Distance- and Trafc-Aware Channel Assignment in

Cognitive Radio Networks

Haythem Bany Salameh and Marwan Krunz Ossama Younis
Department of Electrical and Computer Engineering Applied Research, Telcordia Technologies, Inc.
University of Arizona, Tucson, AZ 85721 Piscataway, NJ 08854
Email: haythem,krunz@ece.arizona.edu Email: oyounis@research.telcordia.com
Abstract The scarcity of unlicensed spectrum has triggered
great interest in cognitive radio (CR) technology as a means
to improve spectrum utilization. An important challenge in this
domain is how to enable nodes in a CR network (CRN) to access
the medium opportunistically. Multi-channel MAC protocols for
typical ad hoc networks assume that frequency channels are ad-
jacent and that there are no strict constraints on the transmission
power. However, a CRN may occupy a wide range of frequencies.
In addition, a power mask is often enforced on the transmission
power of a CR user to avoid corrupting the transmissions of
spectrum-licensed primary-radio (PR) users. Obviously, CR users
operating in different licensed bands will be subject to different
PR-to-CR interference conditions. To avoid unnecessary blocking
of CR transmissions under these constraints, we propose a
novel distance-dependent MAC protocol for CRNs (DDMAC) that
attempts to maximize the CRN throughput. DDMAC introduces a
novel suboptimal probabilistic channel assignment algorithm that
exploits the dependence between the signals attenuation model
and the transmission distance while considering the trafc prole.
The protocol allows a pair of CR users to communicate on a
channel that may not be optimal from one users perspective,
but that allows more transmissions to take place simultaneously,
especially under moderate to high trafc loads. Simulation results
indicate that compared to typical multi-channel CSMA-based
protocols, DDMAC decreases the connection blocking rate of CR
transmission requests by up to 30%, which improves the network
throughput at no additional cost in energy consumption. On the
whole, our protocol is simple yet effective. It can be incorporated
into existing multi-channel systems with little extra processing
overhead.
I. INTRODUCTION
The enormous growth in the number of unlicensed wireless
applications resulted in crowding the unlicensed portion of
the frequency spectrum (e.g., the ISM bands). While this is
happening in the unlicensed bands, the FCC Spectrum Policy
Task Force (SPTF) indicated that at any given time and in any
geographic locality, less than 10% of the licensed spectrum is
being utilized [1]. These two factors motivated the need for a
new spectrum-adaptive technology that improves the spectrum
efciency without degrading the performance of licensed PR
networks (PRNs). To cope with the rising demand in the
unlicensed services, the cognitive radio (CR) technology has
been proposed as a means to improve the spectrum utilization
by allowing an open access policy subject to a predetermined
etiquette. In a CRN, users are aware of the radio frequencies
used by existing legacy networks. They opportunistically adapt
their communication parameters to be able to communicate
without affecting active PR users.
This research was supported in part by NSF (under grants CNS-0721935,
CNS-0627118, CNS-0325979, and CNS-0313234), Raytheon, and Connection
One (an I/UCRC NSF/industry/university consortium).
A CRN has unique characteristics that distinguish it from
traditional multi-channel wireless communication networks.
Unlike traditional wireless networks, which typically occupy
contiguous bands [2][4], a CRN is expected to operate over a
set of highly separated non-contiguous frequency bands, which
exhibit different RF attenuation and interference behaviors.
It is well known that signal attenuation increases with the
distance between the two communicating users and also with
the carrier frequency used for communication [5]. Therefore,
when assigning channels for its transmissions, it is necessary
for a CRN to consider the signal attenuation model and the
interference conditions so as to improve the utilization of the
spectrum. Another characteristic of a CRN is that users must
operate using a relatively low transmission power (i.e., subject
to power mask) to avoid degrading the performance of the PR
users [6], [7]. These peculiar characteristics call for new MAC
protocols that can efciently utilize the available spectrum and
improve the overall network throughput.
Channel assignment approaches in traditional multi-channel
wireless networks typically try to select the best set of
channels for a given transmission (e.g., [2], [4], [7], [8])
1
.
We refer to this approach as the best multi-channels (BMC)
scheme. When BMC is employed in a CRN, the blocking
probability for CR transmissions can actually increase, leading
to a reduction in the network throughput. To illustrate, consider
an environment in which two PRNs and one CRN coexist.
PRN 1 operates at a low frequency band (CH 1), while PRN
2 operates at a high frequency band (CH 2). Suppose that
PRN 2 introduces a higher average PR-to-CR interference
(i.e., it has higher activity factor and transmission power).
Consequently, a CR receiver experiences a higher average
signal-to-interference-plus-noise ratio (SINR) over CH 1 than
over CH 2. Assume that two CR users A and C need to
send data to CR users B and D, respectively (see Figure 1).
Also assume that the distance between A and B is less than
that between C and D. Figure 1(a) shows that when the CR
users employ the BMC scheme, the transmission A B
uses CH 1, whereas the transmission C D uses CH 2.
The transmission A B is allowed to proceed because it
operates over a low carrier-frequency channel (CH 1) with low
PR-to-CR interference for a short transmission distance. On
the other hand, the transmission C D requires relatively
higher transmission power to overcome the high attenuation
associated with the high-frequency/high-interference channel
and the long transmission distance. If the required transmission
power exceeds the specied power mask, C D cannot
proceed. However, both A B and C D have much better
1
The best channel is often dened as the one that supports the highest rate.
978-1-4244-1777-3/08/$25.00 IEEE
This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE SECON 2008 proceedings.
10
chances of proceeding simultaneously if each CR transmitter
selects channels while keeping in mind the constraining power
mask on the other transmitter (Figure 1(b)).
A B
C
D
CH 1
CH 2
X
(a) BMC channel assignment
A B
C
D
CH 2
CH 1
(b) Distance-dependent chan-
nel assignment
Fig. 1. Scenarios in which two CR transmissions can/cannot proceed
simultaneously.
It is worth mentioning that in a given (one-hop) neighbor-
hood, the optimal channel assignment that maximizes the total
number of simultaneous CR transmissions can be structured as
an integer linear programming (ILP) problem. Due to the high
(exponential) complexity of computing the optimal solution for
this ILP [9], [10], heuristic algorithms that attempt to compute
suboptimal assignments with reasonable computational com-
plexities are needed.
