See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330901131

Introduction to Cognitive Radio Systems

Technical Report · July 2008

DOI: 10.13140/RG.2.2.19607.83363

CITATIONS READS
6 4,141

1 author:

Ahmad M. Rateb
Huaiyin Institute of Technology
13 PUBLICATIONS   92 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Ahmad M. Rateb on 06 February 2019.

The user has requested enhancement of the downloaded file.

UNIVERSITI TEKNOLOGI MALAYSIA

Technical report

Introduction to Cognitive Radio Systems

Ahmad M. Rateb

Department of Telematics and Optical Engineering

Faculty of Electrical Engineering
UTM
2008
 ABSTRACT

With today’s ambition for more wireless devices and systems, the need for spectral resources to
accommodate these systems grows everyday, but the remaining spectrum resources seems to be
running out. Cognitive radio technology offers a revolutionary solution to this crisis by looking
at the spectrum in an unconventional manner as a multi-dimensional space. The current wireless
communication systems are governed by fixed spectrum assignment policy, which was proven
by measurement to be inefficient. Cognitive radio enables maximum spectrum utilization
efficiency through opportunistic access to temporarily unutilized portions of spectrum. This
report presents a general introduction to cognitive radio technology; it provides a description of
its basics, structure, and challenges. It discussed the main ideas, concepts and definitions
introduced for this technology, and it explains its main tasks, functions, components and
architecture. Moreover, it describes the cross-layer cooperation of each cognitive function.
Finally, it presents some noticeable ideas proposed for the realization of cognitive radio
functions.

2
Contents
ABSTRACT…...…………………………………………..….………………………...…...……2
1. Introduction ………………………………………………………………………………….6
2. The structure of cognitive radio ...……………...………………………………….……… 8
3. Concepts and definitions introduced for Cognitive Radio ………………………………..9
3.1. Primary/Secondary users/networks ……………………………………………………...9
3.2. Spectrum holes ………………………………………………………………………….10
3.3. Spectrum pooling………………………………………………………………………. 10
3.4. Interference temperature ………………………………………………………………..11
3.5. Multi-dimensional channels in multi-dimensional spectrum space……………………..12
4. Cognitive radio tasks and functions ……………………………………………………....15
4.1. Cognitive capability & Reconfigurability ……………………………………………...15
4.2. Cognitive functions ……………………………………………………………………..15
4.2.1. Spectrum sensing ………………………………………………………………..16
A. Non-Cooperative & cooperative detection ………………………………….19
B. Transmitter detection ………………………………………………………..20
▪ Matched filter detection ……………………………………………..20
▪ Energy detection …………………………………………………….21
▪ Pilot detection ……………………………………………………….22
▪ Cyclostationary feature detection …………………………………...23
C. Interference-based detection ………………………………………………...24
4.2.2. Spectrum management …………………………………………………………..25
A. Spectrum analysis …………………………………………………………...25
B. Spectrum decision …………………………………………………………...25
4.2.3. Spectrum mobility.……………………………………………………………….26
4.2.4. Spectrum sharing....…………...…………………..……………………………..26
5. Cognitive radio system architecture ………………………………………………………27
5.1. CR network elements …………………………………………………………………...27
5.2. CR network access ……………………………………………………………………...28
5.3. The need for a cross layer design.……………………………....……………………....28

3
 5.4. Cross layer operation in cognitive radio functions ……………………………………..29
6. Novel ideas proposed for CR technology …………………………………………………30
6.1. Neural networks for efficient pattern recognition………………………………………30
6.2. Game theory for spectrum sharing……………………………………………………...31
6.3. MB-OFDM as the proposed Physical layer technology for CR………………………...32
6.4. MIMO for high flexibility-spectrum efficient transmission…………………………….34
6.5. spatial filtering for spectrum sensing……………………………………………………34
6.6. Turbo processing for higher QoS……………………………………………………….35
6.7. Deep Spectrum notching using data specific tones for MB-OFDM-based systems……35
7. Conclusions………………………………………………………………………………….37
References……………………………………………………………………………………….39

4
List of Figures

Figure 1: Measured spectrum utilization (0-6 GHz) in downtown Berkeley, USA…………... ..6
Figure 2: SDR/CR transceiver…………………………………………………………………...9
Figure 3: Interference temperature model…………………………………………………… ..11
Figure 4: Spectrum hole distribution on a 2D spectrum space at two instants of time ………..13
Figure 5: 3D spectral space and Virtual cube model for spectral unit ………………………...14
Figure 6: Cognitive cycle ……………………………………………………………………...16
Figure 7: 3D spectrum for CDMA-based systems ………………………………………….....17
Figure 8: Classification of spectrum sensing techniques ……………………………………...18
Figure 9: A schematic example of a binary allocation vector in a MB-OFDM-based
Cognitive radio system……………………………………………………...……….19
Figure 10: Energy detector implementation …………………………………………………….21
Figure 11: Spectrum picture at the input of sensing radio for primary user signal Detection ….23
Figure 12: Curves comparing the probability of detection performance for Energy,
preamble (pilot), and combined detectors for a packet based
communication system ……………………………………………………………...23
Figure 13: CR network communication functionalities.………....……………………………...29
Figure 14: OFDM technique………………………………………………………………….....33
Figure 15: Spatial filtering……………………………………………………………………. ..35
Figure 16: Spectrum notching…………………………………………………………………...37

5
1. Introduction
The exponentially growing demand for higher rate wireless communications to satisfy the
greed for more applications has come to a bottleneck state since the spectral resources are
running out; there are almost no more frequency bands to allocate to new systems.
This can be confirmed if we take a look at the FCC spectrum inventory table [1], where we will
find that for the range of (0-6 GHz), which is the most popularly used band for commercial
wireless communication systems, there is a very limited available spectrum bands for installing
new systems with a broad bandwidth. This led to belief that there is a spectrum availability crisis.
This wrong belief is because we look at the spectrum as a huge band of frequencies that we
distribute over different systems using a fixed spectrum assignment policy, but what if we look at
the spectrum from the time respective for the same frequency range (0-6 GHz) as shown in
Figure 1. We will find that the band utilization in urban areas ranges from less than 1% to
slightly more than 50% as measured experimentally in [2], and it is even less in suburban and
rural areas. This indicates that a huge spectrum opportunity is available for new generation of
wireless communication systems if these systems can be able to be aware enough of the
spectrum in order to recognize and utilize these spectrum opportunities in dimensions other than
frequency.

