1

EC 6014- COGNITIVE RADIO

UNIT III

INTRODUCTION TO COGNITIVE RADIO

A ‗‗Cognitive Radio‘‘ is a radio that can change its transmitter parameters based

on interaction with the environment in which it operates.

A cognitive radio adds both a sensing and an adaptation element to the software

defined and software radios. Four new capabilities embodied in cognitive radios

will help enable dynamic use of the spectrum: flexibility, agility, RF sensing, and

networking.

Flexibility is the ability to change the waveform and the configuration of a device.

Agility is the ability to change the spectral band in which a device will operate.

Sensing is the ability to observe the state of the system, which includes the

radio and, more importantly, the environment

Networking is the ability to communicate between multiple nodes and thus

facilitate combining the sensing and control capacity of those nodes.

Cognitive radio is viewed as a novel approach for improving the utilization of a

precious natural resource: the radio electromagnetic spectrum.

The cognitive radio, built on a software-defined radio, is de-fined as an intelligent

wireless communication system that is aware of its environment and uses the

methodology of understanding- by-building to learn from the environment and

adapt to statistical variations in the input stimuli, with two primary objectives in

mind:

highly reliable communication whenever and wherever needed;

efficient utilization of the radio spectrum.

Electromagnetic radio spectrum is a natural resource, the use of which by transmitters

and receivers is licensed by governments. In November 2002, the Federal

Communications Commission (FCC) published a report prepared by the Spectrum-Policy

Task Force, aimed at improving the way in which this precious resource is managed in

the United States.

Spectrum Hole

A spectrum hole is 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.

2

Spectrum utilization can be improved significantly by making it possible for a

secondary user (who is not being serviced) to access a spectrum hole unoccupied

by the primary user at the right location and the time in question. Cognitive radio

, inclusive of software-defined radio, has been proposed as the means to promote

the efficient use of the spectrum by exploiting the existence of spectrum holes.

Characteristics of CR

Two main characteristics of the cognitive radio are cognitive capability and

Reconfigurability

Cognitive capability: Cognitive capability refers to the ability of the radio

technology to capture or sense the information from its radio environment.

Reconfigurability: The cognitive capability provides spectrum awareness

whereas reconfigurability enables the radio to be dynamically programmed

according to the radio environment.

Comparison SDR Vs Cognitive Radio

SDR Cognitive radio

A radio that includes a transmitter in which

the operating parameters of frequency

range, modulation

type or maximum output power (either

radiated or conducted), or the

circumstances under which the

transmitter operates can be altered by

making a change in software without

making any changes to

hardware components that affect the RF

emissions.

A radio or system that senses and is aware

of its operational environment and can be

trained to dynamically and autonomously

adjust its radio operating parameters

accordingly.

It should be noted that ―cognitive‖ does not

necessarily imply relying on software. For

example, cordless telephones (no software)

have long been able to select the best

authorized channel based on relative

channel availability.]

Cognitive Radios

A CR has the following characteristics: sensors creating awareness of the environment,

actuators to interact with the environment, a model of the environment that includes state

or memory of observed events, a learning capability that helps to select specific actions

or adaptations to reach a performance goal, and some degree of autonomy in action.

Example: The first examples of CRs were modeled in the Defense Advanced Research

Projects Agency (DARPA) NeXt Generation (XG) radio development program.

3

Adaptive Radio

A radio that supports multiple channel bandwidths is not adaptive, but a radio that

changes instantaneous bandwidth and/or system timing parameters in response to offered

network load may be considered adaptive. If a radio modifies intermediate frequency (IF)

filter characteristics in response to channel characteristics, it may be considered adaptive.

In other words if a radio makes changes to its operating parameters, such as power level,

modulation, frequency, and so on, it may be considered an adaptive radio.

Example: Digital European Cordless Telephone (DECT)

Aware Radios

A voice radio inherently has sensing capabilities in both audio (microphone) and RF

(receiver) frequencies. When these sensors are used to gather environmental information,

it becomes an aware radio. The local RF spectrum may be sensed in pursuit of channel

estimates, interference, or signals of interest. Audio inputs may be used for

authentication or context estimates or even natural language understanding or aural

human–machine interface (HMI) interactions. Added sensors enable an aware radio to

gather other information, such as chemical surroundings, geolocation, time of day,

biometric data, or even network quality of service (QoS) measures. The key characteristic

that raises a radio to the level of aware is the consolidation of environmental information

not required to perform simple communications. Utilization of this information is not

required for the radio to be considered aware.

Example: One example of an aware radio is the code division multiple access (CDMA)

based cellular system.. This system is aware of QoS metrics and makes reservations of

bandwidth to improve overall QoS.

Classification /Types of Awareness

Awareness means the understanding of the situation. Both geographical and RF

environment-related information (such as radio propagation characteristics, waveform,

and spectral regulations) play major parts in cognitive radio knowledge. In addition,

policy, goals, and contexts are also important issues for the cognitive radio.