In this work, we develop a novel CSMA-based MAC protocol
that aims at enhancing the throughput of the CRN subject to
a power mask constraint. The proposed protocol (DDMAC)
employs an intelligent heuristic stochastic channel assignment
scheme that exploits the dependence between the RF signals
attenuation model and the transmission distance while consider-
ing the prevailing local trafc conditions. The proposed channel
assignment scheme accounts for the interference conditions and
the power constraints at different PRN bands. In particular, the
proposed scheme assigns channels with lower average SINR
to shorter transmission distances, and vice versa. In addition,
our scheme assigns more preferable channels to the most
frequent transmission distances and less preferable channels to
the less frequent ones. We propose two variants of the channel
assignment mechanism in DDMAC. The rst variant is suitable
for static networks with known trafc patterns, while the other
variant is for dynamic networks with unknown trafc patterns.
The second variant employs a stochastic learning technique that
adapts to network dynamics. DDMAC has several attractive
features. First, it does not make any assumptions about the
activities of the underlying networks or about user distributions.
Second, the operation of DDMAC is simple and thus its
processing overhead is very small. This makes it suitable
for practical implementations. In addition, DDMAC does not
require any coordination with PR users. Finally, under low load
and many available channels, DDMAC gracefully degrades into
the BMC scheme.
In our performance evaluation, we conduct simulations over
a dynamic CRN with mobile users. Our simulation results
show that distance and trafc awareness signicantly improves
network throughput. The results also indicate that compared
with typical multi-channel CSMA-based protocols, the pro-
posed DDMAC decreases the connection blocking rate in a
CRN by up to 30%. By injecting articial errors into the
estimated distances, our evaluation reveals that DDMAC is
robust against estimation errors.
The rest of the paper is organized as follows. Section II
overviews related work. In Section III-A, we introduce our
system model and state our assumptions. The average SINR
for CRN is discussed in Section III-B. Section III-C illus-
trates the effect of the carrier frequency and transmission
distance on the path loss. In Section IV, we formulate the
optimal channel assignment problem. Section V introduces
our proposed distance-aware trafc-aware channel assignment
algorithm. Section VI describes the proposed DDMAC protocol
and outlines its benets and overhead. We evaluate DDMAC
in Section VII. Finally, Section VIII gives concluding remarks.
II. RELATED WORK
Recently, several attempts were made to develop MAC
protocols for CRNs (e.g., [7], [11][17]). In [7], the authors
developed a CRN MAC protocol with a common control chan-
nel. This protocol jointly optimizes the channel/power/rate as-
signment, assuming a given power mask on CR transmissions.
DC-MAC [11] is a cross-layer distributed scheme for spectrum
allocation/sensing. It provides an optimization framework based
on partially observable Markov decision processes, assuming
that PR and CR users share the same slotted transmission
structure. In [12], the authors investigated continuous-time
Markov models for dynamic spectrum access in open spectrum
wireless networks. Using such models and the homo-egualis
anthropological model, they proposed a distributed random
access protocol that is shown to achieve airtime fairness.
The FCC dened the interference temperature model [18],
which provides a metric for measuring the interference expe-
rienced by licensed receivers. In [13], the authors studied the
issue of spectrum sharing among a group of spread-spectrum
users subject to constrains on the SINR and on the interference
temperature. In [14], the interference temperature model was
used for optimal selection of spectrum and transmission powers
for CR users. AS-MAC [16] is a spectrum-sharing protocol
for a CRN that coexists with a GSM network. In AS-MAC,
the GSM network is assumed to provide input to the CRN
over a broadcast channel. Each CR user selects channels based
on the CRNs control packets and information from the GSM
network. Thus, explicit coordination with the PRNs is required.
In [17], the authors proposed a decentralized channel-sharing
mechanism for CRNs based on a game-theoretic approach for
both cooperative and non-cooperative scenarios.
The above protocols were designed without exploiting the
dependence of the number of allowable CR transmissions on
the carrier frequency and the transmission distance. In addition,
these protocols do not account for the trafc and PR-to-
CR interference conditions. They are limited to the analytical
aspects of MAC design, with no complete operational details.
To the best of our knowledge, DDMAC is the rst CRN
MAC protocol that aims at improving the CRN throughput by
exploiting the dependence on the RF signals attenuation model
and the transmission distance while considering the prevailing
trafc and interference conditions.
III. PRELIMINARIES
A. Network Model
We consider a decentralized CRN that coexists geograph-
ically with M different PRNs. PR users are legacy radios
that cannot be controlled by the CRN. Figure 2 shows a
11
Fig. 2. Example of an opportunistic CRN that coexists with two PRNs.
Fig. 3. Operating spectrum in the hybrid network.
conceptual view of the scenario under consideration. The PRNs
are licensed to operate over non-overlapping frequency bands.
We assume that all the PRN bands have the same bandwidth
(B). In reality, a PRN may occupy multiple, non-contiguous,
unequal frequency bands. Such a PRN can be easily represented
in our setup by using multiple equal-bandwidth virtual PRNs,
each operating over its own carrier frequency. For the ith PRN,
we denote its carrier frequency by f
i
. As shown in Figure 3, the
available bandwidth (B) of a PRN is divided into L adjacent
but non-overlapping frequency channels of Fourier bandwidth
W (in Hz). Such L channels are collectively referred to as a
band. Let N denote the total number of channels in all bands;
N = LM.
Without loss of generality, we assume that B is sufcient
to support at least one CR transmission. This is an acceptable
assumption in many wireless systems that are built to operate
in the unlicensed bands, including IEEE 802.11/a/b/g-compliant
devices. Each CR user is equipped with n
t
radio transceivers
n
t
L, that can be used simultaneously. In theory, a CR
user can transmit over an arbitrary segment of the available
bandwidth by using tunable lters. In practice, however, a
CR typically implements a bank of xed lters, each tuned
to a given carrier frequency with xed bandwidth, allowing
the CR user to choose from a xed number of channels. In
our setup, we assume the latter (more practical) capability,
which can be used to approximate the tunable lter scenario.
To avoid corrupting the transmissions of licensed users, a
mask is enforced on the transmission power of a CR user
over each band, i.e, P
(i)
t
P
(i)
mask
, i = 1, 2, . . . , M. The
determination of