Figure 1: Measured spectrum utilization (0-6 GHz) in downtown Berkeley, USA [2]

6
 Hence, it is true we have a spectrum scarcity crisis but only in the frequency dimension, but on
other dimensions such as time and space, we actually have frequency abundance. However, the
current communication technologies do not have the ability to recognize such spectrum
opportunities. For this reason, a new generation of spectrum-aware communication technology
called “cognitive radios” was first proposed in [3].
A cognitive radio is a smart device that merges powerful techniques from taken from artificial
intelligence, statistical signal processing , wideband/multiple antenna techniques, game theory,
information theory and others together to generate an exceptionally dynamic behavior that will
lead to the emergence of a new generation of intelligent wireless dynamic systems for the next
generation (xG) wireless communication networks.
Cognitive radios have the ability to analyze the surrounding radio spectrum to locate spectrum
access opportunities, use a smart technique based on cumulative learning to decide which
opportunity to use, in a manner that does not cause the surrounding systems any harm. A
cognitive radio is highly flexible in a manner that allows it to coexist with any communication
system, adapt to any channel condition by varying its operation parameters such as modulation
scheme and channel coding to achieve the highest possible quality of service.
Cognitive radio is considered as a spectrum sharing technology like ultra-wide-band [4]. The
key difference is that while the UWB signal spectrum overlaps with the primary user signal
spectrum by transmitting very low power levels that hardly affect the primary users, a cognitive
radio transmits over temporarily-unused spectrum segments defined as spectrum holes [5] in a
flexible opportunistic manner. As a result, a cognitive radio device may transmit high signal
power and hence, have a longer range than an UWB device, as long as it can correctly identify
the unused spectrum segments and the spectral harmful interference in the primary user spectrum
bands.
The report is organized as follows: Section 2 briefly introduces the structure of a cognitive
radio, which is mainly a software-defined radio (SDR). Section 3 explains the main concepts and
definitions related to this technology. Section 4 discusses the main cognitive tasks and functions
comprising the closed cognitive cycle. In section 5, the cognitive radio architecture is explained,
including cognitive radio network elements and access types. The need for cross layer design to
achieve cognitive functions and the cross layer operation of each cognitive function are also

7
included in this section. A selection of noticeable novel ideas introduced to realize the operation
of cognitive radios is presented in section 6. Finally, conclusions are drawn in section 7.

2. The structure of cognitive radio

A cognitive radio is essentially a class of software radios (SR’s) with additional capabilities
and functionalities such as environment sensing, learning, and decision-making, which enable it
to reach the required dynamic performance. On software level, cognitive radio enables running
of high-level application software to emulate a personal digital assistant (PDA). In order to
understand the structure of a cognitive radio, we have to explore software radios and their
practical version; the software-defined radios (SDR’s), since they represent the main component
of a cognitive radio. The rest of this section provides a brief overview of software radios and
software defined radios.
A software radio (SR) is defined as a transceiver whose communication functions are realized
as programs running on a suitable processor. Hence, a SR comprises all the protocol stack layers
of a communication system. Based on the same hardware, adaptation of transmitter/receiver
algorithms can be done on software to match different transmission standards. A wide-band
antenna attached to the hardware allows its operation over different bands. A software radio is
hence, highly flexible and adaptable, which meets the essence of the cognitive radio
specifications. An ideal SR directly samples the antenna output and converts these samples to
digital domain, and then the entire base band signal processing is done in digital domain.
Software radio is rather a theoretical model, because directly digitizing the wide-band antenna
output will lead to the digitization of an unnecessary huge bandwidth filled with many different
signals of no interest. This is neither technologically nor commercially desirable. A software-
defined radio (SDR) is a practical version of a SR; the received signals are sampled after a
suitable band selection filter to reduce sampling and digitizing complexity. The difference
between a SDR and a conventional receiver lies in the reconfigurability feature in SDR, where
transmission parameters can be modified via a control bus. Such a configuration is called a
Parameter-controlled (PaC) SDR [6], and shown in Figure 2.

8
 Figure 2: SDR/CR transceiver [6].

For SDR’s, reconfigurability means that the radio is able to process signals of different
standards or even signals that are not standardized but exist in specific applications. This
represents the one of the basic cognitive tasks of a cognitive radio as will be explained in section
4.1.

3. Concepts and definitions introduced for Cognitive Radio technology

The adoption of the cognitive radio technology as the technology for the next generation
wireless networks resulted in the emergence of new concepts, new definitions and new
quantities. In the following sub-sections, we discuss some of the main concepts and definitions
introduced by cognitive radio technology in order to make it easier to comprehend and digest the
discussions in the following sections.

3.1. Primary/Secondary users/networks

The presence of cognitive radio devices divides the wireless spectrum users into two types or
ranks, namely, Primary (licensed) users, and Secondary (unlicensed/cognitive) users.
Consequently, two networks shall coexist; primary network, which contains all primary users
with their respective licensed communication systems, and a secondary independent network that
contains secondary unlicensed cognitive users. Primary users are the original incumbent users
that can have legacy access to their dedicated spectrum bands unconditionally at any time or
place. Secondary users are the cognitive radio users that access spectrum in an opportunistic

9
manner. Primary users do not exhibit any cognitive behavior and their operation or design must
not be affected by the presence of cognitive users. The absolute priority for spectrum access
always goes to the primary user. Even if the secondary user is already transmitting over a certain
channel, it must evacuate this channel immediately if a primary user was sensed to be attempting
to access this channel.

3.2. Spectrum holes

This term was first defined in [5] as “a band of frequencies assigned to a primary user, but, at a
particular time and specific geographic location, the band is not being utilized by that user”. As
will be discussed in section 3.5, spectrum hole definition can expand to have many dimensions.
A spectrum hole can be a band of frequencies unutilized by the primary user at a certain time, or
at a certain location, or polarization, or code, or can be a combination of some or all of these.
In [7], a spectrum holes were classified into three types:
1. Black spaces, which are fully occupied by primary users some of the time
2. Grey spaces, which are partially occupied by low-power interferers such as short-range
UWB devices.
3. White spaces, which are free of RF interference except for ambient noise.
The principle of a cognitive radio operation is mainly based on locating these holes and
transferring data through them. However, this involves more complicated and advanced
functions as will be discussed in the rest of this report.

3.3. Spectrum pooling

Spectrum pooling is a term that represents the coexistence of two independent radio systems;
the primary and secondary systems within the same frequency range. It enables opportunistic
utilization of already licensed frequency bands by the secondary (cognitive) system.
The term “spectrum pool” was first mentioned in [8]. It basically represents the idea of
merging spectral ranges from different spectrum owners (military, trunked radio, etc.) into a
common pool. The goal of spectrum pooling is to enhance spectral efficiency by overlaying a
new radio system (cognitive radio system) on an existing one without requiring any changes to
the actual licensed system.

10
3.4. Interference temperature
The interference temperature is defined to be the RF power measured at a receiving antenna
per unit bandwidth [9]. The interference temperature model is proposed in [10]. It represents a
new technique for measuring interference, where the contribution of a low-level communication
system such as short-range UWB is assumed to be just another noise source that is included in a
link budget. All of these sources are summed together to form a new noise floor as shown in
Figure 3 [10]. The noise level in a given environment can be thought to be composed of three
main components:
1. Natural thermal noise (KTB).
2. Unintentional man-made noise.
3. Intentional man-made noise (e.g. short-range UWB devices).
Slightly increasing the noise level by allowing a small amount of intentional man-made noise can
provide us with a great opportunity for low power systems to operate below this interference cap.
Short-range UWB devices, for example have a very low transmission power, just one or two dB
above the original noise floor, since in most wireless systems, the link budget design takes into
account some link margin, therefore, the performance of these systems almost will not be
affected by the presence of the UWB devices. As we can see in the figure below, the interference
caused by the UWB devices is indicated by the peaks over the original noise floor, in the worst
case, the new interference cap formed by the UWB devices slightly reduces the range of
operation of the primary licensed system. However, this small compromise yields a huge
spectrum opportunity for various systems to operate in the UWB range.

