You are on page 1of 8

Int J Wireless Inf Networks (2011) 18:171–178

DOI 10.1007/s10776-011-0148-y

Opportunistic Spectrum Access Using Fuzzy Logic for Cognitive


Radio Networks
Hong-Sam T. Le • Hung D. Ly • Qilian Liang

Received: 27 March 2009 / Accepted: 15 April 2011 / Published online: 15 May 2011
 Springer Science+Business Media, LLC 2011

Abstract Recent studies and measurements have shown The output of the FLS provide the possibility of accessing
that, with the traditional spectrum access approach, the spectrum band for secondary users and the user with the
radio spectrum assigned to primary (licensed) users is greatest possibility will be assigned the available spectrum
vastly underutilized. While many spectrum methods have band.
been proposed to use spectrum effectively, the opportu-
nistic spectrum access has become the most viable Keywords Cognitive radio  Fuzzy logic system 
approach to achieve near-optimal spectrum utilization by Opportunistic spectrum access
allowing secondary (unlicensed) users to sense and access
available spectrum opportunistically. Opportunistic spec-
trum access approach is enabled by cognitive radios which 1 Introduction
are able to sense the unused spectrum and adapt their
operating characteristics to the real-time environment. Recent studies and measurements have shown that, with
However, a naive spectrum access for secondary users can the traditional spectrum access approach, the radio spec-
make spectrum utilization inefficient and increase inter- trum assigned to primary (licensed) users is vastly under-
ference to adjacent users. In this paper, we propose a novel utilized while the demand for access to the limited radio
approach using Fuzzy Logic System (FLS) to control the spectrum have been growing dramatically. This view is
spectrum access. Three descriptors are used: spectrum supported by actual measurements conducted by the FCC’s
utilization efficiency of the secondary user, its degree of Spectrum Policy Task Force which has determined that, in
mobility, and its distance to the primary user. The linguistic some locations or at some times of a day, about 70% of the
knowledge of spectrum access based on these three allocated spectrum may not be in use [1]. Measurements in
descriptors is obtained from a group of network experts. 27 [2] reveal that spectrum utilization is often heavy in unli-
fuzzy rules are set up based on this linguistic knowledge. censed bands while low in TV bands or medium in some
cellular bands. These observations on actual spectrum
usage have challenged approaches to the radio spectrum
management and fueled interests in the opportunistic
H.-S. T. Le spectrum access problem.
Department of Telecommunications Engineering, Posts and Opportunistic spectrum access has been enabled by
Telecommunications Institute of Technology, Hanoi, Vietnam cognitive radios (CRs). Unlike conventional radios, CRs
have the capability to sense their surroundings and actively
H. D. Ly (&)
Department of Electrical and Computer Engineering, adapt their operation mode to maximize the quality of
Texas A&M University, College Station, TX, USA service for secondary users while minimizing interference
e-mail: hungly@tamu.edu to primary users. Hence, CRs must carry out spectrum
sensing to identify white spaces or spectrum holes which
Q. Liang
Department of Electrical Engineering, The University of Texas are bands of frequencies assigned to primary users, but, at a
at Arlington, Arlington, TX, USA particular time and specific geographic location, these