(a) Location awareness: The cognitive radio knows where it is, in the form of latitude,

longitude, and altitude, or relative location to some reference nodes. Location awareness

concept for wireless systems is used for positioning, tracking, and location-based services

(LBS).

(b) Geographical environment awareness: The cognitive radio knows the terrain and

geographical information related to the radio propagation and channel characteristics.

This awareness is critically important for a cognitive radio to choose the appropriate

spectrum, channel model, or RAT, antenna configuration, and networking techniques.

4

(c) RF environment and waveform awareness: The cognitive radio knows the spectrum

utilization, the existence of PUs and/or SUs, the topology of the user group, the

interference profile, and other RF characteristics that may be of concern.

(d) Mobility and trajectory awareness: The cognitive radio knows its moving speed and

direction. For example, in conjunction with geographical awareness, the cognitive radio

can know it is moving south along Main Street at a speed of 45 miles per hour, and it can

―foresee‖ the radio environment ahead, such as the available channel after the user passes

over the next hill or the radio standards supported along the route.

(e) Power supply and energy efficiency awareness: The cognitive radio knows the source

of its power supply, the remaining battery life, and the energy efficiency of alternative

adaptation schemes.

(f) Regulation awareness: The cognitive radio knows the spectrum allocation and

emission masks at specific locations and frequency bands, which are regulated by

government authorities such as the US Federal Communications Commission (FCC).

(g) Policy awareness: The cognitive radio knows the policy defined by the user and/or

the service provider. For example, the user may prefer to use the wireless local area

network (WLAN) from a specific service provider at some locations for quality of service

(QoS) or security reasons.

(h) Capability awareness: The cognitive radio knows its own capabilities as well as those

of its team members and/or the network. Such awareness may include knowing which

waveforms are supported, the maximum transmit power, and the sensitivity of the

cognitive radio.

(i)Mission, context, and background awareness: The cognitive radio understands the

intent of the user, and knows what mode and volume of traffic it is going to generate and

what the impact of that traffic will be to the local networks. The cognitive radio

understands the QoS requirements, and how overhead activities may trigger additional

network traffic and latency.

(j) Priority awareness: The cognitive radio knows the user‘s priorities and habits. For

example, the user may prefer to use low-cost services whenever possible (e.g., to switch

to WLAN from a third-generation (3G) system when entering a Wi-Fi® zone), or may

prefer reliability over cost.

(k) Language awareness: The cognitive radio knows the signs, ontologies, and etiquette

used among cognitive radios to communicate with each other.

(l) Past experience awareness: The cognitive radio remembers the past experience

and learns from it

5

3.1 Making Cognitive Radio –Self Aware

Today‘s digital radios have considerable flexibility, but they have little

computational intelligence. For example, the equalizer taps of a GSM SDR reflect

the channel impulse response.

If the network wants to ask today‘s handsets ―How many distinguishable

multipath components are in your location?‖ two problems arise.

First, the network has no standard language with which to pose such a question.

Second, the handset has the answer in the structure of its time-domain equalizer

taps internally, but it cannot access this information.

It has no computationally accessible description of its own structure. Thus, it does

not ―know that it knows.‖ It cannot tell an equalizer from a vocoder. To be termed

―cognitive,‖ a radio must be self-aware. It should know a minimum set of basic

facts about radio and it should be able to communicate with other entities using

that knowledge. For example, it should know that an equalizer‘s time domain taps

reflect the channel impulse response.

To plan and schedule a task for execution, a PDA must know whether the task is

within its own capabilities. It needs to understand what it does and does not know,

as well as the limits of its capabilities. This is referred to as self-awareness.

For instance, the radio should know its current performance, such as bit error rate

(BER), signal-to-interference and noise ratio (SINR), multipath, and others. In a

more advanced case, the agent might need to reflect on its previous actions and

their results. For instance, for the radio to assess its travel speed a fortnight ago

between locations A and B, it might be able to extract parameters from its log file

and do the calculation.

For the radio to decide whether it should search for the specific entries in the log

and then perform appropriate calculations (or simply guess), it needs to know the

effort required to perform such a task and the required accuracy of the estimate to

its current task.

The following basic functionalities a cognitive radio should include:

● information collection and fusion;

● self-awareness;

● awareness of constraints and requirements;

● query by user, self or other radio;

● command execution;

● dynamic interoperability at any stack layer;

● situation awareness and advise;

● negotiation for resources.

6

3.1.1 ONTOLOGY-SELFAWARENESS

An ontology is an explicit mechanism for capturing the basic terminology and

knowledge (the concepts) of a domain of interest as well as the relationships

among the concepts . Ontologies are an increasingly important mechanism for

the integration of disparate software systems. Indeed, a shared ontology is a

fundamental prerequisite for meaningful communication between systems.

The advantages include support for interoperability, flexible querying, run-time

modifiability, validation against specifications, and consistency checking.