Figure 3: Interference temperature model [10].

11
 Interference temperature limit (interference cap) is the fixed limit for the amount of
interference that secondary users are allowed to cause to the primary users, Hence low power
devices are allowed to operate as long as their transmission do not exceed the interference cap.
The presence of this interference cap would assure that the licensed operation would not
experience any further degradation or loss of service from new interference, and thereby provide
licensed systems greater certainty regarding the maximum permissible level of interfering RF
energy in the bands in which they operate. This will allow a more reliable link budget design.
Two ways were proposed for reliable interference temperature estimation [7]:
1. Using the multi-taper method [11] to estimate the power spectrum of the interference
temperature due to the cumulative distribution of both internal sources of noise and
external sources of RF energy.
2. Using distributed sensors that can efficiently sniff the RF environment, large number of
sensors can be used to compensate for the spatial variations of interference temperature
from one place to another.

3.5. Multi-dimensional channels in multi-dimensional spectrum space.

Spectrum is considered a physical quantity; hence, it must be measured in units. A channel is
considered the basic spectrum unit in any communication system. However, the definition of a
channel varies from one respect to another. Conventionally, a channel is simply defined in the
frequency dimension as a certain frequency band; this is a limited single-dimensioned definition
that does not suit the cognitive radio technology as a technology for the future. A broad and
comprehensive definition defines the spectrum to be partitioned into a set of orthogonal non-
interfering channels [12]. This definition expands a channel to have more than one dimension
such as power, location, direction of arrival (DOA), code, etc. Moreover, since a channel is
basically, a part of the spectrum space, the definition of a spectrum space needs consequently to
be expanded to include the multi-dimensional nature.
Dynamic spectrum access in cognitive radio systems demands increased consideration of the
use of time as a dimension of the signal space, since at every time instant, the channel utilization
distribution changes, and consequently the distribution of spectrum holes changes as shown in
Figure 4. This demands very high rate of spectrum data update to avoid affecting primary user
transmission.

12
 Figure 4: Spectrum hole distribution on a 2D spectrum space at two instants of time ©.
(a) t = t0, (b) t = t0+ τ.
Typically, when defining the spectrum space dimensions, the dimensions are supposed to be
orthogonal, that is, the values uniquely define a point in the space. However, several of the
possible parameters (e.g. modulation), do not necessarily define an orthogonal parameter but
they can still be used to distinguish signals. An efficient technique for modulation scheme
identification is proposed in [13] and discussed in section 6.1.
The usage of power as one proposed dimension of the signal space is done to address the
underlying low power, non-interfering systems such as short range UWB devices mentioned in
section 3.4. The possible dimensions of the spectrum space are arranged in Table 1 [14].

Table 1: Possible dimensions of spectrum space [14]

13
 A 3D spectrum space model for modeling network resources was proposed in [15], the
dimensions are time, rate and power/code, as shown in Figure 5. The time dimension models the
time required to transfer information. The rate dimension models the data rate of the network
The power/code dimension models the energy consumed for transmitting information through
the network, in the case of CDMA networks, the bandwidth increase due to the multi-code
transmissions is also captured in this dimension.

Figure 5: 3D spectral space and Virtual cube model for spectral unit [15]

The primary and cognitive networks both comprise heterogeneous spectrum access and
utilization techniques. To guarantee successful interoperability between these networks, a
universal model for a spectral unit must be specified to create a common spectral “currency” that
simplifies spectrum “trading” between different communication systems.
From the cognitive radio respective, a spectrum unit will help to efficiently model the multi-
dimensional spectrum space, specify precisely the available spectrum opportunities, efficiently
distribute them over CR users, and be able to evaluate the performance of both the primary and
cognitive network.
In [15], Virtual Cube concept depicted in Figure 5 was proposed for network performance
evaluation. The Virtual Cube concept defines a unit structure based on the resource allocation
techniques used in existing networks. The resource is modeled in a three dimensional resource-
space with time, rate, and power/code dimensions as shown in Fig. 5.

14
4. Cognitive tasks and functions
The extraordinary operation of a cognitive radio requires extraordinary features as well,
comprising unique tasks and functions and unique challenges. In the next sub-sections, we will
have an overview of the main tasks and functions required for the cognitive radio to achieve the
desired expectations.

4.1. Cognitive capability & Reconfigurability.

Cognitive capability refers to the ability of the radio device to achieve “awareness” of the
surrounding environment by extracting useful knowledge about the surrounding radio
environment in a smart and efficient manner, while exerting minimum interference to the
surrounding users. This knowledge should comprise many dimensions not only frequency and
time, but also space, power, interference levels, different codes, etc...
Through this knowledge, the cognitive radio device can identify useful spectrum segments
(spectrum holes), and use a smart algorithm to choose the best and most appropriate hole for
operation.

Reconfigurability enables the cognitive radio use the gained knowledge by its cognitive
capability to dynamically reconfigure its transmission/receive parameters to adapt with the radio
environment variations. The embedded SDR in a cognitive radio makes this task easier by
enabling the cognitive radio to dynamically configure a variety of parameters such as frequency
of operation, modulation and coding scheme, etc…
For reconfigurability, the software-defined radio entity performs this task. For other tasks of a
cognitive kind, cognitive radio assigns them to signal-processing and machine-learning
procedures.

4.2. Cognitive functions

The cognitive tasks mentioned in the previous section can be distributed over four main
cognitive functions, namely, Spectrum sensing, spectrum management, spectrum mobility and
spectrum sharing [16]. The four functions cooperation represents the cognitive cycle shown in
figure 6 and discussed in the following sub-sections.

15
 Figure 6: Cognitive cycle ©

4.2.1. Spectrum sensing

On the startup of a cognitive radio device operation, the first task to be done is to get
acquainted to the surrounding radio atmosphere by identifying the available spectrum
opportunities for transmission. Such opportunities are available in the form of spectrum holes as
explained in section 3.2. The wireless communication spectrum has many dimensions, a
communication channel (or spectrum hole) can be defined not just by frequency and bandwidth;
a channel can be defined over space (spatial diversity), code (CDMA systems), polarization
(polarization diversity) or direction of arrival DOA (multiple antenna systems). Hence, spectrum
sensing is not confined to opportunities in just time or frequency domain, it extends also to many
other domains.
Consequently, cognitive spectrum sensing requires fast and efficient techniques for locating
spectrum holes. The sensing technique should be smart enough to identify different
communication technologies in the surrounding medium, and be able to expand its scanning

16
 dimensions to cover all available spectrum holes in all possible dimensions. For example if the
cognitive radio is operating in a CDMA technology radio environment, the cognitive radio
should be smart enough to identify this and expand its scanning dimensions to cover the code
domain also. Figure 7 depicts and example of such a 3D spectrum.
Another thing to be considered, is the time varying nature of the radio spectrum a cognitive
radio tries to access, this necessitates very frequent update of the spectrum hole information. This
adds more signal processing burden on a cognitive radio and reduces the bandwidth utilization,
since, typically at the moment a cognitive radio is sensing the spectrum; it cannot be
transmitting.

