Wireless Sensor Network, 2010, 2, 649-654
doi:10.4236/wsn.2010.28077 Published Online August 2010 (http://www.SciRP.org/journal/wsn)
Copyright © 2010 SciRes. WSN
A Spectrum Sensing Framework for UWB-Cognitive
Network
Deah J. Kadhim, Saba Q. Jobbar
Electrical Engineering Department, College of Engineering University of Baghdad, Baghdad, Iraq
E-mail: deya_naw@yahoo.com
Received May 30, 2010; revised June 17, 2010; accepted June 28, 2010
Abstract
Due to that the Ultra Wide Band (UWB) technology has some attractive features like robustness to multipath
fading, high data rate, low cost and low power consumption, it is widely use to implement cognitive radio
network. Intuitively, one of the most important tasks required for cognitive network is the spectrum sensing.
A framework for implementing spectrum sensing for UWB-Cognitive Network will be presented in this pa-
per. Since the information about primary licensed users are known to the cognitive radios then the best spec-
trum sensing scheme for UWB-cognitive network is the matched filter detection scheme. Simulation results
verified and demonstrated the using of matched filter spectrum sensing in cognitive radio network with
UWB and proved that the bit error rate for this detection scheme can be considered acceptable.
Keywords: Cognitive Radio, UWB, Spectrum Sensing, Matched Filter
1. Introduction
With respect to the FCC regulation, Ultra Wideband
(UWB) is a future technology for medium and short
scope of wireless networks with a different of throughput
choices containing very high data rates. The most im-
portant property of UWB is that it can presence in the
same temporal, spatial, and spectral domains with li-
censed/unlicensed users because it is an underlay system.
Other seductive features of UWB include the flexibility
of adapting pulse shape, bandwidth, data rate, and trans-
mit power. At the head of these characteristics, UWB has
low power consumption so the complexity of the trans-
ceiver will be reduced and this will lead to form a limited
system cost. Note that there is another important feature
of UWB represented by providing secure communica-
tions as a result detecting UWB transmission will be very
hard since the power spectrum will embed into the noise
floor. There are two commonly proposed means of im-
plementing UWB. These two technologies are the Or-
thogonal Frequency Division Multiplexing based UWB
(UWB-OFDM) and the impulse radio based UWB
(IR-UWB) [1,2].
Most of today's wireless networks are characterized by
a fixed spectrum assignment policy. Unfortunately, only
few frequency resources are currently available for future
applications. Recent measurements indicate that many
portions of licensed spectrum are not used for significant
periods of time. Cognitive radio is a new technology
used to improve the spectrum utilization by allowing
secondary users (cognitive radios) to borrow unused ra-
dio spectrum from primary users (licensed users) or to
share the spectrum with them. A network with a cogni-
tive process that observes its environment current condi-
tions and learn from these conditions in order to decide
and plan then make some adaptations for their work to
qualify its communication environment, this network
called cognitive radio network [3].
UWB communication systems can be presented as a
first example of technology that is coincidence for the
implementation of cognitive radio mechanism. UWB
systems are preferable because of their large bandwidth,
their low-power noise-like signaling, which can be ex-
ploited in the transmission over (licensed) bands pro-
ducing a controlled level of interference on existing com-
munication systems. Thus recent researches has been
focused on the investigation of coexistence issues related
to the UWB technology, assuming that such systems will
operate in an environment characterized by the presence
of heterogeneous interfering users. The essential com-
ponent of Ultra Wideband-based Cognitive Network
(UWB-CN) is the spectrum sensing in which a cognitive
radio user can find only an unused band of the spectrum,
so it should closely monitor the whole spectrum bands,
note their information, and then detect spectrum holes
[4,5].
D. J. KADHIM ET AL.
650
In this paper we will focus our work on implementing