In the case of software-defined radio, ontologies offer the additional advantage of

self-awareness: communication nodes can understand their own structure and can

modify their functioning at run-time. Furthermore, nodes can query the

capabilities and current state of other nodes, allowing them to modify the

processing of packets during a communication session both at the source and the

destination. The ontology specifies not only the structure of communication

packets but also the processing of those packets according to the communication

protocol.

3.1.2 COGNITIVE RADIO FRAMEWORK

The radio hardware consists of a set of modules: antenna, RF section, modem,

INFOSEC module, baseband/ protocol processor, and user interface. This could

be a software radio, SDR, or PDR. In the figure, the baseband processor hosts the

protocol and control software.

The modem software includes the modem with equalizer, among other things. In

addition, however, a cognitive radio contains an internal model of its own

hardware and software structure. The model of the equalizer shown would contain

the codified knowledge about equalizers, including how the taps represent the

channel impulse response.

Variable bindings between the equalizer model and the software equalizer

establish the interface between the reasoning capability and the operational

software. The model-based reasoning capability that applies these Radio

Knowledge Representation Language (RKRL) frames to solve radio control

problems gives the radio its ―cognitive‖ ability.

The approach, then, is to represent radio knowledge in RKRL and to structure

reasoning algorithms to use that knowledge for the control of software-radios. The

radio‘s model of itself should contain a representation of its functions (e.g.

transmission, reception, coding, etc). The figure 3.1 shows CRFN

Fig.3.1 Cognitive Radio Frame Work (CRFN)

7

In addition, however, if the radio is to be context aware, it must interact

with the outside world. This is accomplished via the cognition cycle.

3.1.3 COGNITIVE ARCHITECTURE

The cognitive radio architecture shown in shown in Figure 3.2. Here, the intelligent core

of the cognitive radio exists in the cognitive engine. The cognitive engine performs the

modeling, learning, and optimization processes necessary to reconfigure the

communication system, which appears as the simplified open systems interconnection

(OSI) stack. The cognitive engine takes in information from the user domain, the radio

domain, the policy domain, and the radio itself. The user domain passes information

relevant to the user‘s application and networking needs to help direct the cognitive

engine‘s optimization. The radio domain information consists of radio frequency (RF)

and environmental data that could affect system performance such as propagation or

interference sources. The policy engine receives policy-related information from the

policy domain. This information helps the cognitive radio decide on allowable (and legal)

solutions and blocks any solutions that break local regulations.

8

Fig.3.2 Generic Cognitive Radio Architecture

The first problem in dealing with cognition in a system is to understand (1) what

information the intelligent core must have and (2) how it can adapt. In radio, we can think

of the classical transmitters and receivers as having adjustable control parameters (knobs)

that control the radio‘s operating parameters.

PHY- and Link-Layer Parameters Knobs The knobs of a radio are any of the parameters that affect link performance and radio

operation. Some of these are normally assumed to be design parameters, and others are

usually assumed to be under real-time control of either the operator or the radio‘s real-

time control processes. Figure 3.3 shows a simple system diagram of the PHY- and link-

layer portions of a transmitter. In the PHY layer, center frequency, symbol rate, transmit

power, modulation type and order, pulse-shape filter (PSF) type and order, spread

spectrum type, and spreading factor can all be adjusted. On the link layer are variables

that will improve network performance, including the type and rate of the channel coding

and interleaving, as well as access control methods such as flow control, frame size, and

the multiple access

9

Table 3.1 Knobs and Meters

Table 3.2 Knobs and Meters for GNU Radio Simulation

10

Fig. 3.3 PHY-link Layer of Generic Transmitter

Meters

In optimization theory, the meters represent utility and cost functions that must be

maximized or minimized for optimum radio operation. All of these performance analysis

functions constitute objective functions. In an ideal case, we can find a single-objective

function whose maximization or minimization corresponds to the best settings. However,

communication systems have complex requirements that cannot be subsumed into a

single-objective function, especially if the user or network requirements change. Metrics

of performance are as different for voice communications as they are for data, e-mail,

web browsing, or video conferencing. The types of meters represent performance on

different levels. On the PHY layer, important performance measurements deal with bit

fidelity. The most obvious meters are the signal-to-noise ratio, or a more complex SINR.

The SINR has a direct consequence on the bit error rate (BER), which has different

meanings for different modulations and coding techniques, usually nominally determined

by the SINR ratio, Eb/(N0I0), where Eb is energy per bit, N0 is noise power per bit, and I0

is interference power per bit. On the link layer, the packet fidelity is an important metric,

specifically the packet error rate (PER). There are more external metrics to consider as

well, such as the occupied bandwidth and spectrum efficiency (number of bits per hertz)

and data rate.