Spectrum usage Spectrum holes over

over code plane 1 different code planes

Spectrum usage
over code plane 2

Figure 7: 3D spectrum for CDMA-based systems ©

Mainly, the spectrum sensing techniques can be classified with respect to weather sensing data
is exchanged between users or not as cooperative and non-cooperative techniques. Classification
according to the detection techniques typically covers transmitter detection or interference based
detection. Receiver detection, despite being more likely to be accurate, is impractical (see sub-
section B). The classification is shown in Figure 8.

17
 Figure 8: Classification of spectrum sensing techniques ©.

A brief discussion of these sensing techniques is done below. Another classification [17]
classifies spectrum sensing into two categories, namely, power-based sensing, and waveform-
based sensing, but this classification does not cover all sensing aspects.

A. Non-Cooperative & Cooperative detection

For non-cooperative detection, each CR device detects primary user signals independently
through their local observations. In most cases, a CR network is physically separated from the
primary network so there is no interaction between them, but still the CR device records their
presence. This may lead to inaccurate sensing data. At the same time, some primary devices
might be inside the CR’s range but due to shadowing, the CR device cannot detect them.
Cooperative detection presents one possible solution that can add more accuracy to the
detection process. Sensing information is exchanged between CR devices to assure maximum

18
accuracy. Cooperative detection refers to spectrum sensing methods where information from
multiple CR users is incorporated for primary user detection. An example of such information
for MB-OFDM-based systems (see section 6.3) is the binary allocation vector [18]. In this
vector, each bit represents one OFDM sub-carrier, if a bit has a value of one, this means that the
corresponding sub-carrier is occupied, if it has the value of ‘0’, this means the sub-carrier is
vacant and available for cognitive access. Figure 9 shows an example this vector [18]. By
applying a logical OR the collected vectors from different CR devices, maximum protection for
Primary users with respect to interference can be obtained.

Figure 9: A schematic example of a binary allocation vector in a MB-OFDM-based cognitive

radio system [18]

Cooperative detection can be implemented either in a centralized or in a distributed manner

[19, 20]. In the centralized method [19], it is assumed that there is a common base station
capable of both receiving and sending with which all CR devices communicate. Each CR device
individually detects local primary users, if found, it informs other CR devices through the base
station. Distributed solutions [20] require exchange of observations among CR users through an
assigned common control channel (CCC).
✓ Advantages of cooperative detection:
o Theoretically more accurate since the uncertainty in a single user’s detection can be
minimized [21], especially in a high shadowing/fading atmosphere.
✓ Disadvantages:
o Requires additional operations and traffic overhead.

19
 o For centralized approach, CR base station is required for operation. This requires a CR
infrastructure.
o For non-centralized approach, common control channel reduces the bandwidth efficiency.
o This architecture can result in suboptimal spectrum decision due to possible conflicts
between data transmission and sensing.

B. Transmitter detection
The first thing that would come to our mind when thinking of a technique for spectrum sensing
is to detect active transmission channels between primary transmitters and receivers. Downlink
channel detection is usually more efficient to detect, since the transmitter is most probably a base
station hence, downlink signals are stronger and easier to detect. While uplink signals from
primary users are usually weak due to the limited battery power capability.
In practice, however, it is hard to use downlink channel detection for spectrum sensing
purposes, since it is too difficult to determine which primary users are inside the scanning range,
since simply we can detect a channel but the primary device is out of the CR communication
range, so this might result in exaggerated sensing data.
This leaves us with the only possible solution, which is the detection of the weak transmitted
signals from primary users. This solution, although it needs more receiver sensitivity, can give us
a better idea about the spectrum utilization in the surrounding area, and hence be able to locate
and characterize the available spectrum holes.
Detection of such transmission can be done most simply by detecting radiated energy over the
band of interest, or by using a matched filter, but these techniques, despite being simple have
many drawbacks, another technique was proposed, namely, cyclostationary feature detection
[22], which outperforms the first two techniques. The three techniques are discussed in the
following sub-sections.

▪ Matched filter detection

In Stationary Gaussian noise channel, usually, a matched filter gives optimum detection results
and a maximum SNR, provided that information about the shape of the transmitted signal is
available.

20
 ✓ Advantages:
o The main advantage of the matched filter is that it requires less time to achieve high
processing gain due to coherency.
o Can distinguish between primary user signals and other CR user’s signal. This allows
the CR device to have deeper knowledge of the surrounding radio environment and
helps in taking better spectrum sharing decisions.
✓ Disadvantages:
o It requires a priori 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.
✓ Proposed solutions:
o Since most wireless network systems have pilot, preambles, synchronization words or
spreading codes, these can be used for the coherent detection.

▪ Energy detection
If no enough information is available about the primary user signal, which is usually the case in
a CR environment, energy detection can be used. This can be done simply by band-pass filtering
the required bandwidth, then squaring and integrating the output over the observation interval,
and then the result is compared with a certain threshold to decide whether a primary user is
present or not [23]. A schematic diagram for an energy detector is shown in Figure 10 [24].

Figure 10: Energy detector implementation [24]

21
 ✓ Advantages:
o Very simple technique.
o Does not require any a priori knowledge of the primary user signal, which is mostly the
case in a CR environment.
✓ Disadvantages:
o Noise and ACI affect the performance of the energy detector, a very poor performance
at low SNR was observed in [17].
o The energy detector cannot differentiate signal types but can only determine the
presence of the signal. Thus, the energy detector cannot differentiate between primary
and secondary user signals. This makes it prone to false alarm when another secondary
user attempts to access the same channel.
✓ Proposed solutions: A pilot tone from the primary transmitter is used to help improve the
accuracy of the energy detector in [25]. However, this violates the basic requirement of the
CR system, which is not supposed to intervene in the performance or modify the design of
the primary user’s hardware, i.e. we cannot “touch” the primary user.

▪ Pilot detection
In mostly all current communication system, a deterministic pilot signal is transmitted for
synchronization and channel acquisition. The power of the pilot tone is typically 1-10% of the
total transmitted power. This makes transmitter detection based on pilot signal detection a very
attractive technique.
The benefit of pilot signals, if they are perfectly known to cognitive radio sensor, is that it can
be processed coherently using a matched filter, hence achieve a maximal SNR as explained
before. In most primary communication systems, a sine wave tone is used as a pilot signal as
shown in Figure 11. However, for packet-based systems, a short preamble packet containing a
high autocorrelation pseudo-noise sequence is used as a pilot. If a cognitive radio sensor knows
the packet structure of the system, then it can perform coherent processing of a preamble through
a set of matched filter correlators [24].

22
 Figure 11: Spectrum picture at the input of sensing radio for primary user signal
Detection [24]

This method can also be used in conjunction with energy detection method to provide an
efficient detection method for asynchronous packet-based communication systems. The
performance improvement in probability of detection is shown in Figure 12 below [24].

Figure 12: Curves comparing the probability of detection performance for Energy, preamble
(pilot), and combined detectors for a packet based communication system [24].

▪ Cyclostationary feature detection

In cyclostationary feature detection [22], modulated signals are in general coupled with sine
wave carriers, pulse trains, repeating spreading, hopping sequences, or cyclic prefixes, which
result in built-in periodicity. These modulated signals are characterized as cyclostationarity since
their mean and autocorrelation exhibit periodicity. These features are detected by analyzing a
spectral correlation function.

