x
KIpeak
0
(
(
) )(ms +msp )
1
(1 x)ms 1 1 1
x
dx. (47)

Then, using the following expression [20]:

1
(a)( a)
F1 (a; ; ; ; x, y) =
ta1
()
0

(48)

(1 t)a1 (1 tx) (1 ty) dt,

for Re(a) > 0, Re( a) > 0, |x| < 1 and |y| < 1.
and inserting (48) into (47) when setting a = msp +, = ,
= ms + msp , = ms + msp + , x = 1 KI1peak , and
y = 1 1 , we get
(KIpeak ) (ms )(msp + ) (
F1 msp + ,
(ms + msp + )
(49)
1)
1
;1
.
, ms + msp ; ms + msp + ; 1
KIpeak

I=

Note that the condition |x| < 1 and |y| < 1 imply that
KIpeak > 0.5 and > 0.5, respectively.
We now obtain an alternative solution for the closed-from
expression of the effective capacity when the above-mentioned
inequalities on KIpeak and do not hold. We first apply the
w
change of variable x = 1+w
on I:
1
I = ms +msp
xmsp 1 (1 x)ms +1
(50)
0

(1 (1 KIpeak ) x)

(ms +msp )

(1 (1 ) x)

dx.

Now, by setting a = msp , = , = ms + msp , =

ms + msp + , x = 1 KIpeak , y = 1 , and inserting (48)
into (50), we get
ms +msp (ms )(msp + )
(51)
(ms + msp + )
F1 (ms , , ms + msp ; ms + msp + ; 1 KIpeak ; 1 ) ,

I=

where the conditions |x| 1 and |y| 1 imply that KIpeak <
2 and < 2, and as such (51) is correct when 0 KIpeak < 2
and 0 < 2. This concludes the proof for (45).
R EFERENCES
[1] A. J. Goldsmith and P. P. Varaiya, Capacity of fading channels with
channel side information, IEEE Trans. Inf. Theory, vol. 43, no. 6, pp.
19861992, Nov. 1997.
[2] A. J. Goldsmith and S.-G. Chua, variable-rate variable-power MQAM
for fading channels, IEEE Trans. Commun., vol. 45, no. 10, pp. 1218
1230, Oct. 1997.
[3] T. A. Weiss and F. K. Jondral, Spectrum pooling: An innovative strategy
for the enhancement of spectrum efficiency, IEEE Commun. Mag.,
vol. 42, no. 3, pp. S8S14, Mar. 2004.
[4] D. Wu and R. Negi, Effective capacity: A wireless link model for
support of quality of service, IEEE Trans. wireless Commun., vol. 2,
no. 4, pp. 630643, July 2003.
[5] C.-S. Chang, Stability, queue length, and delay of deterministic and
stochastic queueing networks, IEEE Trans. Automatic Control, vol. 39,
no. 5, pp. 913931, May 1994.

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
10

[6] J. Tang and X. Zhang, Quality-of-service driven power and rate

adaptation over wireless links, IEEE Trans. Wireless Commun., vol. 6,
no. 8, pp. 30583068, Aug. 2007.
[7] W. Chen, K. B. Letaief, and Z. Cao, A joint coding and scheduling
method for delay optimal cognitive multiple access, in Proc. IEEE Conf.
Commun. (ICC 2008), China, Beijing, May 2008.
[8] M. Rashid, M. Hossain, E. Hossain, and V. Bhargava, Opportunistic
spectrum scheduling for multiuser cognitive radio: a queueing analysis,
IEEE Trans. Wireless Commun., vol. 8, no. 10, pp. 5259 5269, Nov.
2009.
[9] L. Gao and S. Cui, Power and rate control for delay-constrained
cognitive radios via dynamic programming, IEEE Trans. Veh. Technol.,
vol. 58, no. 9, pp. 4819 4827, Nov. 2009.
[10] M. Gastpar, On capacity under received-signal constraints, in Proc.
42nd Annual Allerton Conf. on Commun. Control and Comp., Monticello, USA, Sept. 29-Oct. 1 2004.
[11] A. Ghasemi and E. S. Sousa, Capacity of fading channels under
spectrum-sharing constraints, in Proc. IEEE Int. Conf. Commun. (ICC),
Istanbul, Turkey, June 2006, pp. 43734378.
[12] L. Musavian and S. Assa, Capacity and power allocation for spectrumsharing communications in fading channels, IEEE Trans. Wireless
Commun., vol. 8, no. 1, pp. 148156, Jan. 2009.
[13] K. T. Phan, S. A. Vorobyov, N. D. Sidiropoulos, and C. Tellambura,
Spectrum sharing in wireless networks via QoS-aware secondary multicast beamforming, IEEE Trans. Signal Proc., vol. 57, no. 6, pp. 2323
2335, June 2009.
[14] L. Musavian and S. Assa, Quality of service based power allocation
in spectrum-sharing channels, in Proc. IEEE Global Commun. Conf.
(Globecom 2008), Neworleans, LA, USA, 30 Nov.-4 Dec. 2008.
[15] J. M. Peha, Approaches to spectrum sharing, IEEE Commun. Mag.,
vol. 43, no. 2, pp. 1012, Feb. 2005.
[16] A. Jovicic and P. Viswanath, Cognitive radio: An information-theoretic
perspective, submitted to IEEE Trans. Inf. Theory, May 2006, available
online at http://arxiv.org/PS_cache/cs/pdf/0604/0604107v2.pdf.
[17] L. Zhang, Y.-C. Liang, and Y. Xin, Joint beamforming and power
allocation for multiple access channels in cognitive radio networks,
IEEE J. Sel. Areas Commun., vol. 26, no. 1, pp. 3851, Jan. 2008.
[18] L. Musavian and S. Assa, Outage-constrained capacity of spectrumsharing channels in fading environments, IET Commun., vol. 2, no. 6,
pp. 724732, July 2008.
[19] M. K. Simon and M.-S. Alouini, Digital Communication over Fading
Channels: A Unified Approach to Performance Analysis. John Wiley
and Sons Inc., 2000.
[20] M. Abramowitz and I. A. Stegun, Handbook of mathematical functions.
New York: Dover, 1965.
[21] E.
W.
Weisstein.
(From
mathWorlda
wolfram
web
resource)
Gamma
distribution.
[Online].
Available:
http://mathworld.wolfram.com/GammaDistribution.html
[22] W. N. Bailey, On the reducibility of appells function, Quart. J. Math.
(Oxford), no. 5, pp. 291292, 1934.