the spectrum sensing in UWB-CN. Intuitively, we could
make a conclusions on using matched filter spectrum
sensing scheme to implement the spectrum sensing for
UWB-CN. This conclusion is that matched filter requires
prior information about the characteristics of primary
users’ signals and this information are known for Cogni-
tive radios in UWB-CN system so it is the optimal spec-
trum sensing scheme for UWB-CN system. Note that
matched filter is used for maximizing the signal to noise
ratio for a given input signal in presence of additive
white Gaussian noise.
This paper is organized as follows. In Section 2, we
describe the framework of spectrum sensing for UWB-
Cognitive network. Section 3, gives the description de-
tails of implementing matched filter detection. Section 4,
shows the presents our computer simulation results. The
conclusions are drawn in Section 5.
2. Spectrum Sensing Framework
To discover which spectrum sensing schemes is the most
efficient and suitable to the UWB-cognitive network
system, for this discovering we need to do the following
discussion; when the receiver cannot gather sufficient
information about primary users, the optimal detector
may be the energy detector. However, due to non-coherent
processing O(1/SNR2) samples are required to meet a
probability of detection constraint. Moreover, energy
detectors often generate false alarms announced by un-
wanted signals or multiple secondary users because they
cannot discriminate signal types.
Generally, modulated signals can be defined as the
signals which are characterized by a periodic and Cyc-
lostationary features. These features can be analyzed and
detected by using a spectral correlation function. The
attractive property of feature detection scheme is its
firmness to uncertainty in noise power. But the main
disadvantage of this scheme is the complex calculations
and requires great effort and more time. Moreover, it
needs additional bandwidth and subject to the radio fre-
quency, spectrum loss of strong signals and timing or
frequency jitters.
When the information of the primary user signal is
known to the CR user, the optimal detector in stationary
Gaussian noise is the matched filter detection. The main
advantage of matched filter is that due to coherency, it
requires less time to achieve high processing gain since
only O(1/SNR) samples are needed to meet a given
probability of detection constraint. However, the matched
filter requires a priori knowledge of the characteristics of
the primary user signal. On the other hand, matched filter
detection requires accurate time and frequency synchro-
nization. From what has been discussed above, we may
draw the conclusion that the matched filter detection is
the optimal spectrum sensing scheme in the UWB-CR
because the priori knowledge of the primary user signals
in the UWB-CR system is known to the CR user.
2.1. Matched Filter
A matched filter is a correlation function between a
known signal or template, with an unknown signal to de-
tect the presence of the template in the unknown signal
and it is equivalent to a convolution function between the
unknown signal with a time-reversed version of the tem-
plate. It is the optimal linear filter for maximizing the
signal to noise ratio (SNR) in the presence of additive
stochastic noise. Matched filters are commonly used in
radar, in which a signal is sent out, and we measure the
reflected signals, looking for something similar to what
was sent out [6].
A matched filter is a linear filter designed to maximize
the output signal-to-noise ratio for a given input signal.
Suppose that a signal s(t) plus additive white Gaussian
noise n(t) is input to an linear time-invariant filter fol-
lowed by a sampler, as shown in Figure 1 [7].
Figure 2 shows the characteristics of a matched filter.
The impulse response of the matched filter h(t) is a de-
layed version of the mirror image of s(t). In general, the
matched filter is not a physically realizable (causal) fil-
ter.
0
()( )htKst t
(1)
2.2. Matched Filter Correlator
If r (t) = s(t) + n(t) is the received signal to the input of a
causal matched filter, then the output of the filter can be
found by convolving the received signal r(t) with the
impulse response of the filter. Therefore;
00
0
()()()*()() ()
s
tnt rthtrhtd