23
 ✓ Advantages:
o It differentiates the noise energy from modulated signal energy, since noise is wide-
sense stationary signal that does not exhibit periodicity signal and had no spectral
correlation, while modulated signals are cyclostationary with spectral correlation due to
the embedded redundancy of signal periodicity. Therefore, a cyclostationary feature
detector can perform better than the energy detector in discriminating against noise due
to its robustness to the uncertainty in noise power [17].
✓ Disadvantages:
o Computationally complex and requires significantly long observation time.
✓ Proposed solutions:
o A scheme that combines spectral analysis with pattern recognition based on neural
networks was proposed in [13]. Distinct features of the received signal are extracted
using cyclic spectral analysis, The neural network, then, classifies signals into different
modulation types By combining these techniques a more efficient and reliable classifier
can be developed where a significant amount of processing is performed offline, thus
reducing online computation. More about this technique is discussed in section 6.1.

Since the current radio environment is highly dynamic, especially in multiple access bands,
therefore, spectrum-sensing data must be updated frequently. Scanning the spectrum for
detection is slow, and it is not a smart approach for spectrum sensing, a cognitive radio is
inherently a smart device; hence, a smarter approach for sensing is required.

C. Interference-based detection
The interference temperature model discussed in section 3.4 has been exploited for the
detection of primary users, and detection/identification of other interferers (secondary users).
Since only primary users have the privilege of exceeding the interference cap limit, by measuring
the interference temperature over different bands; if the measured interference temperature is
below the interference cap limit, this means that this band is not utilized and can be used for
cognitive access. On the other hand, if the interference temperature exceeds the interference cap
limit, this indicates the presence of a primary user (or another cognitive user). By using smart

24
pattern-recognition techniques such as one in [13]. The cognitive radio device can differentiate
between primary users and secondary users.
Two methods for interference temperature were mentioned in section 3.4, namely Multi-taper
method [11], and distributed sensors method [7].

4.2.2. Spectrum management

After spectrum sensing stage, the CR device now has information about the available spectrum
holes that would be spread over different frequency bands or timeslots or locations or codes, etc.
Each hole has its own unique characteristics such as frequency, bandwidth, noise level, etc...
Spectrum management protocol has to analyze data given about spectrum from the sensing
stage, and then decide which band to choose according to the QoS requirements. This requires a
cross-layer co-operation between spectrum sensing (Physical layer) and upper layers which will
do the analysis and decision. The analysis and decision tasks are discussed briefly below. More
details about the cross layer performance for spectrum management are discussed in section 5.4.

A. Spectrum analysis
Detected spectrum holes in the spectrum sensing stage need to be characterized in order to
extract the characterization parameters of each hole that define their eligibility to handle a certain
application. Examples of such parameters include operating frequency, bandwidth, path loss,
interference level, delay spread, channel error rate and holding time. Holding time is defined as
the expected time duration that the CR user can occupy a licensed band before being interrupted.
It is a critical CR-related parameter, which directly affects the quality of service. Holding time
can be estimated through observations of the primary users’ transmission behavior over each
channel. Channel capacity, which can be derived from the parameters explained above, is one of
the most important factors for spectrum characterization; spectrum decision mainly depends on it
as a decision parameter. Interference temperature model also can be exploited for capacity
estimation.

B. Spectrum decision
Each application running on a CR device requires a certain QoS, which in turn sets some
minimum requirements for the desired operating channel. For example, applications such as real-

25
time video transfer demand a high quality of service, large bandwidth, very small delay and a
large holding time in order to guarantee a satisfactory video quality. In other words, we need a
high quality channel. On the other hand, non real-time applications such as file transfer do not
impose very tight constraints on channel quality.
The spectrum decision protocol must first be able to define the minimum acceptable
parameters for each application, and then compare them to the available choices generated by the
spectrum analysis. The decision algorithm must be able to decide which channel best fits the
requirement. Various spectrum decision algorithms are proposed in [12, 26, and 27].

4.2.3. Spectrum mobility (spectrum handoff)

As discussed before, cognitive radio operation is mainly about exploiting unused licensed
spectrum segments in an opportunistic manner. This must be done without affecting the primary
users’ operation. If this was not 100% guaranteed, primary users would not allow cognitive users
to share their spectrum. Therefore, As soon as a primary user starts transmission over a channel
occupied by a secondary user, the secondary user must immediately evacuate the channel in a
smart and efficient manner that would not cause any harm to the integrity of the primary user
transmission. Such a function is called “spectrum mobility” or “spectrum handoff”.
A spectrum mobility protocol must also guarantee the secondary user a smooth transition in
terms of the associated latency, since as the secondary user transits from one channel with certain
parameters to another channel with different parameters; this required a very fast adaptation to
the new parameters like operating frequency, modulation scheme and the adequate associated
source/channel coding. This requires cooperation between all layers in network stack as shown in
figure 13 to quickly adapt to the new operation parameters and provide transparency to the
transition latency. More details about this subject are discussed in section 5.4. Useful information
about this latency can help to improve mobility performance if provided by the spectrum sensing
protocol, where it can give an estimation of the transition process duration in advance; this can
help the upper layers minimize performance degradation during transition.

4.2.4. Spectrum sharing

A wide range of spectrum opportunities is open for CR users to access; this besides being an
advantage, imposes a challenge to the CR technology, since access should be organized between

26
CR users in a fair manner. Spectrum sharing is the function concerned with this subject; it is
similar to the medium access control (MAC) protocol used in most networking systems.
However, due to the dynamic nature of the CR network and the presence of two ranks of users
(primary and secondary users), Spectrum sharing protocol must exploit smarter and more
advanced algorithms to guarantee peaceful co-existence with primary users and fair spectrum
sharing between secondary users. Efforts done on designing spectrum sharing protocols took two
main directions:
1. Game theory, exploiting Nash equilibria or no-regret learning algorithms [28, 7]. More
details are explained in section 6.2.
2. Information theory, mainly through exploiting Iterative water-filling algorithm [29]
which was developed originally for dynamic spectrum management in digital subscriber
lines [30].
Both approaches were well discussed and evaluated in [7], mentioning their advantages and
disadvantages.

5. Cognitive radio system architecture

A cognitive radio network will operate in a unique environment, the spectrum of its operation
contains licensed and unlicensed bands, and licensed bands are allocated to different systems
with various communication technologies. Operation in such a heterogeneous atmosphere
requires a unique network architecture that can adapt quickly, coexist peacefully and operate
efficiently in such environment.

