1

Chapter 1. Introduction
1.1 History and Motivation for Cognitive Radio

Cognitive radio promises a more efficient use of scarce radio spectrum and much

research has been done into applying existing techniques in wireless communications to

achieve the aims of this new technology. Interest in cognitive radio has risen due to the

increasing scarcity of available frequency spectrum. Although licensed spectrum is

scarce, surveys of licensed bands show that they are often unoccupied by the licensed

user [2]. Therefore, it would be desirable if a technology existed that could exploit this

unused spectrum, which is what cognitive radio aims to do. In the literature, the licensed

user is commonly referred to as the primary user and the nodes in the cognitive radio

network are referred to as the secondary users. The goal of the secondary users is to

communicate in the licensed spectrum, while limiting interference to the primary user.

The push for cognitive radio technology is driven by the saturation of the licensed radio

frequency spectrum as well as studies which show that it is being highly underutilized.

The Federal Communication Commission (FCC) is the body responsible for allocating

frequency spectrum and dictating how it is used in the United States. As seen in Figure

1, the radio frequency has been divided into frequency bands, each having a specific use

[2].
 2

Figure 1: Radio Frequency Spectrum Allocation [2]

Testing of some of these licensed bands by the FCC has shown that many are

underutilized. Underutilized can mean that for large portions of time the band is unused

or that in some areas signals from the licensed users are not present. These are referred to

as the time dimension and the spatial dimension of the frequency spectrum. Figure 2

shows the utilization of a 7 megahertz band in three different major US cities.

 3

Figure 2: Utilization of a 7 MHZ band below 1 GHz

The curves represent the percentage of a thirty second window that was occupied in a 7

MHz band below 1 GHz. As can be seen in the plot, even in a major city such as

Chicago, the band was unoccupied for almost half of the time. Tests have also shown

that many channels in some of the licensed bands are more underutilized than others.

This would be considered the frequency dimension of the spectrum. According to a 2002

FCC Spectrum Policy Task Force Report, the existence of such low average occupancy

bands in urbanized areas with increasing spectrum demand raises interest in the

possibility of matching supply and demand better with alternative technology.

Therefore, it would be advantageous if a technology existed that was capable of detecting

the state of the time, space, and frequency dimensions of the RF spectrum and adapting to

this environment to improve spectral efficiency. This is what Cognitive Radio aims to

do.

1.2 Introduction to Cognitive Radio Networks

An ideal Cognitive Radio Network (CRN) has the ability to detect the state of the RF

environment, identify available spectrum, and exploit the available spectrum without

interference to licensed users communication. The users in the licensed network are
 4

referred to as primary users (PUs) and the users in the cognitive network are referred to

as the secondary users (SUs). The SUs must exploit the spectrum of the PU network

autonomously, because it is assumed that the PU network is a separate, independent

network that does not provide feedback to the SU network. Therefore, it is critical that

the SU network accurately assess the channel-state of the PU network and vacate a

channel before the PU reoccupies it. This can be accomplished with spectrum sensing

and predictive modeling [3]. According to [1], there are two primary objectives for

cognitive radio networks:

Highly reliable communications whenever and wherever needed

Efficient utilization of the radio spectrum

In order to accomplish these objectives, a CRN must repeatedly go through a basic

cognition cycle as shown in Figure 3.

 5

Action: transmitted signal

Radio Environment

RF
Stimuli

Spectrum Holes
Noise-floor statistics
Traffic statistics
Transmit-power control and Radio Scene
spectrum management Analysis

Interference Temperature

Channel-state estimation and

predictive modeling

Transmitter Receiver

Figure 3: Cognition Cycle [1]

The figure shows the tasks that must be accomplished by the CRN and the parameters

that must be considered to achieve those tasks. The steps in the Cognition cycle can be

summarized as follows [1]:

Radio scene analysis: Radio scene analysis consists of sensing the interference

temperature of the radio environment and detecting available spectrum.

 6

Channel Identification: Channel identification consists of estimation of the

channel state information and estimation of the channel capacity.

Transmit-power control: Transmit power control consists of controlling the

transmission parameters of the network to manage the spectrum.

This cognition cycle consists of two key components: radio scene analysis and

adaptation of transmission parameters. The adaptation of transmission parameters is

the focus of this thesis.

1.3 Introduction to Beamforming in Wireless Communications

Beamforming (BF) is used in wireless communications in order to produce a high gain

signal or construct a directional beam. Beamforming is accomplished through adjusting

the amplitudes and phases of signals such that they combine constructively or

destructively at a receiver. This can be used to direct a beam to a desired receiver or to

direct a null in the beampattern to a receiver to avoid interference.

Beamforming can be performed by a linear array of transmitters or by a distributed

network of transmitters. Due to the ability to manage the spatial dimension of

spectrum, beamforming lends itself well to Cognitive Radio. Beamforming can be

accomplished by either using the positions of the intended and unintended receivers or

by using channel-state information (CSI) [10]. A more detailed survey of beamforming

techniques is presented in Chapter 2.