P
mask
def
= P
(1)
mask
, P
(2)
mask
, . . . , P
(M)
mask
is
certainly an important topic, but is out of the scope of this
paper. Here, we simply assume that

P
mask
is given. A CR
user transmits data to other CR users using the maximum
allowable power vector

P
mask
. When not transmitting, a CR
user is capable of measuring the total noise-plus-interference
2
I
(i)
over all bands i = 1, 2, . . . , M. This requires the CR
user to have a wideband sensing capability with a narrowband
resolution. The technology to support such capability is readily
available through a wideband antenna, a power amplier, adap-
tive lters, and a DSP technique called cyclostationary feature
detection [6], [19], [20]. Thus, a CR user can simultaneously
sense several GHz-wide bands and estimate the instantaneous
interference over each band [20]. Alternatively, a sequential par-
tial sensing approach can be employed at the cost of negligible
switching/sensing overhead [19], [21]. It is worth mentioning
that off-the-shelf wireless cards (e.g., ICS-572 products [22])
can readily serve as a fully functional wideband multi-channel
CR interface. Such an interface enables a CR user to perform
analysis of the RF spectrum (i.e., sensing) in real time.
B. Analysis of the Average SINR
Based on the aforementioned characteristics of the CRN,
the average measured SINR (SINR) over band i is mainly
determined by: (1) the path loss associated with that band
(P
L
(f
i
)); (2) the average value of the measured I
(i)
over that
band (I
(i)
), which can be estimated based on the sensing history
and the spectrum occupancy statistics [11], [23]; and (3) the
enforced power mask P
(i)
mask
. Formally, SINR
(i)
(dB) is given
by:
SINR
(i)
(dB) = P
(i)
mask
(dB) P
L
(f
i
)(dB) I
(i)
(dB). (1)
C. Carrier Frequency and Distance Effects on Path Loss
In this section, we discuss the effect of the carrier frequency
and transmission distance on the path loss. For a given carrier
frequency f, let d
o
(f) be the close-in distance, i.e., the distance
from the transmitter after which the RF channel can be approx-
imated by the free-space model; d
o
(f) can be determined from
measurements or can be estimated by [24]:
d
o
(f) = max
_
2D
2
f
c
, D,
c
f
_
(2)
where D is the antenna length of the transmitter and c is the
speed of light. Let P
o
(f) denote the received power at the close-
in distance. Then, P
o
(f) can be estimated as follows [24]:
P
o
(f) =
c
2
G
t
(f)G
r
(f)
(4d
o
(f))
2
f
2
P
t
(f) (3)
where G
t
(f) and G
r
(f) are the transmit and receive antenna
gains, respectively. Let P
r
(f) denote the received power at
distance d from the transmitter, d d
o
(f). Then,
P
r
(f) = P
o
(f)
_
d
o
(f)
d
_
n
(4)
where n is the path loss exponent (typically, 2 n 6).
2
The quantity I
(i)
includes the PR-to-CR interference and the thermal noise.
12
Using (2), (3), and (4), the path loss P
L
(f) can be expressed
as:
P
L
(f) = 10 log
P
t
(f)
P
r
(f)
= 10
_

_
log
c
2
D
n2
f
2
d
n
, f s.t. D max
_
c
f
,
2D
2
f
c
_
log
c
n

f
n
d
n
, f s.t.
c
f
max
_
D,
2D
2
f
c
_
log
c
4n
(2D
2
)
n2
f
4n
d
n
, f s.t.
2D
2
f
c
max
_
D,
c
f
_
(5)
where

def
=
G
t
(f)G
r
(f)
(4)
2
. (6)
Note that the dependence of the path loss on d (i.e.,
1
d
n
) is the
same, irrespective of the carrier frequency.
Figure 4 depicts the path loss for a wide range of carrier
frequencies and two values of n at a distance d = 1 meter.
This gure and equation (5) reveal that the signal attenuation
increases as the distance between two communicating users in-
creases, and as the frequency used for communication increases.
These observations provide the motivation for our distance-
dependant channel assignment, discussed in section V.
0 1 2 3 4 5 6
x 10
9
20
30
40
50
60
70
f (Hz)
P
a
t
h

L
o
s
s

(
d
B
)
n = 4
n = 2
Fig. 4. Path loss vs. carrier frequency for two path loss exponents (D = 5
cm, Gt(f) = Gr(f) = 1).
IV. FORMULATION OF OPTIMAL CHANNEL ASSIGNMENT
Our objective is to maximize the number of simultaneous CR
transmissions, and consequently the overall network through-
put, by the means of optimal channel assignment. Toward
this end, we dene the term local spectrum utilization as
the total number of simultaneous CR transmissions that can
be supported in a given (one-hop) locality while meeting a
predened power mask. Before formulating the problem, we
discuss the requirements of a successful CR transmission.
A. CRN Transmission Requirements
Within a given neighborhood, multiple CR users may con-
tend for access to one or more of the available channels. Let
^ and denote the set of all N channels, and the set of all
CR transmission requests in the local neighborhood at a given
time, respectively. We say the jth CR transmission (j ) is
successful if both of the following two conditions are met:
It is possible to nd m
j
available channels from the set
^ such that

mj
i=1
r
(i)
j
R
j
, where r
(i)
j
is the data rate
of the ith selected channel and R
j
is the total rate demand
for the jth CR transmission.
Let /
j
be the set of m
j
selected channels. Then, the
received SINR of every i /
j
(SINR
(i)
j
) must be greater
than the SINR threshold (

i
) that is required at the CR
receiver to achieve a target bit error rate over channel i.
B. Maximizing the Local Spectrum Utilization
Let
(i)
j
be a binary variable denoting whether or not channel
i is assigned for transmission j. Formally,

(i)
j
=
_
1, if channel i is assigned for transmission j
0, Otherwise.
(7)
Similar to [10], [25], the problem of maximizing the total
number of simultaneous CR transmissions in a given neighbor-
hood can be formally stated as follows:
max