 
(2)
Using Equation (2), then we get,
Received Signal
LTI Filter
H(f)
s(t) + n(t)
s
o
(t) + n
o
(t)
t = t
0
s
o
(t
0
) + n
o
(t
0
)
Figure 1. Matched filter presentation.
s(–t)
t
0
0
s(t)
0
t
0
h(t) = s(t
0
t)
0
t
0
Figure 2. Matched filter characteristics.
Copyright © 2010 SciRes. WSN
D. J. KADHIM ET AL.651
00
0
()()( )[()]
s
tnt Krstotd

 
(3)
At t = t0,
0
00 00
0
() ()()()
t
s
tnt Krsd


(4)
The above operation is known as the correlation of r(t)
and s(t). For this reason, the matched filter is often re-
ferred to as a correlator. Figure 3 shows the block dia-
gram of a correlator. It is an alternative way of synthe-
sizing a matched filter [7].
3. Matched Filter Detection Scheme
Since the information of the primary user signal is known
to the cognitive radio user at UWB-cognitive network,
the optimal detector in stationary Gaussian noise is the
matched filter since it maximizes the received sig-
nal-to-noise ratio (SNR). While the main advantage of
the matched filter is that it requires less time to achieve
high processing gain due to coherency, it requires a pri-
ori knowledge of the primary user signal such as the
modulation type and order, the pulse shape, and the
packet format. Hence, if this information is not accurate,
then the matched filter performs poorly. However, since
most wireless network systems have pilot, preambles,
synchronization word or spreading codes, these can be
used for the coherent detection.
The energy detector can sense any signals. It is espe-
cially suitable for sensing signals such as bursting inter-
ference signals, pulsed jamming and tone jamming,
which have distinctive power or energy features instead
of features in the frames or structures. The block diagram
of energy sensing is showed in Figure 4. The received
signal r(t) is filtered and changed into digital signal by
analog-to-digital (A/D) converters. After Fast Fourier
Transform (FFT), it is squared and averaged in the dwell
tine N. The result is decided by comparing the energy
0
0
t
s
o
(t
0
) + n
o
(t
0
r(t) = s(t) + n(t)
s(t)
Figure 3. Matched filter correlator.
r(t)
Input Output
Wide band
Filter
A/
D
FFT
Average
In N
Energy
Detection
Decision
Threshold
Figure 4. Energy detector.
detected with the decision threshold γ.
There are three keys in the energy detection, test sta-
tistic, decision rule and dwell time as well as threshold.
The detection operation is done as flowing [3,8];
1) Test Statistic, gives two hypotheses:
0:( )( ),0,1,...,1HXn nnnN
 (5)
1:( )( )( ),0,1,...,1HXnrn nnnN
  (6)
where r(n) is the received signal and n(n) is the noise, N
is the dwell time. The sensed energy can be written as,
12
0
() ()
N
n
Trrn
(7)
2) Decision Rule,
1
2
0
() (,)
n
H
Tr N
H
(8)
where γ is the decision threshold, 2
n
is the variance of
the noise.
3) Design Parameters, It is obvious that the decision
threshold γ is the function of N and2
n
. So that the per-
formance of the energy sensing is depended on the signal-
to-noise ratio (SNR) of the received signal and the dwell
time.
Almost all modulated signals have the cyclostationar-
ity in their carrier frequency, bit rate, cyclic prefix and so
on [9]. We also exploit the built in cyclostationarity of
the modulated signals to sense the signals. In Figure 5,
the received signal r(t) is filtered and changed into digital
signal by A/D converters. After FFT, the correlation is
computed and averaged in the dwell tine N. The decision
is made out at last.
The spectral correlation function is written as,
12/
0
1
(;)()()
Nv
j
fh L
n
Sfvrnrnve
N


(9)
where r(n) is the received signals. The coherent detection
in feature space is given by,
1*
0
()( ;)( ;)
L
vf
TrS fvSfv
 (10)
If the information of the primary user signal is known to
the CR user, the correlator in Figure 5 would be replaced
by a linear filter. This linear filter should be matched to the
primary user signal, as showed in Figure 6. The signal
processing is same as in feature detection instead of us-
ing correlator to calculate the correlation.
To be specific, let us suppose that the impulse response
of the matched filter is,
Copyright © 2010 SciRes. WSN
D. J. KADHIM ET AL.
652
r(t)
Input
Output
Wideband
Filter
A/D
FFT Correlator
Feature
Detection
Figure 5. Feature detector.
r(t)
Input
Wideband
Filter
A/D
FFT
3
Matched
Filter Output
Figure 6. Matched filter detection.
(1),0, 1,...,1
()
0,
rNn nN
hn
else
 
(11)
where N is the duration time of received signal. The out-
put of the matched filter is,
1
0
()()(),0,1,...,1
N
n
TjrnhjnjN

(12)
If we sample the output of the filter at j=N–1, we ob-
tain the maximum,
1
0
12
0
(1) ()(1)
()
N
n
N
n
T NrnhnN
rn
 