 7

1.4 Introduction to the Genetic Algorithm

Many problems that arise in CR BF do not have a clear analytical solution or require an

analytical solution that is complex due to optimization problems formed with the

constraints of a CRN. The Genetic Algorithm provides a way to arrive at reasonable

solutions to these problems with a low complexity. The GA is an algorithm inspired by

the biological process of natural selection. The basic idea of the GA is that for a given

population, the more fit individuals will produce more offspring and each generation

will be better adapted. The genetic diversity of the population changes due to random

mutations and desirable mutations will be passed on to future generations. The GA is a

simple algorithm that has been proven to be very powerful and widely applicable. [14]

Another benefit of the GA is the potential for parallelization, which is attractive for a

distributed application. A detailed description of the GA and its parameters will be

discussed in Chapter 3.
 8

Chapter 2. Beamforming
Techniques in Wireless
Communications
2.1 Background

One technique that lends itself well to Cognitive Radio is beamforming. Congnitive radio

operates under the condition that the interference to a primary user is kept below a certain

threshold or probability while communication between the secondary cognitive users is

maintained. Studies into beamforming show that N nodes can cooperate together to

achieve a gain approaching N2 [4]. Additionally, beamforming can be used to steer a

beam towards a desired receiver while steering a null to avoid interference to a different

receiver. This capability can be applied to cognitive radio to limit interference to primary

users by steering a null towards them [15].

2.2 Array Factor

The array factor represents the gain of the array in a particular direction. The array

factor, F(), is a function of the azimuth angle, . The array factor of an antenna array

is determined by the weighting applied to each antenna and the geometry of the array.

[13]

2.3 Diversity Gain

Diversity gain refers to the gain achieved through the use of multiple transmit antennas.

Diversity can be desirable in a wireless communication system, because the directivity

 9

increases as N2. For example, an isotropic antenna with a transmit power of N*P

achieves a uniform beampattern with power N*P. An array of N isotropic antennas each

with P transmit power can be phased appropriately in order to achieve a beampattern

with a maximum array factor of N2 although the same power was dissipated.

2.4 Array Beamforming

Array beamforming refers to an array of local antennas that are equally spaced and can

be weighted to achieve the desired radiation pattern. Antenna arrays can be used for

directional of arrival (DOA) estimation and for interference avoidance by steering nulls

toward unintended receivers and steering beams towards intended receivers. Figure 4

shows an array beamformer with M elements spaced a distance, d, apart.

 10

Figure 4: Array Beamformer with M elements with spacing d

The array factor of the beamformer is found by determining the relative phase shift of

the wavefront at each antenna element. The relative phase shift can be found using

simple trigonometry as:

= ( ) cos() /. (1)
 11

Where n is the element number, is the frequency, and c is the speed of light. The

array factor can then be found as

cos() (1) cos()

() = [1, , . . . , ]. (2)

This is also known as the steering vector and the opposite signed array factor can be

applied as element weights to steer a beam towards angle .

2.5 Distributed Beamforming

Distributed beamforming is performed by a distributed network of transmitters that are

not fixed locally in some predictable geometry such as in mobile network. Despite the

unpredictable and complicated geometry of distributed beamforming networks, the

distributed nodes can still cooperate to achieve a desired beampattern. In [6] it was

shown that an array of N distributed nodes can achieve the same diversity gain as a

linear array of N nodes as the number of nodes approaches infinity. A model of a

distributed beamforming network can be seen in Figure 5.

 12

Figure 5: Distributed Beamforming Network

The array factor of this network can be found similarly to that of an antenna array. The

relative phase difference is determined by finding the distance between the receiver

and each node. In the case of the receiver being in the same plane as the network, the

distance between each node and the receiver can be found trigonometrically to be:

= 2 + 2 2 cos(0 ). (3)

If the far-field condition of A>>rk is met, the distance can be approximated as:

cos(0 ). (4)

If the initial phase of each node is set to:

 13

2
= , (5)

Then the array factor is

2
1 [cos(0 ) cos( )]
() =
=1 . (6)

The far-field beampattern is then

() = |()|2 (7)

The authors in [6] then use the uniform distribution of the nodes to find the average

beampattern as:

1 1 1 (()) 2
() = + (1 ) |2 | (8)
(()

Where J1 is the first-order Bessel function of the first kind and

4
() = sin ( 2 ). (9)

The plot of the average beampattern can be seen in Figure 6 for different network

radiuses and transmitting nodes.

 14

-2
R = 2, N = 10
R = 2, N = 100
-4 R = 4, N = 10
R = 4, N = 100
-6
Average Beampattern (dB)

-8

-10

-12

-14

-16

-18

-20
-150 -100 -50 0 50 100 150
Azimuth Angle (degrees)

Figure 6: Average beampattern of N randomly placed nodes in circle of radius R

2.6 Effects of Imperfect Carrier Synchronization

The beampattern produced in a distributed beamforming system uses knowledge of the

relative locations of the nodes and assume a synchronized carrier frequency. If the

carrier frequency is offset due to noise in the phase-locked loop of the node, the

beampattern produced may not be the desired beampattern. [5] In CR, this effect can

cause protection to the primary user to be compromised [7],[8].