(i)
j
{0,1}

jJ
1
_

iN

(i)
j
r
(i)
j
R
j
_
(8)

jJ

(i)
j
1, i ^ (9)

iN

(i)
j
n
t
, j (10)
SINR
(i)
j

i
, j , s.t.
(i)
j
= 1 (11)
where 1[.] is the indicator function. The constraint in (9)
ensures that a channel cannot be assigned to more than one CR
transmission in the same vicinity. The constraint in (10) ensures
that at most n
t
channels can be assigned to a CR transmission.
For an ad hoc CRN, the above optimization problem must
run in a distributed manner at each CR user in the network.
This implies that each CR user must exchange instantaneous
SINR and rate demand information with neighboring CR users
before selecting channels, which incurs high control overhead
and delay (i.e., information may not be up-to-date). Worse yet,
even with perfect knowledge of the SINR of each link and
the rate demands, the above ILP problem belongs to the class
of NP-hard problems [9]. In our paper, we develop a heuristic
channel assignment scheme that provides a suboptimal solution
with lower complexity while still achieving good spectrum
utilization.
Our heuristic exploits distance and trafc awareness. The key
idea behind it is to assign channels with low SINR to short-
distance transmissions. Also, local trafc information is used
to assign more channels to more likely transmission distances.
V. DISTANCE-DEPENDENT CHANNEL ASSIGNMENT
ALGORITHM
In this section, we describe our proposed channel assignment
mechanism. The assignment process identies a preferable
channel list for each CR user. Such a list indicates which
channels are preferable to use depending on the estimated
distance between the transmitter and the receiver. Two variants
of the channel assignment mechanism are proposed. The rst
variant is suitable for static networks with known trafc pat-
terns, whereas the second one is for dynamic (mobile) networks
with unknown trafc patterns.
A. Assignment for a Static CRN with Known Trafc Patterns
Given s CR user with a packet to transmit, let r be the
distance to the intended receiver; r r
c
, where r
c
is the
maximum transmission range
3
. Let F
R
(r)
def
= PrR r. The
3
This is the largest distance from a CR transmitter over which the transmis-
sion at maximum power can be correctly decoded over all selected channels
in the absence of interference from other terminals (CR or PR users).
13
functional form of F
R
depends on both the distribution of users
as well as the trafc prole, which we now assume to be given.
Given F
R
, the channel assignment process is conducted as
follows:
The available bands are divided according to their mea-
sured SINR (given in (1))
4
into M sets S
1
, S
2
, . . . , S
M
,
where each band consists of multiple channels. The set S
1
contains the frequency channels of the band that has the
highest SINR, S
2
contains the next highest SINR, and so
on.
A CR user, say A, divides its maximum transmis-
sion region R
c
def
= r
2
c
into M non-overlapping rings
R
1
, . . . , R
M
. The ith ring contains the CR users whose
distances to A fall in (r
i1
, r
i
], where i = 1, . . . , M and
0 = r
0
r
1
r
2
. . . r
M
= r
c
. The rings are
divided such that the probability of communicating with
a CR receiver that falls within any of the M rings is the
same, i.e.,
F
R
(r
i
) F
R
(r
i1
) =
1
M
, i = 1, . . . , M. (12)
User A computes the radii r
i
, i = 1, . . . M, by substituting
for F
R
(r
i
) in (12) and solving for r
i
.
Finally, A constructs a preferable channel list for each
ring by assigning channels with lower SINR to shorter
transmission distances and channels with higher SINR
to longer transmission distances, i.e., assign S
M
to R
1
,
S
M1
to R
2
, . . ., and S
1
to R
M
.
To illustrate the idea, we consider a uniformly distributed
CRN. We assume that a CR transmitter chooses a destination
for its data randomly within R
c
. Therefore, F
R
(r) is given by:
F
R
(r) =
_
r
2
r
2
c
, r r
c
1, r r
c
. (13)
Using (12) and (13), and noting that r
0
= 0, we arrive at the
following expression for r
i
:
r
i
=

_
1
M
+
r
2
i1
r
2
c
_
r
c
=
_
i
M
r
c
. (14)
Figure 5 illustrates the non-overlapping rings around a CR
transmitter for M = 4. Within these rings, other CR and PR
users exist.
Fig. 5. Different rings for assigning channels around a CR transmitter (M =
4).
4
Note that P
L
s dependence on d is the same for all bands. Thus, for the
purpose of SINR comparison, we set d = 1 meter.
B. Assignment for a Dynamic CRN with Unknown Trafc
Patterns
In the previous analysis, we assumed a xed network and
prior knowledge of the trafc pattern (i.e., the form of F
R
). In
practice, the trafc pattern may change with time, depending
on the network dynamics and user mobility. Because users
can only possess local knowledge of their neighborhoods, it is
hard to reach and maintain the optimal network performance.
Nevertheless, we can develop a stochastic learning algorithm
that performs well with only localized information. Stochastic
learning techniques have been widely used in wireless networks
for online trafc prediction, tracking, and power control [26],
[27]. Our proposed learning approach is a distributed algorithm
that runs at each CR user in the network. A CR user, say A,
evenly divides its maximum transmission region R
c
into mnon-
overlapping regions, where m M. The ith region, D
i
, forms
a ring, dened by the area (x, y) : d
2
i1
< x
2
+ y
2
d
2
i
,
where d
i
= i
rc
m
, and d
i1
< d
i
i = 1, . . . , m. CR user
A maintains an m-entry transmission distance table. The ith
entry in that table corresponds to the region D
i
. It contains
the number of overheard CR packet requests during the recent
observation window T
win
for which the transmitter-receiver
distances fall in the range(d
i1
, d
i
]. Note that the proper setting
of T
win
depends on the dynamics of the network (the effect of
T
win
is studied in Section VII).
Fig. 6. Time diagram of pmfs updating process.
To initialize the assignment algorithm, all CR users employ
the BMC scheme discussed in Section I. At any time t, CR
user A constructs its transmission distance table based on
control packets that have been overheard during the observation
window [t T
win
, t]. Using the transmission distance table,
A estimates the current probability mass function p
i
(t) of
the distance r at time t (see Figure 6). It then computes an
exponentially weighted average of p
i
(t) :
p
i
(t) = p
i
(t) + (1 ) p
i
(t T
win
) (15)
where is a forgetting factor, 0 < 1. Once p
i
(t) is
computed, A computes the preferable channel list for each ring.
Let
i
(A) denote the preferable channel list for ring D
i
at CR
user A (how to construct
i
(A) will be given later). The new
preferable channel lists will be used during the next observation
window time. The proposed channel assignment process merges
the D
i
s into K regions according to p
i
(t), where K M. It
then assigns preferable channels for each region. The process
is now described in detail:
1) User A chooses k such that [