5.1. CR network elements

The spectrum will witness the coexistence of two major networks, the primary network, and
the secondary (CR) network. The primary network represents the already existing licensed
network infrastructure. It has the legitimate right to access its own dedicated band at any time
without any interruption or interference from any intruder. The primary network elements
comprising primary users and primary base stations do not exhibit any cognitive behavior. Their
coexistence with the secondary network elements should not disturb their operation or require
them to add any additional functions or design modifications. However, the primary base-station

27
may be requested to have both legacy and CR protocols for the primary network access of CR
users [16].
On the other hand, the secondary (CR) network, that does not have any legitimate rights for
using any spectrum bands, attempts to access the temporarily unused spectrum segments in an
opportunistic manner that does not affect the primary network operation. CR network can be
either an infrastructure-based network or an ad-hoc-based network.
Because of the opportunistic nature of the CR network operation, the CR network elements
comprise elements with more complicated functionalities than those of the primary network.
In [31], a CR network architecture called DIMSUM-Net was proposed. It consists of four
elements: base stations, clients, a Radio Access Network Manager (RANMAN) that obtains
spectrum leases, and a per-domain spectrum broker that controls spectrum access and plays a
role in sharing spectrum resources among different CR networks.

5.2. CR network access

CR users can communicate with each other either in a multi-hop manner or through the CR
base-station. Thus, in CR networks, there are three different access types [16]:
1. CR network access: CR users can access their own CR base-station both on licensed and
unlicensed spectrum bands.
2. CR-ad hoc access: CR users can communicate with other CR users through ad hoc
connection on both licensed and unlicensed spectrum bands.
3. Primary network access: The CR users can also access the primary base-station through the
licensed band [16].

5.3. The need for a cross layer design

From the discussion of cognitive radio functions in section 4.2, it can be concluded that cross
layer design is required for all cognitive functions, because the continuous change in the
operating spectrum parameters results in the dynamic nature the CR operation. Hence, fast
adaptation and smart decisions based upon learning and awareness are required to attain proper
operation. Physical layer parameters must be adapted to satisfy the QoS requirements from the
application layer. Hence, knowledge has to be shared between all layers in between in order to
obtain the highest possible adaptability.

28
5.4. Cross layer operation in cognitive radio functions.
In cross-layer design-based functions, each function requires the cooperation of more than one
layer to achieve the required function. Figure 13 shows the layers involved with each cognitive
function [16].

Figure 13: CR network communication functionalities [16]

▪ As we can see from the figure, Spectrum sensing is mainly done on Physical layer, but some
help from the Link (MAC) layer is required, especially instructing the Physical layer to
choose the appropriate spectrum hole among the detected ones.
▪ Spectrum sharing is mainly a Link (MAC) layer function, but still requires some help from
the Physical layer, which provides it with data about the link status, and informs it if it senses
a primary user activity over the channel, so that the MAC layer decides what to do.
▪ Spectrum management and Spectrum mobility, on the other hand, require full cooperation
between all layers in the protocol stack. Spectrum management takes sensing data from the
bottom layer of the stack (Physical layer) and tries to match them with the QoS required for
the application running on the top of the stack (Application layer). The software on

29
 application layer identifies the spectrum opportunities by analyzing the given sensing data,
and decides the next spectrum hole for transmission with the aid of scheduling information
provided by the Link layer to guarantee fair spectrum sharing. According to this decision,
transport layer reconfigures the transmission parameters such as modulation scheme and
carrier frequency, and the Network layer re-routes the data stream. The protocol then
reconfigures the Link layer and Physical layer to handle the new transmission parameters. As
we notice, the spectrum management process is an up-down process, where the flow goes up
the protocol stack in the spectrum analysis stage, and goes down in the decision stage.
▪ Similarly, for spectrum mobility, Physical layer gives an alert on the detection of a primary
user starting transmission; Spectrum management function decides the next spectrum hop to
the next spectrum hole, according to this decision, link layer provides assumed latency for
the transition. Network layer provides new route data, data loss due to this transition is
specified by the transport layer, and the application layer tries to overcome this loss or at
least alleviate the abrupt QoS degradation caused by the transition.
It is too obvious that all cognitive functions are highly correlated, we cannot mention one
function without mentioning the other, protocol stack layers all work in conjunction in a cross-
layer manner to fulfill these cognitive functions. No layer can work independently without the
help of other layers.

6. Novel ideas proposed for CR technology

During the last few years, a lot of research has been done on cognitive radio, covering a lot of
aspects and considerations. Some works introduced novel and unique ideas that I present a few
of them in this section.

6.1. Neural networks for efficient pattern recognition

As previously mentioned in this report, and confirmed in many other articles [13, 16, 17, and
32] that cyclostationary feature detection is the optimum detection scheme for spectrum sensing,
since it provides a way for classifying signals for applications where the carrier frequency and
bandwidths are unknown. However, it is computationally complex and requires a significant
amount of observation time for adequate performance. On the other hand, neural networks have
been used for signal classification, but only for situations where the base band signal is present.

30
 In [13] A combination of the two techniques to get a more efficient and reliable classifier was
proposed, where distinct features of the received signal are extracted using cyclic spectral
analysis and represented by both spectral coherent function and spectral correlation density
function. The neural network, then, classifies signals into different modulation types.
Using the proposed approach, signal modulation types can be determined quickly and reliably.
Advantages of the proposed method merges the advantages of cyclostationary feature detection
such as robustness to stationary noise, separating signals with overlapping power spectral
densities, and keying component extraction, with the flexibility granted by using neural
networks, since the network can be retrained easily in order to incorporate new signal types.
However, in this article, the set of signals was limited to BPSK, QPSK FSK, MSK, and AM,
while more realistic digital modulation types such as QAM and PSK were not considered. Also,
the case of two or more signals present at a time was not considered. The proposed system
estimates only the modulation type, while realistic CR spectrum sensing must specify other
parameters such as carrier frequency and bandwidth. Eventually, a question arises which is
weather the use of neural networks would enhance or simplify the sensing process or would it
give no considerable gain.

6.2. Game theory for spectrum sharing

In spectrum sharing process, it is required that spectrum usage opportunity to be fairly shared
between CR users, so that no users monopolize the spectrum opportunities available (always
winning), while others cannot find any opportunity (always losing). This calls for the need for
shared spectrum environment where “everybody wins”. Game theory then can be applied to
provide the criteria by which this objective can be guaranteed.
In [28], Game theory was used to provide a framework to analyze the behavior of CR for
distributed adaptive channel allocation, in this framework, the cognitive radios are the players
and their actions are the selection of new transmission parameters and new transmission
frequencies, etc., which influence their own performance, as well as the performance of their
neighboring players. The article compared the performance of cooperative and non-cooperative
(selfish) spectrum sharing techniques.
The cooperative case was modeled as an exact potential game, which converges to a pure
strategy Nash equilibrium solution. However, this framework was shown not to be applicable for

31
non-cooperative spectrum sharing and a no-regret learning algorithm was proposed instead.
Alternatively, a no-regret learning implementation was proposed for both cooperative and non-
cooperative cases, and it was shown to have similar performance with the potential game when
cooperation is enforced [28].
Simulation results showed that cooperative spectrum sharing approach can obtain an improved
operation and fairness, but on the expense of high signaling overhead due to the requirement of a
high environment knowledge. On the other hand, non-cooperative approach does not require
complete environment knowledge and need less implementation complexity, but on the expense
of degraded performance and fairness. The flexibility of the No-Regret learning-based scheme
used provides potential benefits on the design of cognitive radios for heterogeneous networks.