123
172 Int J Wireless Inf Networks (2011) 18:171–178

bands are not being utilized by those users [3]. Some 2 Preliminaries
methods on spectrum sensing has been proposed in [4, 5],
and [14]. Once spectrum holes are identified, CRs 2.1 Cognitive Radios
opportunistically utilize these holes for communication
without causing interference to primary users. Assume Cognitive Radios have been seen as the key technology that
that a spectrum band is available for secondary users. If enables cognitive radio networks or xG networks to use
only one secondary user, in a particular location and at a spectrum efficiently by allowing secondary (cognitive) users
specific time, can sense this available spectrum, this to sense and utilize available spectrum opportunistically.
secondary user can use this band right after the primary Cognitive radios have two intrinsic characteristics [3]:
user finishes the communication session. What will hap-
• Cognitive capability: Cognitive capability implies the
pen, however, and which secondary user will be chosen to
ability of cognitive radios to sense information from
use the available band if multiple secondary users try to
their surroundings in order to figure out spectrum
access the spectrum? Of course, for the former question,
portions that are unused at a specific time or location.
since these secondary users have the same rights to access
The most suitable portion will be selected for commu-
the spectrum, they have to compete with each other in a
nication without causing interference to other users.
collaborative and fair manner. This paper will give more
• Reconfigurability: Reconfigurability enables cognitive
detail answer for these questions and propose a novel
radios to be dynamically reprogrammed according to the
approach using Fuzzy Logic System (FLS), an artificial
real environment. This means that cognitive radios can
intelligence system which is capable of making real time
change the operating frequency, modulation scheme,
decisions, to decide the suitable secondary user which
transmission power, communication protocol, etc. on the
will use the available band.
fly without any modification of hardware components.
In research literature on opportunistic spectrum access,
some work uses game theoretical analysis [6] to find Cognitive radios, in order to use spectrum opportunis-
strategies for spectrum sharing. In [7], spectrum allocation tically, experience four main procedures, i.e.,
using a graph coloring algorithm is proposed but mobility
1. Spectrum sensing: A cognitive radio monitors spec-
of the secondary users is not considered. Moreover,
trum bands and detects unused bands, i.e., spectrum
authors assumed that if two secondary users within dis-
holes which are time-varying and location-dependent.
tance of each other use the same spectrum band, they fail
Cognitive radio can use the spectrum sensing tech-
to access spectrum. With this approach, some secondary
niques such as transmitter detection, cooperative
users will lose the rights to compete for using spectrum
detection, and interference-based detection.
and monitoring secondary users conflicting in using
2. Spectrum access: Assume that multiple cognitive users
spectrum band is also a challenging issue. In our
trying to use the spectrum coexist in an area. This
approach, we use the rule-based FLS to assign the
procedure is used to prevent multiple users from
available spectrum to secondary users efficiently and
colliding in overlapping spectrum portions.
guarantee that the secondary user using assigned band
3. Communication: Once a cognitive radio is assigned a
will not interfere with the primary users. To achieve these
spectrum band for communication, it will inform its
objectives, we use three descriptors which are spectrum
receiver about the chosen band. After the receiver-
utilization efficiency of the secondary user, its degree of
transmitter handshake procedure is completed, the
mobility, and its distance to the primary user. The lin-
cognitive radio begins receiving and/or transmitting
guistic knowledge of spectrum access based on these
information.
three descriptors is obtained from a group of network
4. Spectrum mobility: A cognitive radio must move to
experts. 27 fuzzy rules are set up based on this linguistic
another spectrum hole to keep doing communication
knowledge. The output of the FLS provide the possibility
once it detects the signal from the primary user. Hence,
of each secondary user which will be assigned spectrum
spectrum mobility occurs when a cognitive radio
band and the user with the greatest possibility will be
change its operating band.
assigned the available spectrum band.
The rest of this paper is organized as follows. In Sect. 2, we There have been many open research problems to develop
briefly introduce the fuzzy logic system and cognitive radios. these procedures. In this paper, we use the fuzzy logic
Opportunistic spectrum access using the FLS which is based system, i.e., an optimization technique, to give solutions for
on experiences from a group of network experts is proposed the opportunistic spectrum access problem. The most
in Sect. 3. In Sect. 4, we discuss the simulation results. suitable secondary user having the rights to access the
Conclusions and future works are presented in Sect. 5. spectrum is chosen based on three descriptors, i.e., spectrum