k1
i=0
p
i
(t)

m
i=k
p
i
(t)[
is minimized, i.e., it divides the regions into two groups;
short-distance and long-distance groups. The probabilities
of the short-distance and long-distance groups are given
14
by:
P
short
=
k1

i=0
p
i
(t) (16)
and
P
long
=
m

i=k
p
i
(t). (17)
2) User A divides the M bands into two frequency sets:
low SINR frequency set and high SINR frequency set. It
assigns the low SINR frequency set to the short-distance
group and the high SINR frequency set to the long-
distance group. The numbers of bands in the high (n
H
)
and low (n
L
) frequency sets depend on P
short
and P
long
as follows:
n
H
=
_
P
short
P
short
+P
long
M
_
n
L
= M n
H
(18)
where x| is the smallest integer x.
3) Step 1 and 2 are repeated for every group until either
only one band is assigned to that group or the group
contains only one region. Note that when repeating the
above process for a group, m in (17) and M in (18) are
replaced with the number of regions in that group and the
number of channels assigned to that group, respectively.
By this recursive procedure, the preferable channel list

i
(A), for all i, is computed for one observation window.
C. Example
We illustrate the previously discussed channel assignment
process via a simple example. Consider four PRNs and one
CRN. Each PRN occupies two adjacent non-overlapping chan-
nels. The PRNs are labeled such that f
1
< f
2
< f
3
< f
4
.
Consider a CR user A with SINR
(1)
> SINR
(2)
> SINR
(3)
>
SINR
(4)
. Suppose that A divides its transmission region R
c
into 8 rings, D
1
, D
2
, . . . , D
8
. At a given time t, assume that
the weighted average pmf p
i
(t) : i = 1, . . . , 8 is given
by 0.25, 0.1, 0.15, 0.05, 0.05, 0.15, 0.05, 0.2. Figure 7 shows
how the proposed channel assignment process is conducted.
The outcome of this process is as follows:
Band 4, which includes two channels, is assigned to all CR
transmissions whose distances are in D
1
(i.e.,
1
(A) =
4).
Band 3, which includes two channels, is assigned to all
CR transmissions whose distances are in D
2
and D
3
(i.e.,

2
(A) =
3
(A) = 3).
Band 2, which includes two channels, is assigned to all
CR transmissions whose distances are in D
4
, D
5
, and D
6
(i.e.,
4
(A) =
5
(A) =
6
(A) = 2).
Band 1, which includes two channels, is assigned to all
CR transmissions whose distances are in D
7
and D
8
(i.e.,

7
(A) =
8
(A) = 1).
VI. DDMAC PROTOCOL
Based on the channel assignment process presented in Sec-
tion V, we now propose a distributed, asynchronous MAC
protocol for CRNs. The proposed DDMAC uses contention-
based handshaking for exchanging control information. We rst
state our main assumptions. Then, we describe our protocol in
detail.
D
i
1 0.25
2 0.1
3 0.15
4 0.05
5 0.05
6 0.15
7 0.05
8 0.2
0.5
0.5
0.25 {f4, f3}
{f4}
{f3}
0.25
{f2, f1}
0.25
0.25
{f2}
{f1}
i
p
~
Fig. 7. Example that illustrates the channel assignment process in a dynamic
CRN.
A. Assumptions
For each frequency channel, we assume that the channel
gain is stationary for the duration of a few control and one
data packet transmission periods. This assumption holds for
typical mobility patterns and transmission rates [28]. We as-
sume symmetric channel gains between two CR users. This is
a typical assumption in any RTS/CTS-based protocol, including
the IEEE 802.11 scheme. We also assume that a CR user
transmits data to other CR users at a xed rate using the
maximum allowable power vector (

P
mask
). Finally, we assume
the availability of a prespecied control channel with Fourier
bandwidth B
c
, where B
c
B. This channel need not to
be reserved for the CRN. It can, for example, be one of the
subchannels in an ISM band.
B. Channel Access Mechanism
The purpose of the channel access mechanism is to allow
the CR transmitter and receiver to agree on the set of channels
to use for communication and to determine the rate allocation
across these selected channels in a manner that ensures that
the power mask and the rate demand are met. A CR user A
views its transmission region as K non-overlapping regions,
where each region is associated with a preferable channel list

i
(A), i = 1, . . . , K, determined according to the discussion
in Section V. This user maintains an N-entry channel list and
an m-entry transmission distance table (as described in Section
V). The jth entry of the channel list indicates the status of the
jth channel. It maintains a value of 1 if the channel is available
and 0 if the channel is occupied or reserved by any of As CR
neighbors. Every CR user listens to the control channel, and
accordingly updates its channel list and transmission distance
table.
Suppose that CR user A has data to transmit to another CR
user B at a constant aggregate rate demand R
A
, which corre-
sponds to the aggregate rate for the A B communication
over all selected channels. Then, A reacts as follows:
If user A does not sense a carrier over the control channel
for a random duration of time, it sends an RTS message
at the maximum (known) power P
max
. This P
max
is con-
strained by the power mask imposed on the prespecied
control channel. The RTS includes R
A
and the list of all
available channels at A. The RTS message is sent even
if As channel list contains no free channels. In this case,
the purpose of the RTS is to help CR users predict the
network trafc pattern.
15
The neighbors of A other than B that can correctly decode
the RTS refrain from accessing the control channel until
they receive one of two possible control packets, denoted
by EPCA and ENCA (explained below).
Upon receiving the RTS packet, B uses the received signal
strength of the RTS to estimate the distance between
A and B (d
AB
). It identies the preferable channel list

i
(B) that corresponds to d
AB
. Based on the available
channels at A and B, and the instantaneous interference
level over these channels as measured at B, user B
removes any channel that has a received SINR less than
its threshold SINR and determines the common channel
list that is potentially available for A B transmis-
sion, denoted by CCL(A, B). User B then computes the
intersection between
i
(B) and CCL(A, B) to identify
the preferable set of available channels for the A B
communication ((A, B)). To achieve good throughput,
B sorts the channels in (A, B) in a descending order
of their maximum possible data rate (calculated accord-
ing to the Shannons formula) and then appends the
rest of the common available channels that are not in
(A, B)
_
i.e., CCL(A, B)