6.3. MB-OFDM as the proposed Physical layer technology for CR

The dynamic performance and continuous adjustments of radio operating parameters of
cognitive radio requires highly flexible and adaptable Physical layer, especially with respect to
the spectral shape of the transmitted signal. This property is necessary for flexible spectrum
access by efficiently filling any of the available spectrum holes, no matter how small they could
be.
A special case of multi-carrier transmission, known as orthogonal frequency division
multiplexing (OFDM) is one of the most widely used technologies in current wireless
communications systems. Because of its attractive features, OFDM has been successfully used in
numerous wireless technologies including wireless local area network (WLAN), wireless
metropolitan area network (WMAN), and the European terrestrial digital video broadcasting
(DVB-T).
By dividing the spectrum into low-rate parallel data streams that are modulated with
orthogonal sub-carriers, OFDM removes the need for equalizers and thus reduces the complexity
of the receiver. Low symbol rate on each sub-carrier results in lower sensitivity to inter-symbol
interference and hence, a simple equalizer structure, and high immunity to multipath fading. The
basic principles of OFDM are depicted in Figure 14.

32
 Figure 14: OFDM technique (a) Multi-carrier transmission (b) Orthogonality of sub-carriers (c)
OFDM transceiver structure [18].

The basic idea behind OFDM’s flexibility in spectrum access is to match the bandwidth of the
chosen spectrum hole with an integer multiple of the carrier spacing Δf used in the CR system.
As Δf gets smaller, the resolution and flexibility increases, but of course on the expense of more
signal processing complexity.
OFDM has the potential of fulfilling the requirements of cognitive radios inherently or with
minor changes for many reasons such as [33]:

1. OFDM enables easy shaping easily shaping of the transmitted signal, by carefully
adjusting the signal shape; interference to other neighboring bands can be reduced. Deep
notches can be inserted into the transmission band to allow primary user operation within
the OFDM operation band, see section 6.7 for more details.

33
 2. By using the FFT engine, the MB-OFDM device can loosely approximate a spectrum
analyzer. The average power levels transmitted by other devices within the MB-OFDM
receiver band can be measured by the averaging the received signal over time. This
approach will not only measure the power levels of the interferers, but that of other MB-
OFDM devices.
3. By inserting quiet periods into the MB-OFDM network's operation, it is possible to
obtain a relatively accurate estimate of the signal to interference-plus-noise ratio (SINR)
in the band of operation.

6.4. MIMO for high flexibility-spectrum efficient transmission.

It is widely recognized that the use of a multiple input multiple output (MIMO) antenna
architecture can provide for a spectacular increase in the spectral efficiency of wireless
communications. With improved spectrum utilization as one of the primary objectives of
cognitive radio, it seems logical to explore building the MIMO antenna architecture into the
design of cognitive radio. The end-result is a cognitive MIMO radio that offers the ultimate in
flexibility, which is exemplified by four degrees of freedom: carrier frequency, channel
bandwidth, transmitting power, and multiplexing gain.

6.5. Spatial filtering for spectrum sensing

Spectrum sensing process requires reliable detection of the primary user signal, to achieve this
function, a wide band RF front end and a signal processor are required. In order to ease this
processing, its easier to work in the digital domain, hence, received analog signals must be
converted into digital ones using an Analog to digital converter (ADC).
However, since the received signal is wideband with potentially large dynamic range, the ADC
is required to have high speed and high resolution to meet the system specifications. A novel
idea of a dynamic range reduction algorithm was proposed in [32], where the dynamic range of
the received signal is reduced before conversion by spatially filtering strong signals using a
phased antenna array with weights computed in the digital domain and fed back to adjust the
gains and phases of the antenna elements. System architecture and plots of the results are shown
in figure 15.

34
 (a)

(b) (c)
Figure 15: Spatial filtering (a) Diagram of the spatial filtering system, (b) Large dynamic range
signal before filtering, (c) Signal after filtering. [32]

6.6. Turbo processing for higher QoS

In [7], Building turbo processing into the design of a cognitive radio was strongly suggested.
Turbo processing highly improves channel decoding and channel equalization, which result in a
reduction of the frame error rate (FER), and hence a better QoS. While a high QoS is one
essential requirement of a CR system, exploiting turbo processing can yield remarkable
improvements in the system performance. More details about Turbo processing can be found in
[34].

6.7. Deep Spectrum notching using data specific tones for MB-OFDM-based systems
As discussed in section 6.3, Multi-Band OFDM system is a very strong candidate to be adopted
by the next generation CR systems, due to its immunity to multi-path fading, ease of

35
implementation in digital domain, bandwidth efficiency, flexibility, and many other advantages.
One of the most interesting advantages of MB-OFDM is the ability to easily do spectrum
shaping, by controlling the sub carriers, which can be done adaptively in a CR atmosphere to
give a maximized flexibility. The first CR-useful spectrum shaping aspect that would come to
our mind is spectrum notching; a CR device can adaptively insert a notch into the spectrum in
the band occupied by a licensed user while continuing transmission in the surrounding band.
This notch can be temporary or permanent for fixed bands such as the Japanese radio astronomy
bands within the UWB range.
A straightforward technique for spectrum notching is just by turning-off (sending zeros over)
the sub-carriers (tones) that overlap with the licensed user channel. For this technique, the
receiver does not need to have a priori information that some of the tones have been zeroed. This
fact can be estimated during the channel estimation phase and will be viewed as fading within
the channel. Note that this technique can only provide limited notching capabilities because of
spectral leakage that is dependent on the data transmitted on the adjacent non-zero tones. This is
exactly the reason why a large number of zero tones must be used in order to create notches with
significant depth.
Typically, this approach requires a large number of zero tones adjacent to the victim receiver
band in order to create a deep notch, but the implementation is extremely simple and the receiver
requires no a priori knowledge of which tones have been zeroed.

A deeper notch can be obtained using fewer tones by inserting data-specific dummy tones on
either side of the victim band [35]. By accounting for the contribution due to the non-zero tones,
dummy tones can be inserted to minimize the spectral leakage, and thereby increase the depth of
the notch. However, this technique increases the ripple of the power spectral density due to
peaking at the band edges. This approach requires that MB-OFDM devices back-off an
additional 1.8 dB in radiated power to be compliant with the FCC regulations [36], resulting in
lower range. A method for eliminating this peaking by constraining the amount of energy placed
on the data-specific dummy tones is discussed in [33, 36].
Figure 16-a shows an example where 9 tones (~45 MHz) are zeroed-out to achieve a 15 dB
deep notch for a 7 MHz channel within 3260-3267 MHz band. While Figure 16-b shows the

36
notch generated using this technique, where only 5 tones are used (~25 MHz) to create a notch
with depth of 24 dB for the 7 MHz channel within 3260-3267 MHz band [33].

(a) (b)
Figure 16: Spectrum notching (a) Spectrum notching by tone nulling (b) Spectrum notching
by using data-specific dummy tone insertion [33].

7. Conclusions
Cognitive radio technology provides us with a promising solution to the spectrum scarcity
crisis by exploiting new dimensions of spectrum for maximized spectrum utilization. In this
report, cognitive radio technology was introduced and its main issues were explored. Unique
features and behavior were discussed, and a rough description of the expected cross-layer
behavior was stated. Finally, some novel ideas for realization of cognitive functions were
presented.
It can be noticed in this report that most of the research done on cognitive functions
concentrate mainly on spectrum sensing issues, very few work; mostly theoretical was done on
other functions. In addition, in most of this work, cross-layer design concept was not deeply
rooted into the proposed ideas and designs. Moreover, artificial intelligence and learning
capability are not yet well applied to the proposed designs. Hence, more effort must be done in
these fields to enhance the performance of cognitive radio.