123
Int J Wireless Inf Networks (2011) 18:171–178 173

FUZZY LOGIC SYSTEM


Expert knowledge for selecting the best suitable secondary
user to access the available band is collected based on the
RULES following three antecedents, i.e., descriptors:
CRISP CRISP
INPUT OUTPUT 1. Antecedent 1: Spectrum utilization efficiency.
FUZZIFIER DEFUZZIFIER
x X y f (x) Y 2. Antecedent 2: Degree of mobility.
INFERENCE
3. Antecedent 3: Distance to the primary user.
FUZZY INPUT FUZZY OUTPUT
SETS SETS
Generally, the secondary user with the furthest distance
to the primary user or the secondary user with maximum
Fig. 1 The structure of a Fuzzy Logic System spectrum utilization efficiency can be chosen to access
usage efficiency of the secondary user, its degree of spectrum under the constraint that no interference is cre-
mobility, and its distance to the primary user. ated for the primary user. In our approach, using the rule-
based FLS, we combine the above three descriptors to find
2.2 Fuzzy Logic Systems optimal solutions to assign spectrum opportunistically. We
see that different users will perceive different available
Figrue1 shows the structure of a fuzzy logic system (FLS). spectrum and using spectrum efficiently is the main pur-
When an input is applied to a FLS, the inference engine pose of the opportunistic spectrum access schemes. Hence,
computes the output set corresponding to each rule. The spectrum utilization efficiency gs is introduced in our
defuzzifier then computes a crisp output from these rule design. gs is defined as the ratio between the spectrum band
output sets. Consider a p-input 1-output FLS, using sin- which will be used by the secondary user and the available
gleton fuzzification, center-of-sets defuzzification and band, i.e.,
‘‘IF-THEN’’ rules of the form [8] BWs
gs ¼  100% ð3Þ
l
R : IF x1 is Fl1 and x2 is Fl2
and . . . and xp is Flp ; BWa
THEN y is Gl where BWs and BWa are the spectrum band which will be
Assuming singleton fuzzification is used, when an input used by the secondary and the available band, respectively.
x0 ¼ fx01 ; x02 ; . . .; x0p g is applied, the degree of firing Mobility of the secondary user plays an important role in
corresponding to the lth rule is computed as our design. When the secondary user is moving at a
velocity v m/s, it causes the Doppler effect.
lF1l ðx01 ÞHlF2l ðx02 ÞH    HlFpl ðx0p Þ ¼ T pi¼1 lFil ðx0i Þ ð1Þ
vcosh
fD ¼ fc ð4Þ
where H and T both indicate the chosen t-norm. There are c
many kinds of defuzzifiers. In this paper, we focus, for where fD is the Doppler shift, h is the arrival angle of the
illustrative purposes, on the center-of-sets defuzzifier. It received signal relative to the direction of motion, c is the
computes a crisp output for the FLS by first computing the wave velocity, and fc is carrier frequency. Mobility can
centroid, cGl ; of every consequent set Gl, and, then computing a reduce capability of detecting signal from the primary
weighted average of these centroids. The weight corresponding users. If the secondary user is not capable of detecting the
to the lth rule consequent centroid is the degree of firing primary signal, it will incorrectly determine that the spec-
associated with the lth rule, T pi¼1 lFil ðx0i Þ; so that trum is unused; thereby leading to potential interference to
PM p 0 adjacent users, i.e., the signal transmitted by the secondary
0 l¼1 cGl T i¼1 lFil ðxi Þ
ycos ðx Þ ¼ PM p ð2Þ user will interfere with the signal that the primary user is
0
l¼1 T i¼1 lFil ðxi Þ trying to decode. This situation is often referred as the
hidden node problem.
where M is the number of rules in the FLS.
Besides, we also consider the distance from the sec-
ondary user to the primary user. Actually, the location of
3 Knowledge Processing and Opportunistic Spectrum the primary users is unknown. We can consider signal-to-
Access noise ratio (SNR) as a proxy for distance [9]. Assume the
primary user at the distance R from the secondary user
3.1 Designing the Fuzzy Logic System transmits signal at power P1 and the power gain between
for Opportunistic Spectrum Access the primary user and secondary user, g(R), is a continuous,
nonnegative, strictly decreasing function of R defined on
We design a fuzzy logic system to solve the opportunistic the interval ½0; 1: SNR at the secondary user, cs, is given
spectrum access problem in cognitive radio networks. by