(A, B)
_
, also listed in a
descending order of their maximum possible data rate,
to the bottom of the sorted preferable channels.
5
Then,
B cummulatively adds channels from the top of the new
sorted list until either the aggregate rate R
A
is satised or
the list is exhausted (i.e., no feasible channel assignment
can be found).
User B determines whether or not there exists a feasible
set of channels that can support the aggregate rate R
A
.
If not, then B responds by sending a Negative-Channel-
Assignment (NCA) message that includes the distance
d
AB
. The purpose of this packet is to help Bs neighbors
estimate the network trafc pattern and prompt A to back
off and retransmit later. If B can nd a set of available
channels that can support a total demand R
A
, it sends a
Positive-Channel-Assignment (PCA) message to A, which
contains the assigned channels for the transmission A
B, the distance d
AB
, and the duration needed to hold the
assigned channels for the ensuing data transmission and
corresponding ACK packet. The PCA implicitly instructs
Bs CR neighbors to mark the set of assigned channels
as unavailable for the indicated transmission duration. It
also helps these neighbors in estimating the network trafc
pattern.
Depending on which control message is received, user A
reacts as follows:
If A receives an NCA message, it responds by sending
an Echo-NCA (ENCA) message, which includes the
distance d
AB
. The purpose of this packet is to help
As neighbors estimate the network trafc pattern.
If A receives a PCA message, it replies back with an
Echo-PCA (EPCA) message, informing its neighbors
of the selected channel list, the distance d
AB
, and the
transmission duration. This EPCA also announces the
success of the control packet exchange between A and
B to As neighbors, which may not have heard Bs
PCA.
Once the RTS/PCA/EPCA exchange is completed, the data
5
As demonstrated in [6], [7], minimizing the number of assigned channels
results in the best throughput performance.
transmission A B proceeds. Once completed, B sends
back an ACK packet to A over the best assigned channel
that has the highest rate.
It is worth mentioning that there is no interference between
data and control packet transmissions because the two are
separated in frequency. Therefore, a CR user that hears the
RTS packet from A defers its attempt to access the control
channel until it receives an EPCA or an ENCA packet from
A. In addition, a CR user that receives only a PCA or an
NCA should defer its attempt to access the control channel
until the expected time for the EPCA/ENCA packet expires
(to avoid collision of control packets). This allows for more
parallel transmissions to take place in the same neighborhood.
VII. PROTOCOL EVALUATION
We now evaluate the performance of the DDMAC protocol
and contrast it with the BMC approach. Our results are based
on simulation experiments conducted using CSIM programs
(CSIM is a C-based process-oriented discrete-event simulation
package [29]). For simplicity, data packets are assumed to be
of a xed size (2 Kbytes). Since the maximum transmission
ranges under the DDMAC and BMC protocols are the same, it
is safe to assume that both protocols achieve the same forward
progress per hop. Consequently, our performance metrics are
the one-hop throughput, i.e., the packet destination is restricted
to one hop from the source, and the connection blocking rate.
The latter metric is dened as the percentage of CR packet
requests that are blocked due to the unavailability of a feasible
channel assignment. The signal propagation model in (4) is
used with n = 4 and G
t
(f) = G
r
(f) = 1 for every carrier
frequency f.
A. Simulation Setup
We consider four PRNs and one CRN. Users in each PRN
are uniformly distributed over a 500 500 meters
2
area. The
PRNs operate in the 600 MHz, 900 MHz, 2.4 GHz, and 5.7
GHz bands, respectively. Each PRN band consists of three non-
overlapping 1-MHz channels. The number of PR users in each
PRN is 300. We divide the time into slots, each of length 3.3
ms. A time slot corresponds to the transmission of one CR
packet at a xed data rate (5 Mbps). In any given slot, each
PR user in the ith PRN attempts to transmit over its own band
with probability
i
. The probabilities for the four PRNs are
0.5, 0.3, 0.3, 0.1, respectively. The transmission power for each
PR user is 0.5 Watt, and the antenna length (D) is 5 cm.
For the CRN, we consider a random-grid topology, where
225 mobile CR users are placed within the 500 500 meters
2
eld. The eld is split into 225 smaller squares, one for each
CR user. The location of a mobile user within the small square
is selected randomly. For each generated packet, the destination
is selected randomly from the one-hop neighbors. Each CR user
generates packets according to a Poisson process with rate
(same for all users) and requires an aggregate transmission rate
of 5 Mbps. The random waypoint model is used for mobility,
with the speed of a CR user uniformly distributed between
0 and 2 meters/sec. We set the CRN SINR threshold to 5
dB and the thermal noise to P
(i)
th
= 10
21
Watt/Hz for all
channels. We assume that a CR user can use up to three
data channels simultaneously. We set the interference mask to
P
(1)
mask
= P
(2)
mask
= . . . = P
(12)
mask
= 50 mW, which results in a
maximum transmission range of r
c
= 75 meters. The reported
results are the average of 100 experiments.
16
0 0.1 0.2 0.3 0.4 0.5
0.3
0.35
0.4
0.45
Packet genaration rate
B
l
o
c
k
i
n
g

r
a
t
e

(
%
)


BMC
DDMAC
(a) Blocking rate vs. .
0 0.1 0.2 0.3 0.4 0.5
0
5
10
15
20
25
30
35
40
Packet generation rate
S
i
n
g
l
e

h
o
p

t
h
r
o
u
g
p
u
t

(
P
a
c
k
e
t
/
t
i
m
e

s
l
o
t
)
=0.6
BMC
DDMAC
(b) Throughput vs. .
0 0.1 0.2 0.3 0.4 0.5
0
0.02
0.04
0.06
0.08
0.1
0.12
Packet generation rate
S
i
n
g
l
e

h
o
p

t
h
r
o
u
g
h
p
u
t

(
p
a
c
k
e
t
/
t
i
m
e

s
l
o
t
)
BMC
DDMAC
(c) Per-user single-hop throughput.
Fig. 8. Performance of a CRN.
B. Results
We rst compare the performance of DDMAC to that of the
BMC scheme. We set the forgetting factor to = 0.6, the
observation window to T
win
= 0.5 second, and the number
of rings around a CR user to m = 12. For a fair comparison,
we let both schemes use the maximum allowable power vector