Internal metrics also are involved in decision-making. To decrease the FER, a stronger

code can be used but this increases the computational complexity of the system,

increasing both latency as well as the power required to perform the more complex

forward error correction (FEC) operation. Decreasing the symbol rate or modulation

order will decrease the FER as well without increasing the demands of the system, but at

the expense of the data rate

11

GNU Radio

GNU Radio is the GNU (the clever recursive acronym for ―GNU is Not Unix") project of

the Free Software Foundation (FSF) to provide a General Purpose Processor (GPP)-based

open source software defined radio. The GNU Radio is a pure software package that

provides signal processing blocks, discrete components to perform a specific task. Each

of these components is a C++ class which a developer can connect to other blocks to

create a own graph. A block can be a source with only output ports, a sink with just input

ports, or a general block with both inputs and outputs. Currently, the GNU Radio

supports many signal processing blocks and a number of waveforms.

3.1.4 Modeling Outcome as a Primary Objective

The basic process followed by a cognitive radio is that it adjusts its knobs to

achieve some desired (optimum) combination of meter readings. Rather than

randomly trying all possible combinations of knob settings and observing what

happens, it makes intelligent decisions about which settings to try and observes

the results of these trials.

Based on what it has learned from experience and on its own internal models of

channel behavior, it analyzes possible knob settings, predicts some optimum

combination for trial, conducts the trial, observes the results, and compares the

observed results with its predictions, as summarized in the adaptation loop of

Figure 3.4.

Fig. 3.4 Adaptation Loop

3.1.4.1Definition of MODM and Its Basic Formulation

Multi-Objective Decision-Making (MODM) theory is used to analyze the radio‘s

performance.

At their core, MODMs are a mathematical method for choosing the set of parameters that

best optimizes the set of objective functions. The below equation is a basic representation

of a MODM method :

12

Here all objective functions are defined to either minimize or maximize y, depending on

the application. The x values (i.e., x1, x2, etc.) represent inputs and the y values represent

outputs. The equation provides the basic formulation without prescribing any method for

optimizing the system.

3.1.4.2 The Pareto-Optimal Front: Finding the Nondominated Solutions

In an MODM problem space, a set of solutions optimizes the overall system, if there is no

one solution that exhibits a best performance in all dimensions. This set, the set of

nondominated solutions, lies on the Pareto-optimal front (hereafter called the Pareto

front). All other solutions not on the Pareto front are considered dominated, suboptimal,

or locally optimal.

The most important concept in understanding the Pareto front is that almost all solutions

will be compromises. There are few real multi-objective problems for which a solution

can fully optimize all objectives at the same time. This concept has been referred to as the

utopian point. Fig. 3.5 shows Pareto Front.

Fig. 3.5 Pareto Front of BPSK Curve

3.1.4.3 GA Approach to the MODM

Analyzing the radio by using a GA is inspired by evolutionary biological techniques.

If we treat the radio like a biological system, we can define it by using an analogy

to a chromosome, in which each gene of the chromosome corresponds to some

trait (knob) of the radio. We can then perform evolutionary-type techniques to

create populations of possible radio designs (waveform, protocols, and even

hardware designs) that produce offspring that are genetic combinations of the

parents. In this analogy, we evolve the radio parameters much like biological

evolution to improve the radio ―species‖ through successive generations, with

selection based on performance guiding the evolution. The traits represented in

the chromosome‘s genes are the radio knobs, and evolution leads toward

improvements in the radio meters‘ readings.

13

GAs are a class of search algorithms that rely on both directed searches

(exploitation) and random searches (exploration). The algorithms exploit the

current generation of chromosomes by preserving good sets of genes through the

combination of parent chromosomes, so there is a similarity between the current

search space and the previous search space. If the genetic combination is from

two highly fit parents, it is likely that the offspring is also highly fit.

The algorithms also allow exploration of the search space by mutating certain

members of the population that will form random chromosomes, giving them the

ability to break the boundaries of the parents‘ traits and discover new methods

and solutions.

1. Initialize the population of chromosomes (radio/modem design choices)

2. Repeat until the stopping criterion

(a) Choose parent chromosomes

(b) Crossover parent chromosomes to create offspring

(c) Mutate offspring chromosomes

(d) Evaluate the fitness of the parent chromosomes

(e) Replace less fit parent chromosomes

3. Choose the best chromosome from the final generation

3.1.4.4 Multi-objective GA for Cognitive Radios

Cognition Loop

The primary goal of the cognitive engine is to optimize the radio, and the secondary

functions are to observe and learn in order to provide the knowledge required to perform

the adaptation. A cognitive radio becomes a learning machine through a tiered algorithm

structure based on modeling, action, feedback, and knowledge representation, as shown

in the cognition loop of Figure 3.6.

Fig. 3.6 Cognition Loop

14

Another goal of the modeling system is to monitor the data sent by the user and use it to

determine the QoS parameters the cognitive radio must provide. This is a learning

domain concept that a neural network could be employed to help solve, and the research

group at VT is currently investigating the possibilities. The regulatory policy modeling

and interpretation will come from other efforts, such as the Defense Advanced Research

Projects Agency (DARPA) NeXt Generation (XG) project .