123
174 Int J Wireless Inf Networks (2011) 18:171–178

  1.5
P1 gðRÞ
cs ¼ 10log ð5Þ
r21
where P1 is the transmit power of the primary user and r21
is noise power measured at the secondary user. From (5), Low Moderate High
1
we can derive the distance R between the primary user and
the secondary user.
The linguistic variances used to represent the spectrum
utilization efficiency and degree of mobility are divided
into three levels: low, moderate, and high while we use 3 0.5
levels, i.e., near, moderate, and far to represent the dis-
tance. The consequence, i.e., the possibility that the sec-
ondary user is chosen to access the spectrum is divided into
five levels which are very low, low, medium, high and very
high. We use trapezoidal membership functions (MFs) to 0
0 10 20 30 40 50 60 70 80 90 100
represent near, low, far, high, very low and very high, and (a)
triangle MFs to represent moderate, low, medium and high. 1.5
MFs are shown in Fig. 2. Since we have 3 antecedents and
3 fuzzy subsets, we need set up 33 = 27 rules for this FLS.
Then, we design questions, which will be used in our
survey, according to rules as follows: Low, Near Moderate High, Far

IF the spectrum utilization efficiency of the secondary 1

user is moderate, its degree of mobility is low, and its


distance to the primary user is far THEN the possibility
that this user is selected to access the spectrum is .
0.5
3.2 Knowledge Processing and Opportunistic Spectrum
Access

As pointed out in [10], ‘‘words mean different things to


different people’’, and in [11], ‘‘the decision makers may 0
0 1 2 3 4 5 6 7 8 9 10
have the same preferences to a particular alternative, e.g., (b)
highly preferred but with different degrees;’’ so, we created
1.5
one survey for the network experts. We used rules obtained
from the knowledge of 5 network experts. These experts
were requested to choose a consequent, using one of the
five linguistic variables. Different experts gave different Very Low Low Medium High Very High

answers to the questions in the survey. Table 1 summarizes 1


the questions used in this survey. As an example, we also
give an expert’s answer in this table. Since we chose a
single consequent for each rule to form a rule base, we
averaged the centroids of all the responses for each rule and
0.5
used this average in place of the rule consequent centroid.
Doing this leads to rules that have the following form:
Rl: IF spectrum utilization efficiency of the secondary (x1)
is F1l, and its degree of mobility(x2) is F2l, and its distance to
the primary user (x3) is F3l, THEN the possibility (y) that this 0
0 10 20 30 40 50 60 70 80 90 100
secondary user is chosen to access the available spectrum is
clavg. where l ¼ 1; 2; . . .; 27 and clavg is defined as follows: (c)
P5 Fig. 2 The membership functions (MF) used to represent the
l wli ci
cavg ¼ Pi¼1 5
ð6Þ linguistic labels: a MF for antecedent 1, b MFs for other antecedents,
l
i¼1 wi c MF for consequence

123
Int J Wireless Inf Networks (2011) 18:171–178 175

Table 1 Questions for opportunistic spectrum access problem in Table 2 cavg Corresponding to each rule
cognitive radio networks
Rule # cavg
Rule # Antecedent 1 Antecedent 2 Antecedent 3 Consequence
1 17.222
1 Low Low Near Very low 2 23.611
2 Low Low Moderate Low 3 43.333
3 Low Low Far Low 4 23.611
4 Low Moderate Near Very low 5 30
5 Low Moderate Moderate Low 6 43.333
6 Low Moderate Far Medium 7 10.833
7 Low High Near Very low 8 23.611
8 Low High Moderate Low 9 36.667
9 Low High Far Medium 10 30.278
10 Moderate Low Near Very low 11 56.667
11 Moderate Low Moderate Medium 12 76.389
12 Moderate Low Far High 13 23.889
13 Moderate Moderate Near Very low 14 43.333
14 Moderate Moderate Moderate Medium 15 63.33
15 Moderate Moderate Far High 16 17.22
16 Moderate High Near Very low 17 30.278
17 Moderate High Moderate Low 18 56.667
18 Moderate High Far High 19 50
19 High Low Near Low 20 70
20 High Low Moderate High 21 89.167
21 High Low Far Very high 22 36.667
22 High Moderate Near Low 23 63.333
23 High Moderate Moderate High 24 82.778
24 High Moderate Far Very high 25 30.287
25 High High Near Very low 26 56.667
26 High High Moderate High 27 63.333
27 High High Far High