P
mask
. Figures 8(a) and (b) show that under moderate and
high trafc loads, DDMAC signicantly reduces the connection
blocking rate and improves the overall one-hop throughput by
up to 30%. This improvement is attributed to the increase in
the number of simultaneous transmissions due to the proper
channel assignment algorithm. Note that under low trafc load,
a minor reduction in the throughput of DDMAC may occur
compared with BMC. This is mostly attributed to the larger
control overhead of DDMAC.
In Figure 8(c), we focus on the performance of one CR user
under DDMAC (other CR users depicted similar behaviors).
This gure shows that, although DDMAC requires a pair of CR
users to communicate over a set of channels that may not be
optimal from one users perspective, the per-user throughput
of DDMAC under moderate and high trafc loads is greater
than that of the BMC scheme. This is attributed to the fact
that DDMAC attempts to serve a given CR transmission rst
using the preferable channel list and preserves the better
channels for other transmissions. However, if the aggregate rate
of this transmission cannot be satised using the preferable
list, DDMAC attempts to serve this transmission using the
remaining available channels.
The effect of dividing the CR users transmission range is de-
picted in Figure 9(a) for different values of . As m increases,
the throughput also increases up to a certain point. For m 12,
no signicant improvement is observed in the CRN throughput.
This is because our proposed channel assignment merges the
m regions into K m regions. Thus, increasing m beyond a
certain value (over-dividing R
c
) becomes unnecessary.
In Figure 9(b), we study the impact of and T
win
on the
performance of DDMAC. We set = 0.3 packet/slot. The
throughput versus for different values of T
win
is shown in
the gure. It is clear the throughput depends on the choice
of and T
win
. As T
win
increases, should increase to give
much more importance to recent observations without entirely
discarding older observations. Table I shows the best throughput
performance and the associated optimal value of (

) for
Scheme

Best throughput(packet/slot)
BMC - 25
DDMAC(Twin = 0.03 s) 0.1 26
DDMAC(Twin = 0.3 s) 0.6 33.6
DDMAC(Twin = 0.4 s) 0.6 33.85
DDMAC(Twin = 1 s) 0.8 33.89
DDMAC(Twin = 4 s) 1.0 28
TABLE I
PERFORMANCE OF DDMAC AT THE OPTIMAL AS A FUNCTION OF T
win
.
different values of T
win
. It is clear that if T
win
is too small or
too large, the throughput reduces signicantly.
5 10 15
20
25
30
35
40
m
S
i
n
g
l
e

h
o
p

t
h
r
o
u
h
p
u
t

(
P
a
c
k
e
t
/
t
i
m
e

s
l
o
t
)
= 0.2
= 0.25
= 0.3
(a) Throughput vs. number of rings
(m) around a CR user.
0.2 0.4 0.6 0.8 1
22
24
26
28
30
32
34
36
38
Forgitting factor
S
i
n
g
l
e

h
o
p

t
h
r
o
u
p
u
t

(
P
a
c
k
e
t
/
t
i
m
e

s
l
o
t
)
T
win
= 0.03 s
T
win
= 0.3 s
T
win
= 0.4 s
T
win
= 1 s
=0.3
(b) Throughput vs. for different
T
win
values.
Fig. 9. Performance of DDMAC.
Finally, we investigate the robustness of DDMAC to inac-
curate distance estimation, which mainly results from multi-
path propagation, reection, and fading effects. We introduce
uniform estimation errors ( Uniform[, ]) into the dis-
tance d. Thus, the estimated distance

d is given by

d
(1 +) d. Figure 10(a) shows the effect of inaccurate distance
estimation on the perceived throughput as a function of under
different trafc loads. As the gure indicates, there are no
signicant difference in the throughput for different values of .
17
0.05 0.1 0.15 0.2 0.25
0
5
10
15
20
25
30
35
40

S
i
n
g
l
e

h
o
p

t
h
r
o
u
g
h
p
u
t

(
P
a
c
k
e
t
/
t
i
m
e

s
l
o
t
)
Low load
Moderate load
High load
(a) Throughput vs. .
0 0.1 0.2 0.3 0.4 0.5
0
1
2
3
4
5
6
Packet generation rate
A
b
s
o
l
u
t
e