Leila Musavian (S05-M07) obtained her Ph.D.

degree in Electrical and Electronics Engineering
from Kings College London, London, UK, in 2006
and her B.Sc. degree in Electrical Engineering from
Sharif University of Technology, Tehran, Iran, in
1999. she was a post doctoral fellow at National Institute of Scientific Research-Energy, Materials, and
Telecommunications (INRS-EMT), University of
Quebec, Montreal, Canada. Afterwards, she joined
the Advanced Signal Processing Group, Electronics
and Electrical Engineering, Loughborough University, Leicestershire, UK as a research associate. She has been TPC member
of ICC2008, ICC2009, Globecom2008, WCNC2009 and Globecom2009.
Dr. Musavians research interests include radio resource management for
next generation wireless networks, Cognitive radios, performance analysis of
MIMO systems, and cross-layer design.

Sonia Assa (S93-M00-SM03) received her Ph.D.

degree in Electrical and Computer Engineering
from McGill University, Montreal, QC, Canada, in
1998. Since then, she has been with the National
Institute of Scientific Research-Energy, Materials,
and Telecommunications (INRS-EMT), University
of Quebec, Montreal, Canada, where she is currently
a Full Professor.
From 1996 to 1997, she was a Researcher with
the Department of Electronics and Communications
of Kyoto University, Kyoto, Japan, and with the
Wireless Systems Laboratories of NTT, Kanagawa, Japan. From 1998 to 2000,
she was a Research Associate at INRS-EMT, Montreal. From 2000 to 2002,
while she was an Assistant Professor, she was a Principal Investigator in
the major program of personal and mobile communications of the Canadian
Institute for Telecommunications Research (CITR), leading research in radio
resource management for code division multiple access systems. From 2004
to 2007, she was an Adjunct Professor with Concordia University, Montreal.
In 2006, she was Visiting Invited Professor with the Graduate School of
Informatics, Kyoto University, Japan. Her research interests lie in the area of
wireless and mobile communications, and include radio resource management,
cross-layer design and optimization, design and analysis of multiple antenna
(MIMO) systems, and performance evaluation, with a focus on Cellular, Ad
Hoc and Cognitive Radio networks.
Dr. Assa was the Founding Chair of the Montreal Chapter IEEE Women
in Engineering Society in 2004-2007, a Technical Program Cochair for the
Wireless Communications Symposium (WCS) of the 2006 IEEE International
Conference on Communications (ICC 2006), and PHY/MAC Program Chair
for the 2007 IEEE Wireless Communications and Networking Conference
(WCNC 2007). She was also the Technical Program Leading Chair for the
WCS of the IEEE ICC 2009, and is currently serving as Cochair for the WCS
of the IEEE ICC 2011. She has served as a Guest Editor of the EURASIP
journal on Wireless Communications and Networking in 2006, and as Associate Editor of the IEEE W IRELESS C OMMUNICATIONS M AGAZINE in 20062010. She is currently an Editor of the IEEE T RANSACTIONS ON W IRELESS
C OMMUNICATIONS, the IEEE T RANSACTIONS ON C OMMUNICATIONS and
the IEEE C OMMUNICATIONS M AGAZINE, and Associate Editor of the Wiley
Security and Communication Networks Journal. Awards and distinctions to
her credit include the Quebec Government FQRNT Strategic Fellowship for
Professors-Researchers in 2001-2006; the INRS-EMT Performance Award
in 2004 for outstanding achievements in research, teaching and service; the
IEEE Communications Society Certificate of Appreciation in 2006, 2009 and
2010; and the Technical Community Service Award from the FQRNT Center
for Advanced Systems and Technologies in Communications (SYTACom)
in 2007. She is also co-recipient of Best Paper Awards from IEEE ISCC
2009, WPMC 2010 and IEEE WCNC 2010; and recipient of NSERC (Natural
Sciences and Engineering Research Council of Canada) Discovery Accelerator
Supplement Award.

Dr. Sangarapillai Lambotharan holds a Readership in Communications within the Advanced Signal
Processing Group, Department of Electronic and
Electrical Engineering, Loughborough University,
UK. He received a PhD degree in Signal Processing from Imperial College, UK in 1997, where
he remained until 1999 as a postdoctoral research
associate working on an EPSRC funded project on
mobile communications. He was a visiting scientist
at the Engineering and Theory Centre of Cornell
University, USA in 1996. Between 1999 and 2002,
he was with the Motorola Applied Research Group, UK and investigated
various projects including physical link layer modelling and performance characterization of GPRS, EGPRS and UTRAN. He has been with Kings College
London, UK and Cardiff University, UK as a lecturer and senior lecturer
respectively from 2002 to 2007. His current research interests include spatial
diversity techniques, wireless relay networks and cognitive radios. He serves
as an associate editor for EURASIP Journal on Wireless Communications and
Networking.

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing [email protected].