4 Simulation Results and Discussion


in which wli is the number of experts choosing linguistic label
i To validate our approach, we randomly generated 20 sec-
i for the consequence of rule l and c is the centroid of the ith
ondary users over an area of 100 9 100 meters. The pri-
consequence set (i ¼ 1; 2; . . .; 5; l ¼ 1; 2; . . .; 27). Table 2
mary user was placed randomly in this area. Three
provides cavg for each rule from the completed survey.
descriptors were randomly generated for each secondary
For every input (x1, x2, x3), the output y(x1, x2, x3) of the
user. More specifically, the spectrum utilization efficiency
designed FLS is computed as
P27 of each secondary user was a random value in the interval
l
l¼1 lF l ðx1 ÞlF2l ðx2 ÞlF3l ðx3 Þcavg [0,100], and its mobility degree in [0,10]. Distances to the
yðx1 ; x2 ; x3 Þ ¼ P27 1 ð7Þ
primary users were normalized to [0,10].
l¼1 lF1l ðx1 ÞlF2l ðx2 ÞlF3l ðx3 Þ
di
We recognize that (7) can be represented in a 4-D surface. Di ¼ ð8Þ
max20 i¼1 fdi g
Since it is impossible to plot visually, we fix one of three qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
variables. More specifically, we fixed the distance to the di ¼ ðxi  xp Þ2 þ ðyi  yp Þ2 ð9Þ
primary user x3. Two cases, i.e., x3 = 1 and x3 = 9, were
considered. Figure 3 represents the opportunistic spectrum where (xp, yp) and (xi, yi) represent the coordinate of the
access decision surface for the cognitive user for these two primary user and the ith secondary user ði ¼ 1; 2; . . .; 20Þ:
cases. From Fig. 3, we see clearly that, at the same spec- The values of descriptors corresponding to each sec-
trum utilization efficiency and mobility degree, secondary ondary user were passed through the FLS. The output of
users further from the primary user have higher chance to the FLS, i.e., the possibility that a secondary user was
access the spectrum. selected to access the available spectrum, was computed as

123
176 Int J Wireless Inf Networks (2011) 18:171–178

Table 3 Three descriptors and possibility for four secondary users


Parameters SU1 SU2 SU3 SU4
55
50 Mobility degree 2.4966 3.0715 6.5382 0.9135
45
40
Distance to PU 8.4852 3.0036 10 1.6437
y(x ,x ,1)

35 Possibility 82.2944 54.5189 45.3072 53.1432


1 2

30
Spectrum usage efficiency (%) 88.7104 97.9340 24.2160 92.4424
25
20
15
10 100
10
8 100 90
6 80
4 60
x2 40 80
2 20 x1
0 0
70
(a)
60

50

90 40

80
30
70
y(x ,x2,9)

20
60
1

10
50

40 0
0 10 20 30 40 50 60 70 80 90
30
10
8
Fig. 4 An opportunistic spectrum access scenario in a specific space
100
6 80 and a particular time: SU1, SU2, SU3, and SU4 are denoted using
60
x2 4
40 H; 5; ; and }; respectively. The primary user is denoted using h
2 x1
20
0 0
(b)
Fig. 3 The opportunistic spectrum access decision surface for the Guard Symbols Unique Word Payload Guard Symbols
cognitive user with a fixed distance to the primary user: a when the (3 symbols) (40 symbols) (500 symbols) (3 symbols)
distance to the primary user x3 = 1, b when the distance to the
primary user x3 = 9 Fig. 5 Burst structure