p
e
r
c
e
n
t
a
g
e

e
r
r
o
r
= 5%
= 15%
= 25%
(b) Relative error vs. .
Fig. 10. Impact of inaccurately estimating d on performance.
Furthermore, Figure 10(b) shows that the maximum throughput
reduction due to inaccurate d is less than 6%.
Our results suggest that the assignment is reasonably robust
against estimation errors. This robustness arises from the fact
that DDMAC does not need accurate distances but only the
rough user distributions and local trafc conditions in order
to dynamically adapt the channel assignment to transmission
distances and prevailing trafc conditions.
VIII. CONCLUSIONS
In this paper, we proposed a novel distance-dependent
MAC protocol for opportunistic CRNs, known as DDMAC.
DDMAC improves the CRN throughput through a proper
channel assignment process. We presented a heuristic stochas-
tic channel assignment scheme that dynamically exploits the
dependence between the frequencys signal attenuation model
and the transmission distance, while considering the prevailing
trafc conditions, to enhance the throughput. The proposed
scheme assigns channels with lower average SINR to shorter
transmission distances. We integrated the proposed channel
assignment process in the design of DDMAC. To the best of
our knowledge, DDMAC is the rst CRN MAC protocol that
utilizes the radio propagation characteristics to improve the
overall network throughput under interference mask constraints.
We compared the performance of DDMAC with an optimum
multi-channel MAC protocol that is designed for typical multi-
channel systems (BMC). We showed that, under moderate and
high trafc loads, DDMAC achieves about 30% increase in
throughput over the BMC scheme. Although DDMAC requires
a pair of CR users to communicate on a channel that may not be
optimal from a users perspective, we showed that the average
per-user throughput of DDMAC under moderate and high trafc
loads is greater than that of the BMC scheme. In summary,
our simulation results showed that DDMAC provides better
spectrum utilization in terms of smaller connection blocking
probability and larger system throughput.
REFERENCES
[1] FCC, spectrum policy task force report, ET docket no. 02-155, Nov.
2002.
[2] A. Nasipuri and S. Das, Multichannel CSMA with signal power-based
channel selection for multihop wireless networks, in Proceedings of
the IEEE Fall Vehicular Technology Conference (VTC), Sept. 2000, pp.
2428.
[3] H. Bany Salameh, T. Shu, and M. Krunz, Adaptive cross-layer MAC
design for improved energy efciency in multi-channel wireless sensor
networks, Ad Hoc Network, vol. 5, no. 6, pp. 844854, 2007.
[4] N. Jain, S. Das, and A. Nasipuri, A multichannel CSMA MAC protocol
with receiver-based channel selection for multihop wireless networks,
in Proceedings of the 9th Int. Conf. on Computer Communications and
Networks (IC3N), Oct. 2001, pp. 432439.
[5] M. Riback, J. Medbo, J. Berg, F. Harrysson, and H. Asplund, Carrier
frequency effects on path loss, in Proceedings of Vehicular Technology
Conference (VTC), May 2006, pp. 27172721.
[6] I. Akyildiz, W.-Y. Lee, M.C. Vuran, and S. Mohanty, Next generation
dynamic spectrum access cognitive radio wireless networks: A survey,
Computer Networks, vol. 50, no. 13, pp. 21272159, 2006.
[7] T. Shu, S. Cui, and M. Krunz, Medium access control for multi-channel
parallel transmission in cognitive radio networks, in Proceedings of the
IEEE GLOBECOM Conference, Nov. 2006.
[8] A. Nasipuri and S. Das, Performance of multi-channel ad hoc networks,
International Journal of Wireless and Mobile Computing, vol. 1, no. 3/4,
pp. 191203, 2006.
[9] E. Arikan, Some complexity results about packet radio networks, IEEE
Transactions on Information Theory, vol. 30, no. 4, pp. 681685, 1984.
[10] A. Behzad and I. Rubin, Multiple access protocol for power-controlled
wireless access nets, IEEE Transactions on Mobile Computing, vol. 3,
no. 4, pp. 307316, 2004.
[11] Q. Zhao, L. Tong, and A. Swami, Decentralized cognitive MAC
for dynamic spectrum access, in Proceedings of the IEEE DySPAN
Conference, Nov. 2005, pp. 224232.
[12] Y. Xing, R. Chandramouli, S. Mangold, and S. Shankar, Dynamic
spectrum access in open spectrum wireless networks, IEEE Journal
on Selected Areas in Communications, vol. 24, no. 3, pp. 626637, 2006.
[13] Y. Xing, C. Mathur, M. Haleem, R. Chandramouli, and K. Subbalakshmi,
Dynamic spectrum access with QoS and interference temperature con-
straints, IEEE Transactions on Mobile Computing, vol. 6, no. 4, pp.
423433, 2007.
[14] T. C. Clancy, Achievable capacity under the interference temperature
model, in Proceedings of the IEEE INFOCOM Conference, May 2007,
pp. 794802.
[15] R. Menon, R. Buehrer, and J. Reed, Outage probability based compari-
son of underlay and overlay spectrum sharing techniques, in Proceedings
of the IEEE DySPAN Conference, Nov. 2005, pp. 101109.
[16] S. Sankaranarayanan, P. Papadimitratos, A. Mishra, and S. Hershey, A
bandwidth sharing approach to improve licensed spectrum utilization, in
Proceedings of the IEEE DySPAN Conference, Nov. 2005, pp. 279288.
[17] N. Nie and C. Comaniciu, Adaptive channel allocation spectrum etiquette
for cognitive radio networks, in Proceedings of the IEEE DySPAN
Conference, Nov. 2005, pp. 269278.
[18] S. Haykin, Cognitive radio: Brain-empowered wireless communica-
tions, IEEE Journal on Selected Areas in Communications, vol. 23,
no. 2, pp. 201220, 2005.
[19] D. Cabric, S. Mishra, and R. Brodersen, Implementation issues in
spectrum sensing for cognitive radios, in Proceedings of the 38th
Asilomar Conference on Signals, Systems and Computers, Nov. 2004,
pp. 772776.
[20] D. Cabric and R. Brodersen, Physical layer design issues unique to
cognitive radio systems, in Proceedings of the IEEE International
Symposium on Personal, Indoor and Mobile Radio Communications
(PIMRC), Sept. 2005, pp. 759763.
[21] A. Sabharwal, A. Khoshnevis, and E. Knightly, Opportunistic spectral
usage: Bounds and a multi-band CSMA/CA protocol, IEEE/ACM
Transactions on Networking, vol. 15, no. 3, pp. 533545, 2007.
[22] Application Note (AN-SR-7), Building a versatile low latency cognitive
radio for multi-mission applications with the ICS-572, Tech. Rep., 2003,
http://www.gefanucembedded.com/news-events/whitepapers/3067.
[23] S. Jones, N. Merheb, and I. Wang, A cognitive MAC protocol using
statistical channel allocation for wireless ad-hoc networks, in Proceed-
ings of Wireless Communications and Networking Conference (WCNC),
March 2007, pp. 105110.
[24] T. S. Rappaport, Wireless Communications-Principles and Practice,
Prentice-Hall Press, 2001, 2nd edition.
[25] A. Hoang and Y. Liang, Maximizing spectrum utilization of cognitive
radio networks using channel allocation and power control, in Proceed-
ings of Vehicular Technology Conference (VTC), Sept. 2006, pp. 15.
[26] S. Kiran and R. Chandramouli, An adaptive energy-efcient link layer
protocol using stochastic learning control, in Proceedings of the IEEE
ICC Conference, May 2003, pp. 11141118.
[27] Y. Chang, T. Ho, and L. Kaelbling, Mobilized ad-hoc networks: A
reinforcement learning approach, in Proceedings of the International
Conference on Autonomic Computing, May 2004, pp. 240247.
[28] A. Muqattash and M. Krunz, POWMAC: A single-channel power control
protocol for throughput enhancement in wireless ad hoc networks, IEEE
Journal on Selected Areas in Communications, vol. 23, no. 5, pp. 1067
1084, May 2005.
[29] Mesquite Software Incorporation, www.mesquite.com.
18