(13)
From (7) and (13), it is obvious that the maximum of
the matched filter is the same as in the energy detection
but the performance of the matched filter is independent
of the dwell time.
4. Simulation Results
In the following simulation, we will investigate the sig-
nal detection process by match filter and we will study
the performance of match filter by measuring the prob-
ability of bit error. In this simulation, we will generate
two signal to present r(t) function as a transmitter, first
one is a rectangle pulse with unit amplitude and the sec-
ond signal is a triangle pulse with 1 msec pulse width
with unit peak amplitude. Then noise is added to these
two unmodulated signals as shown in Figure 7.
Now these noisy unmodulated signals are filtered us-
ing matched filter at receiving side, we will use for this
filtering a rectangle matched filter and a triangle matched
filter to show their behavior with two type of unmodu-
lated signals as shown in Figure 8 below.
The above simulations supposed unmodulated signals,
in the other hand let us suppose BPSK modulated signals
are sent at the transmitter for pulse train with unit ampli-
tude, Figure 9 shows the pulse train signal before and
after modulation with BPSK at a transmitter side with
additive noise. At the receiver side the matched filter
Figure 7. Noisy rectangle and triangle pulses.
Figure 8. Rectangle and triangle filters.
filtered the modulated signal as shown in Figure 10.
Now to measure the performance of matched filtered
used above, let us using the derivation of the Bit Error
Rate (BER) for BPSK modulated signal which is pro-
vided at [10]. The binary digits 1 and 0 at BPSK are rep-
resented by the analog levels and
bb
EErespec-
tively. Using [10] we have the following BER function
for BPSK modulated signal,
0
1(
2
b
b
E
Perfc
N
)
(14)
where erfc(.), is the complementary error function. Fig-
ure 11 simulates the BER for BPSK modulated signal.
This simulation is done by generating of random BPSK
signal symbols +1’s and –1’s, then passing them through
additive Gaussian noise channel. Demodulation of the
received symbol based on the location in the constella-
tion, is occurred after that counting the number of errors,
Copyright © 2010 SciRes. WSN
D. J. KADHIM ET AL.
Copyright © 2010 SciRes. WSN
653
Figure 9. Pulse train before and after modulation in time and frequency domains.
Figure 10. Receiver matched filter output in time and frequency domains.
then repeating the same procedure for multiple Eb/N0
value.
some of the main requirements for implementing cogni-
tive radio networks, these features such as avoidance the
interference with licensed primary users, dynamically
adaptable bandwidth, data rate and transmitted power.
The second reason is concerned with the objective issues
that can come with implementing UWB technology in
cognitive radio networks, these objective issues are such
as locating cognitive nodes via UWB, exchanging the
sensing information between cognitive nodes using UWB
5. Conclusions
UWB technology implementation plays a great role in
cognitive radio networks; this conclusion comes from
two main reasons. The first reason is the features of
UWB which are presented as a good chance to achieve
D. J. KADHIM ET AL.
654
-2 0246810
10
-5
10
-4
10
-3
10
-2
10
-1
Eb/No, dB
Bit Error Rate
E
b
/N
0
, dB
Figure 11. Bit error probability curve for BFSK modulation.
and the receiving sensitivity of the nodes in the network
that has an integral role in determining the range and size
of communication networks.
Simulation results showed that the matched filter de-
tection scheme is a suitable for detecting signals through
UWB-cognitive radio network especially all information
required for sensing primary users are known for cogni-
tive radios. However, in the UWB system, the architec-
ture of the UWB signals is known, but the band location
and occupancy level is unknown. Thus, new techniques
are required to measure or estimate the precise locations
of primary users at nearby cognitive radios or secondary
users.
6. References
[1] D. L. Sostanovsky and A. O. Boryssenko, “Experimental
UWB Sensing and Communication System,” IEEE Aero-
space and Electronic Systems Magazine, Vol. 21, No. 2,
February 2006, pp. 27-29.
[2] K. Wallace, B. Parr, B. Cho and Z. Ding, “Performance
Analysis of A Spectrally Compliant Ultra-Wideband
Pulse Design,” IEEE Transactions on Wireless Com-
munications, Vol. 4, September 2005, pp. 2172-2181.
[3] D. Cabric, S. M. Mishra and R. W. Brodersen, “Impleme-
ntation Issues in Spectrum Sensing for Cognitive Radios,”
Proceedings of 38th Asilomar Conference on Signals,
Systems and Computers, Pacific Grove, November 2004,
pp. 772-776.
[4] H. Arslan and M. E. Sahin, “Ultra Wideband Wireless
Communication, ch. Narrowband Interference Issues in
Ultrawideband Systems,” Wiley, Hoboken, September
2006.
[5] S. Zhang and Z. Bao, “Strategy of Anti-Interference
Based on Cognitive Radio in Ultra-Wideband System,”
4th International Conference on Wireless Communi-
cations, Dalian, 2008.
[6] I. Akyildiz, W. Lee, M. Vuran and S. Mohanty, “Next
Generation/Dynamic Spectrum Access/Cognitive Radio
Wireless Networks: A Survey,” Computer Networks, Vol.
50, No. 13, September 2006, pp. 2127-2159.
[7] I. Akyildiz, W. Lee, M. Vuran and S. Mohanty, “A
Survey on Spectrum Management in Cognitive Radio
Networks,” IEEE Communications Magazine, Georgia
Institute of Technology, Atlanta, April 2008.
[8] H. Urkowitz, “Energy Detection of Unknown Deterministic
Signals,” Proceedings of IEEE, Vol. 55, April 1967, pp.
523-531.
[9] J. Lund´en, V. Koivunen, A. Huttunen and V. Poor,
“Spectrum Sensing in Cognitive Radios Based on Multiple
Cyclic Frequencies,” Signal processing Laboratory, Helsinki
Univercity of Technology, Helsinki, July 2007.
[10] J. G. Proakis, “Digital Communications,” 4th Edition,
McGraw-Hill, New York, 2001.
Copyright © 2010 SciRes. WSN