The cognitive system module (CSM) is responsible for learning and the wireless system

genetic algorithm (WSGA) handles the behavioral adaptation of the radio, based on what

it is told to do by the CSM. The modeling system observes the environment from many

different angles to develop a complete picture. The CSM holds two main learning blocks:

the evolver and the decision maker, which takes feedback from the radio that allows the

evolver to properly update the knowledge base to respond to and direct system behavior.

Radio Parameters as Genes in a Chromosome

Knobs with smaller search spaces can be realized with smaller genes, and some traits can

be combined. Modulation is segmented into the type of modulation—such as amplitude

shift keying (ASK), frequency shift keying (FSK), phase shift keying (PSK), QAM, and

so forth—and the order (2-, 4-, 8-, 16-point constellations, etc.). A 16-bit gene

representation (65,536 possibilities) is far too large for either type or order, but the gene

could be split into two 8-bit pieces.

3.2 Cognitive Techniques – Classification

1) Position Awareness Awareness

2) Environment Awareness

3.2.1 Position Awareness (NOV 2016, APRIL 2017)

Location and environment awareness are two prominent features of cognitive radios and

networks enabling them to interact with and learn the operating environment.

Features of Cognitive Radio Receiver -Cognitive radio transceiver uses sensing,

awareness, learning, decision, adaptation, reconfigurability, goal driven autonomous

operation. The physical place occupied by an object (e.g. designated user) is referred as

location. Furthermore, position term is defined as the coordinates of a single point in

space that represents the location of an object. On the other hand, environment is briefly

defined as the volume oriented at a specific location. Detailed definition of environment

is provided in a later section.

3.2.1.1Location and Environment Awareness in Nature

15

Location and Environment Awareness concept can be defined as being cognizant of

location and associated environment. The creatures in the nature have been considered as

models for most of the innovations in science history. Similarly, most of the creatures in

the nature have already location and environment awareness capabilities to some extent

and they have been considered as models for incorporating such capabilities to electronic

devices. For instance, bat has location and environment awareness capability, which is

known as echolocation, for the navigation and prey capturing. The bats emit high

frequency ultrasonic signals (20–200 KHz) from their mouths (transmitter) and listen to

the echoes from the environment using their ears (receivers). The received echoes are

processed by these animals for different purposes such as navigation, object recognition,

and ranging.

3.2.1.2 Location and Environment Awareness in Wireless Systems

(NOV 2016, APRIL 2017)

Location and environment awareness in wireless systems Location and environment

awareness features can be introduced to electronic systems and such approaches have

been investigated extensively for biologically inspired robotics . However, this is not the

case for wireless systems. Utilization of location and environment information in wireless

systems have been limited to positioning systems and location based systems (LBS).

Nevertheless, the aforementioned advanced location and environment awareness

capabilities of human being or bat can be introduced to wireless systems . This can be

accomplished by using cognitive radio technology According to the definition, cognitive

radio has sensing, awareness, and adaptation features, which are the main ingredientsof

location and environment awareness conceptual model shown in Fig 3.7 .

Fig. 3.7 conceptual Model of Location and Environment Awareness Cycles

16

3.2.1.3 Architecture – Location and Awareness Engine (NOV 2016, APRIL 2017)

Fig. 3.8 Architecture - Location and Awareness Engine

The above Fig. 3.8 shows architecture of location and awareness engine where it has

sensing interface, cognitive engine core and spectrum awareness engine.

Sensing Interface

Sensing process is composed of mainly two components, which are sensors and

associated data post-processing methods. Similar to the creatures in the nature, different

17

sensors have been used in wireless systems for sensing. Sensors are utilized to convert

the signals acquired from environment to electrical signals so that cognitive radios can

interpret. The acquired signals can be in different format such as electromagnetic, optic,

and sound. Therefore, sensors can be categorized under three types; electromagnetic,

image, and acoustic sensors. The corresponding data post-processing algorithm for each

sensing technique is different. Inspiring from the sensing features of the creatures, it is

possible to classify the sensing mechanisms in cognitive radios under three main

categories based on the type of sensors used: Radiosensing, Radiovision, and

Radiohearing.

Radiosensing is a sensing technique utilizing electromagnetic sensors and the associated

post-processing schemes. Similarly, radiovision is a sensing approach using image

sensors and the corresponding post-processing schemes. Finally, radiohearing is a sensing

method employing acoustic sensors and the associated post-processing schemes.

Radiosensing sensors

Although light can be considered as an electromagnetic wave, the image sensors are used

here. The most widely used radiosensing (electromagnetic) sensor in wireless systems is

antenna, which is the focus of this section. Antenna is a transducer that converts

electromagnetic signal into electrical signals and vice versa. For instance, in antenna-

based wireless positioning systems, location information is estimated from the received

signal statistics such as time-of-arrival (TOA), receive signal strength (RSS), and

angle-of-arrival (AOA)

Radiovision sensors

Radiovision sensor such as image sensor is a device that captures optic signals from the

environment and converts them to electrical signals in order to construct the

corresponding image. These sensors have been already used in different areas such as

digital cameras and computer vision systems.