37
 Finally, from sections 6.3, 6.4, 6.6 and 6.7; a combined OFDM-MIMO structure with turbo
processing and high spectrum sculpting capabilities can serve as an ideal physical layer
architecture for cognitive radio.

38
References

[1] http://www.fcc.gov/oet/info/database/spectrum/
[2] J. Yang, “Spatial channel characterization for cognitive radios,” Master’s thesis, University
of California, Berkeley, 2004
[3] J. Mitola et al., “Cognitive radio: Making software radios more personal,” IEEE Pers.
Commun., vol. 6, no. 4, pp. 13–18, Aug. 1999.
[4] FCC, “Fcc first report and order: Revision of part 15 of the commission’s rules regarding
ultra-wideband transmission systems,” ET Docket No. 98-153, April 2002.
[5] P. Kolodzy et al., “Next generation communications: Kickoff meeting,” in Proc. DARPA,
Oct. 17, 2001.
[6] F.K. Jondral, Software-defined radio-basic and evolution to cognitive radio, EURASIP
Journal on Wireless Communication and Networking 2005.
[7] S. Haykin, Cognitive radio: brain-empowered wireless communications, IEEE Journal on
Selected Areas in Communications 23 (2) (2005) 201–220.
[8] J. Mitola III, Cognitive radio for flexible mobile multimedia communication, in: Proc. IEEE
International Workshop on Mobile Multimedia Communications (MoMuC) 1999, November
1999, pp. 3–10.
[9] Y. Xing, N. Mathur, M. A. Haleem, R. Chandramouli, K. P. Subbalakshmi, “Dynamic
Spectrum Access with QoS and Interference Temperature Constraints” IEEE
TRANSACTIONS ON MOBILE COMPUTING, VOL. 1, NO. 8, AUGUST 2006
[10] See FCC document FCC 03-289a1, available from www.fcc.gov
[11] T.P. Bronez, “On the performance advantage of multitaper spectral analysis,” IEEE Trans.
Signal Processing, pp 2941-2946, 1992.
[12] H. Zheng, L. Cao, Device-centric spectrum management, in: Proc. IEEE DySPAN 2005,
November 2005, pp. 56–65.
[13] Fehske, J.D. Gaeddert, J.H. Reed, A new approach tosignal classification using spectral
correlation and neural. networks, in: Proc. IEEE DySPAN 2005, November 2005, pp. 144–
150.
[14] W.D. Horne, Adaptive spectrum access: using the full spectrum space, in: Proc.
Telecommunications Policy Research Conference (TPRC), September 2003.

39
[15] M.C. Vuran, I.F. Akyildiz, AMAC: adaptive medium access control for next generation
wireless terminals, IEEE/ACM Transactions on Networking, 2007
[16] F. Akyildiz, W. Y. Lee, M. C. Vuran, and S. Mohanty, “NeXt generation/ dynamic
spectrum access/cognitive radio wireless networks: a survey,” Elsevier Computer Networks
Journal, vol. 50, no. 13, pp. 2127–2159, Sept. 2006.
[17] H. Tang, Some physical layer issues of wide-band cognitive radio system, in: Proc. IEEE
DySPAN 2005, November 2005, pp. 151–159.
[18] T.A. Weiss, F.K. Jondral, Spectrum pooling: an innovative strategy for the enhancement of
spectrum efficiency, IEEE Radio Communication Magazine 42 (March) (2004) 8–14.
[19] G. Ganesan, Y.G. Li, Cooperative spectrum sensing in cognitive radio networks, in: Proc.
IEEE DySPAN 2005, November 2005, pp. 137–143.
[20] J. Zhao, H. Zheng, G.-H. Yang, Distributed coordination in dynamic spectrum allocation
networks, in: Proc. IEEE DySPAN 2005, November 2005, pp. 259–268.
[21] Ghasemi, E.S. Sousa, Collaborative spectrum sensing for opportunistic access in fading
environment, in: Proc. IEEE DySPAN 2005, November 2005, pp. 131–136.
[22] W. A. Gardner, C. M. Spooner, ”Signal interception: performance advantages of cyclic-
feature detectors”, IEEE Trans. On Communications, vol.40. January 1992.
[23] F. Digham, M. Alouini, M. Simon, On the energy detection of unknown signals over fading
channels, in: Proc. IEEE ICC 2005, vol. 5, May 2003, pp. 3575–3579.
[24] D. Cabric, R. Brodersen, “Cognitive Radios: System Design Perspective ,” PhD thesis,
University of California, Berkeley, 2007.
[25] Sahai, N. Hoven, R. Tandra, Some fundamental limits in cognitive radio, Allerton Conf. on
Commun., Control and Computing 2004, October 2004.
[26] V. Kanodia, A. Sabharwal, E. Knightly, MOAR: a multichannel opportunistic auto-rate
media access protocol for ad hoc networks, in: Proc. IEEE BROADNETS 2004, October
2004, pp. 600–610.
[27] S. Krishnamurthy, M. Thoppian, S. Venkatesan, R. Prakash, Control channel based MAC-
layer configuration, routing and situation awareness for cognitive radio networks, in: Proc.
IEEE MILCOM 2005, October 2005.
[28] N. Nie, C. Comaniciu, Adaptive channel allocation spectrum etiquette for cognitive radio
networks, in: Proc. IEEE DySPAN 2005, November 2005, pp. 269–278.

40
 [29] T. Cover and J. Thomas, Elements of Information Theory, New York, Wiley, 1991.
[30] W. Yu, G. Ginis, and J. Cioffi, “Distributed multiuser power control for digital subscriber
lines,” IEEE Journal on Selected Areas in Communication, vol. 20, no. 5, pp. 1105–1115,
June 2002.
[31] M.M. Buddhikot, P. Kolody, S. Miller, K. Ryan, J. Evans, DIMSUMNet: new directions in
wireless networking using coordinated dynamic spectrum access, in: Proc. IEEE
WoWMoM 2005, June 2005, pp. 78–85.
[32] D. Cabric, R.W Brodersen, Physical layer design issues unique to cognitive radio systems,
in: Proc. IEEE Personal Indoor and Mobile Radio Communications (PIMRC) 2005,
September 2005.
[33] A. Batra, S. Lingam, J. Balakrishnan, “Multi-band OFDM: A Cognitive Radio for UWB”
IEEE, ISCAS 2006
[34] A. Berrou, “The ten-year old turbo codes are entering into service,” IEEE J. Commun.
Mag., vol. 41, no. 8, pp. 110–116, Aug. 2003.
[35] H. Yamaguchi, "Active Interference Cancellation Technique for MB-OFDM Cognitive
Radio," European Microwave Week, Amsterdam, 2004.
[36] J. Balakrishnan, "Dummy Tone Insertion for Spectral Sculpting of the Multi-Band OFDM
UWB System.

41

View publication stats