in (7). Then, the secondary user with the highest possibility


would be chosen to access the spectrum. At a particular and Doppler shift fd = 41.6100, 51.1917, 108.9700, and
time, values of three descriptors and possibility for four 15.2250 Hz for SU1, SU2, SU3, and SU4, respectively.
secondary users, i.e., the secondary user chosen to access QPSK modulation scheme was used at transmitter side and
the available spectrum (SU1), the secondary user with the block phase estimation algorithm proposed in [13] was
highest spectrum utilization efficiency (SU2), the second- applied at receiver side. Burst structure is depicted in Fig. 5
ary user having the furthest distance to the primary user with 546 QPSK symbols (500 symbols payload, 40 sym-
(SU3), and the secondary user with the lowest mobility bols Unique Word, and 6 symbols guards) were used.
degree (SU4) are listed in Table 3. Position of these users Performance of secondary users in term of mobility
is shown in Fig. 4. We see that, in Table 3, SU2 has the degree and Eb/N0 is depicted in Fig. 6. Based on Fig. 6, we
highest spectrum utilization efficiency with 97.93%, SU4 note that, at the same Eb/N0, SU4 with the lowest Doppler
with 92.4424% while SU1 only achieves 88.7104%. shift can achieve the best performance while SU3 with the
Although SU3 is the secondary user with the furthest dis- highest Doppler shift has the worst performance. Perfor-
tance, it has the lowest spectrum utilization efficiency and mance of SU1 and SU2 is similar since there is a small
highest mobility degree. difference between Doppler shifts. From above results, we
We also ran the Monte-Carlo simulation to analyze the can confirm that spectrum access decision is tradeoffs
performance of secondary users with different mobility among three descriptors chosen to design the FLS. There-
degrees (see Table 3). In this simulation, we used the fore, the secondary user with the highest spectrum utili-
Rician flat fading channel with fading factor K = 12 dB zation or the secondary user furthest from the primary user

123
Int J Wireless Inf Networks (2011) 18:171–178 177

SU1 represent the opportunistic spectrum access decision sur-


SU2
SU3 face. An opportunistic spectrum access scenario was
SU4
10
−1 analyzed and simulated to validate our approach. In our
approach, moreover, we can modify the membership
functions of descriptors in accordance to requirements of
the primary network and the spectrum using policy. Hence,
BER

our approach is promising to be implemented practically in


future cognitive radio networks.
We see that it is better if the secondary user uses mul-
tiple bands simultaneously for transmission since multi-
−2
band transmission provide better performance than single
10
band transmission during the spectrum handoff [14]. This
means that if a primary user returns to utilize a specific
band, the secondary user must vacate this band. Since
1 2 3 4 5 6 7
transmission is still continuing in other bands, quality of
Eb/N0 [dB]
service (QoS) degradation can be mitigated. Using our
scheme, multiple spectrum band decision for a secondary
Fig. 6 Performance analysis of secondary users in term of mobility
degree and Eb/N0 user is also obtained. However, some secondary users try to
get as much spectrum bands as possible and they can keep
is not guaranteed to access the spectrum. The secondary some of these bands for future transmission when the
user selected to access spectrum has the distance to PU spectrum handoff occurs. By this way, spectrum utilization
8.4852, spectrum utilization efficiency 88.7104% and is not efficient. So, it is important to investigate some
mobility degree 2.4966. solutions to prevent some users from using spectrum
Until now, someone may have two more questions: (1) ineffectively and solve the mobility management problem
who will decide the spectrum access rights for the sec- in order to keep a high QoS of cognitive radio networks.
ondary users? (2) if there are N users competing for
M spectrum bands (N [[ M), how can we control the
spectrum access? Since we use the centralized spectrum
sharing architecture, a centralized entity such as base sta- References
tions in cognitive wireless networks or clusterheads in
1. Spectrum policy task force report, Technical report 02-135,
sensor networks collects information about three descrip- Federal communications commission, Nov. 2002.
tors and available spectrum bands from secondary users 2. M. McHenry, Report on spectrum occupancy measurements,
through a common control channel and builds a spectrum http://www.sharedspectrum.com/
map. Then, it uses our designed FLS to control the spec- 3. S. Haykin, Cognitive radio: Brain-empowered wireless commu-
nication, IEEE Journal on Selected Areas in Communications,
trum assignment and access procedures in order to prevent Vol. 23, No. 5, pp. 201–220, 2005.
multiple users from colliding in overlapping spectrum 4. R. Tandra and A. Sahai, Fundamental limits on detection in low
portions. In case, N users competing for M spectrum bands, SNR under noise uncertainty, International Conference on
the centralized processor also takes advantage of our FLS Wireless Networks, Communications and Mobile Computing,
Vol. 1, pp. 464–469, 2005.
for each band to allow the best secondary user to access
5. B. Wild and K. Ramchandran, Detecting primary receivers for
each spectrum. cognitive radio applications, IEEE Proc. on DySPAN 2005,
pp. 124–130, 2005.
6. N. Nie and C. Comaniciu, Adaptive channel allocation spectrum
atiquette for cognitive radio networks, IEEE Pro. on DySPAN
5 Conclusion and Future Works
2005, pp. 269–278, 2005.
7. H. Zheng and C. Peng, Collabraton and fairness in opportunistic
We propose a novel approach using the rule-based fuzzy spectrum access, IEEE International Conference on Communi-
logic system to control the opportunistic spectrum access cations (ICC) 2005.
8. J.M. Mendel, Uncertainty rule-based fuzzy logic systems,
for secondary users in cognitive radio networks. The sec-
Prentice-Hall, Upper Saddle Rever, NJ, 2001.
ondary user is selected based on three descriptors, i.e., 9. Hoven, N. and Sahai, A., Power scaling for cognitive radio,
spectrum utilization efficiency of the secondary user, its International Conference on Wireless Networks, Communica-
degree of mobility, and its distance to the primary user. The tions and Mobile Computing, Vol. 1, pp. 250–255, 2005.
10. J. M. Mendel, Computing with words when words can mean
linguistic knowledge of spectrum access is based on
different things to different people, Int’l ICSC Congress on
experiences from a group of network experts, so that an Computational Intelligence: Methods and Applications, NY,
acceptable decision can be obtained. As a result, we 1999.