Radiohearing sensors

One of the radiohearing sensors is acoustic sensor, which is a transducer that converts

acoustic signals into electrical signals and vice versa. This type of sensor has already

been used in different wireless systems. The main idea behind acoustic technique is

utilizing sound propagation to navigate, detect objects, and communicate.

Radio hearing Methods

Radio hearing based position methods utilize acoustic sensors for interacting with

environments. Similar to radio based position sensing techniques, radio hearing based

position methods can be implemented using three group schemes range-based, range-free

and pattern matching based techniques.

18

3.2.2 Environment Awareness (APRIL 2017, NOV 2017)

Environment awareness is one of the most substantial and complicated task in cognitive

radios since channel environment is the bottleneck of wireless systems. Creatures with

environment awareness capabilities such as human being and bats can be considered as

models for the realization of environment awareness in cognitive radios. For instance,

human being has different sophisticated senses such as observing and learning the

surrounding environment and bats utilize their echolocation systems for object and

environment identification, and target detection and tracking. Environment awareness

engine governs the technique of environment awareness as given in Fig. 3.9.

Fig.3.9 Conceptual Model for Environment Awareness Engine

Topographical Information

Topography of a local region provides information about not only the relief (Earth

surface features), but also vegetation, human –made structures, history of a particular

region. Assuming that central environment awareness engine has topographical map

including the aforementioned information, numerous advanced location based services

(LBS) can be developed. There are some efforts towards the realization of topographical

map such as Google Maps TM.

Object Recognition and Tracking

19

Objects are defined as the human-made entities present in the target local environment

temporarily or permanently in this study. The large and permanent human-made

structures such as buildings and bridges are considered as part of topography of

environment, hence, such human-made structures are included in the topographical

information. On the other hand, relatively small and movable human made entities such

as vehicles, home and office appliances are considered as objects. Object detection,

identification and tracking are important features of environment awareness engine since

they can affect the dynamic of environment.

Meteorological information

This entity provides information on the weather of target local region, which can affect

the signal propagation. The current and future weather parameters such as rain, snow,

temperature, humidity, and pressure can be acquired either using radio auxiliary sensors

or from central cognitive base station. By having current and forecasted meteorological

information, cognitive radio can adapt itself accordingly.

For instance, rain can have significant affects on the performance of broadband fixed

wireless access links (e.g. Fixed WiMAX), especially operating at higher carrier

frequencies. One of the performance parameters that can be affected from rain is the

carrier-to-interference ratio (C/I) and this performance metric depends on the rain

intensity of the location of desired signal path and interferer signal paths.

Propagation Characteristics

This entity provides information on the characteristics of signal progression through a

medium (channel environment). Basically, propagation characteristics of channel

environment shows that how the channel affects transmitted signal. The statistical

characteristics of wireless channel are given by two group of characteristics such as large

scale and small scale.

3.3 Global Positioning System(GPS)

GPS is without a doubt the best-known location system in the world. GPS is a satellite

navigation system funded and controlled by the US Department of Defense (DoD) and

the Department of Commerce (DoC). The system comprises a constellation of satellites,

ground control stations, and GPS receivers. At most points on Earth (other than in the

deep urban canyons between skyscrapers), there is a high probability of line-of-sight

(LOS) contact with multiple GPS satellites. Given LOS with four or more satellites,

three-dimensional (3-D) position and time can be measured.

Satellite System Architecture The GPS system is readily divided into three segments: space, control, and user.

Space Segment

20

Not counting orbiting spares, there are normally 24 active GPS satellites. The orbital

period is nominally 12 hours. The 24 satellites are distributed evenly in six orbital planes

with 60-degree separation between each of the four satellites in each plane. The

inclination is about 55 degrees off the equator. This geometric distribution provides

between five and eight satellites in view from any point on Earth. A line-of-sight (LOS—

meaning unobstructed) view of four or more satellites is needed to process the signals and

calculate location.

Control Segment

Ground tracking stations are positioned worldwide to monitor and operate the

constellation of GPS satellites. The master control station is located at Schriever Air

Force Base in Colorado. The stations monitor the satellites‘ signals, incorporate them into

orbital models, and calculate ephemeris data that are transmitted back to the space

vehicles. Ephemeris data are, in turn, transmitted to GPS receivers.

User Segment

GPS receivers and their operators form the user segment. The receivers process the

signals from four or more satellites into 3-D position and time. In the differential mode, a

reference GPS receiver communicating with another GPS receiver, where the position of

one node is known to high accuracy, can then improve upon the inherent accuracy of

another stand-alone GPS receiver. GPS receivers also produce a precise one-pulse-per-

second signal

Accuracy Obtained and Coordinate System

The two classes of GPS geolocation capabilities are the precise positioning service (PPS)

and the standard positioning service (SPS). The PPS capability requires cryptographic

technology and achieves 22 m horizontal accuracy, 27.7 m vertical accuracy, and 200 ns

time accuracy. The SPS capability is available to any user and achieves 100 m horizontal

accuracy,

GPS Satellite Signals

GPS satellites transmit two spread spectrum signals, one at 1575.42 MHz (L1 for SPS)

and the other at 1227.60 MHz (L2 for PPS). There is a unique 1-MHz-wide, 1023-chip-

long coarse acquisition pseudorandom (Gold code) spreading code for each satellite. A

Gold code is a spreading code synthesized by exclusive ORing of the output of two linear

feedback shift register (LFSR) pseudorandom (PN) generators together. Each LFSR uses

a carefully selected tap and initialization to assure that the autocorrelations of the Gold

code are small at all delays except perfect time alignment, and are not confused with

cross correlation from other spreading codes.

GPS Navigation Message

21

A data frame (1500 bits) is transmitted every 30 seconds and consists of five 300-

bit subframes.

● Subframe 1 is Telemetry Word | Handover Word | Space Vehicle Clock Correction

Data.

● Subframe 2 is Telemetry Word | Handover Word | Space Vehicle Ephemeris Data part

1.

● Subframe 3 is Telemetry Word | Handover Word | Space Vehicle Ephemeris Data part

2.

● Subframe 4 is Telemetry Word | Handover Word | Other Data.

● Subframe 5 is Telemetry Word | Handover Word | Almanac Data for All Space

Vehicles.

Signal Processing of GPS Signals

The GPS receiver correlates known coarse acquisition spreading codes (with a 1-

millisecond period of 1023 chips) from each of the GPS satellites with the

processed signal from the GPS satellites. The known spreading codes are very

short and may be generated or stored in memory. Because each satellite uses a

different Gold-word spreading code, when the receiver has a peak correlation it

knows which satellite sent the signal. This despreading produces a full-power

signal.

This signal is tracked using a phase locked loop (PLL), and the 50 Hz navigation

message is demodulated from each satellite. Time of Arrival (ToA) information is

extracted when a correlation peak is measured.

Given the ToA information measured from the correlation peak and the GPS time

embedded in the signal, the GPS receiver can measure range to each satellite in

view. An intersection of multiple range spheres determines where the GPS

receiver is located. Four satellites must be in view to estimate x, y, and z

coordinates along with a time estimate. A precise estimate of the position of each

space vehicle in view is determined from the broadcast ephemeris data.

Reference Axes

The x, y, and z estimates are computed in Earth-centered fixed (ECF) coordinates.

ECF is a right-hand orthogonal Cartesian coordinate system with the origin at the

center of Earth, the z-axis increasing through the rotational North Pole of Earth,

the x-axis increasing through the prime meridian (Greenwich, England) at latitude

zero and longitude zero, and the y-axis increasing through 90 degrees longitude

and zero degrees latitude.

Differential GPS

Position accuracy may be improved through the use of differential GPS

processing. Correcting bias errors using a known location accomplishes this. A

known location receiver measures its position and calculates a correction for each

satellite that is passed to other GPS receivers in the local area. This is a

22

sophisticated solution that requires more capability at both the reference and

mobile GPS receiver and a data link between the reference receiver and the

mobile receiver. Another form of differential GPS is the measurement of carrier

phase. This capability is used in surveying and can generate subfoot accuracies

over short distances.

Coordinate System Transformations

Satellite positions and GPS receiver position are reported in ECF coordinates (x, y, z).

Navigators, however, are frequently interested in latitude, longitude, and height. is the

conversion from ECF to latitude, longitude, and height (_, , and h, respectively):

GPS Geolocation Summary

A GPS receiver in a CR is one way to let a CR know where it is. Adding this information

to an inter-radio data stream enables other CRs to know where a particular radio is

located.

Geolocation-Enabled Routing

In addition to the user‘s geolocation support, the radio network functionality may benefit

from geolocation knowledge. Chapter 11 provides a detailed analysis of the radio

environment map (REM), which is an infrastructure to enable cognitive network

functionality.

However, in addition to the REM, the routing functionality of an ad hoc network, and the

cellular handover of cellular networks, may be improved by explicit geolocation

knowledge, velocity vector knowledge, and planned route path knowledge.

23

One can envision that an ad hoc network could use destination location addressing rather

than medium access control (MAC) and IP addressing. In such a network, messages

propagate only to nodes that know they are on a path to that location.

Additional Geolocation Approaches

Terrestrial radio geolocation is accomplished through one or more of the following

techniques: ToA, TDoA, angle of arrival (AOA), or received signal strength (RSS).

Normally, a strong LOS signal is needed for accurate measurements.

Time-Based Approaches

Time-based approaches for geolocation may be divided into ToA and TDoA approaches.