123
178 Int J Wireless Inf Networks (2011) 18:171–178

11. Marimin, et al., Linguistic labels for expressing fuzzy preference Qilian Liang is a Professor at
relations in fuzzy group decisio making, IEEE Trans. on Systems, the Department of Electrical
Man, and Cybernetics—Part B: Cybernetics, Vol. 28, No. 2, Engineering, University of
pp. 205–218, 1998. Texas at Arlington. He received
12. Q. Liang, Clusterhead election in mobile ad hoc networks, IEEE the B.S. degree from Wuhan
Proc.on Personal, Indoor and Mobile Radio Communications, University in 1993, M.S. degree
Vol. 2, pp. 1623–1628, 2003. from Beijing University of Posts
13. A.J. Viterbi and A.M. Virterbi, Nonlinear estimation of PSK- and Telecommunications in
modulated carrier phase with application to burst digital trans- 1996, and Ph.D degree from
mission, IEEE Trans. on Information Theory, Vol. 29, No. 4, University of Southern Califor-
pp. 543–551, 1983. nia (USC) in May 2000, all in
14. Ian F. Akyildiz, et al., NeXt generation/dynamic spectrum Electrical Engineering. Prior to
access/cognitive radio wireless networks: A survey, Computer joining the faculty of the Uni-
Networks Journal (Elsevier), Vol. 50, pp. 2127–2159, 2006. versity of Texas at Arlington in
August 2002, he was a Member
of Technical Staff in Hughes Network Systems Inc at San Diego,
California. His research interests include compressive sensing, radar
Author Biographies sensor networks, wireless sensor networks, wireless communications,
communication system and communication theory, signal processing
Hong-Sam T. Le received her for communications, fuzzy logic systems and applications, etc.
B.S. degree in Electronics and Dr. Liang has published more than 170 journal and conference papers,
Telecommunications Engineer- 7 book chapters, and has 6 U.S. patents pending. He received 2002
ing from Posts and Telecommu- IEEE Transactions on Fuzzy Systems Outstanding Paper Award, 2003
nications Institute of Technology, U.S. Office of Naval Research (ONR) Young Investigator Award,
Hanoi, Vietnam in 2003 and her 2005 UTA College of Engineering Outstanding Young Faculty
M.S. degree in Electrical Engi- Award, and 2007, 2009, 2010 U.S. Air Force Summer Faculty
neering from the University of Fellowship Program Award.
Texas at Arlington in 2007. Her
research interests are in the
areas of wireless communica-
tion and signal processing.

Hung D. Ly received the B.S.


degree in Electronics and Tele-
communications Engineering
from Posts and Telecommuni-
cations Institute of Technology,
Hanoi, Vietnam in 2002, and the
M.S. degree in Electrical Engi-
neering from the University of
Texas at Arlington in 2007. He is
currently pursuing the Ph.D.
degree in Electrical and Com-
puter Engineering at Texas
A&M University. His research
interests include information
theory, wireless communication,
and signal processing.

123

You might also like