Both approaches require a high-resolution system clock. ToA and TDoA approaches to

geolocation are based on the propagation speed of light, which is defined to be 3 x108m/s,

and ―straight line‖ LOS propagation paths so that the signal time delay relates directly to

the LOS distance. Propagation is analyzed by multiple receivers or by multiple antennas

converting time delay to phase difference, or time difference, or frequency shift. These

are then translated into equations of range, range ratio, AOA, and/or other parameters.

Time of Arrival (ToA) Approach

The ToA approach is centered on the ability to time-tag a transmitted signal and measure

the exact ToA of that signal at a receiver. The propagation time, at the speed of light

assuming LOS propagation, is a direct measure of the propagation distance. This provides

a receiver with an iso-range sphere for a given transmitted and received signal. If

multiple receivers at known locations receive the same signal, generally at different

times, the multiple iso-range spheres intersect at the transmitter‘s location. It requires

four receivers to geolocate one transmitter in three dimensions.

Long Range Navigation (LORAN)

LORAN systems transmit a known burst signal from multiple transmitters with a known

and published periodicity. Furthermore, the exact location of each transmitter is known.

Three such transmitters cooperate to enable TDoA measurements. Ships at sea receive

these transmissions and measure the time difference between each received signal. From

these time differences, ships are able to calculate the TDoA hyperbolas.

Timing Estimates

Time can be derived from a number of sources, including atomic clocks, standard clocks,

GPS time, disciplined GPS, phase estimation techniques, and correlation techniques in

wideband transmission environments. The accuracy of the time estimate directly

influences the ranging estimate accuracy at approximately 1 ns 30cm_1

. A popular way to

obtain Time Direction of Arrival (TDoA) information is through cross correlation. At a

24

fixed point in time (maybe on the GPS one pulse per second boundary), two stations

digitize their received signals and pass the measurements to a common processing

location. The processor calculates the cross correlation of the two signals. The peak of the

cross correlation reveals the TDoA.

Angle of Arrival (AOA) Approach

The AOA approach requires an antenna array at the receivers. Multiple receivers estimate

the AOA of a signal. Combining the bearing to the signal with the known location of

multiple receivers yields an intersection point of the transmitter. This is simple

triangulation.

Geometry of AOA Approach Geometry, of course, affects how well any time or angle measurements work, including

the AOA approach in Fig.3.10 If the ―baseline‖ of receivers does not have a sufficiently

large angle of observation, the result is poor. If the accuracy of the AOA sensors is poor

and the range to the receiver is long, the ―angular dispersion‖ impacts the measurement.

However, the computations are simple if the AOA is known to high accuracy and the

interprocessor communications data volume is low.

Fig.3.10 Angle of Arrival Approach

3.4 Radio Environment Map (REM)

The radio environment map (REM) has been proposed as a vehicle of network

support for cognitive radio. The REM is an abstraction of real-world radio

scenarios; it characterizes the radio environment of cognitive radios in multiple

domains, such as geographical features, regulation, policy, radio equipment

capability profile, and radio frequency (RF) emissions. The REM, which is

essentially an integrated spatiotemporal database, can be exploited to support

cognitive functionality of the user equipment, such as situation awareness (SA),

reasoning, learning, and planning, even if the subscriber unit is relatively simple.

The REM can also be viewed as an extension to the available resource map

25

(ARM), which is proposed to be a real-time map of all radio activities in the

network for cognitive radio applications in unlicensed wide area networks

(UWANs).

From the cognitive radio user‘s point of view, the network support to the

cognitive radio can be classified into two categories: internal network support and

external network support. The internal network refers to the radio network with

which the cognitive radio is associated. Along with various communication

services, the internal network can provide some cognitive functionality

The external network refers to any other networks that can provide meaningful

knowledge to support the cognitive functionalities of the radio. For example, a

separate sensor network could be dedicated to gather information for cognitive

radio networks. The external network could be a legacy network or other

cognitive radio networks.

Both internal and external networks can contribute to building up the REM and can be

employed in a collaborative way. For instance, location information needed for a

cognitive radio can be obtained either from internal network support through a network-

based positioning method for indoor scenarios, or from external network support through

the global positioning system (GPS) for outdoor scenarios

As depicted in Figure 3.11, network support can be realized through a global REM and

local REMs. In this figure 3.11, the cognitive radio is symbolized as a brain -

empowered radio. The global REM maintained on the network keeps an overview of the

radio environment, and the local REMs stored at the user equipment only present more

specific views to reduce the memory footprint and communication overhead. The local

REMs and global REM may exchange information in a timely manner so as to keep the

information stored at different entities current. In this figure 3.11, a regional REM can be

aggregated from the combined experiences of several local REMs.

Fig. 3.12 Radio Environment Map

26

REM for Cognitive Radio

REM is a comprehensive spatiotemporal database and an abstraction of realworld

radio scenarios.

The REM is essentially a database that stores information utilized in the decision

process of a cognitive network. As such, various elements in conjunction with the

REM are needed to create the cognitive network. These elements include the

learning, reasoning, and decision algorithms.

Similar to how a city map helps a traveler, the REM can help the cognitive radio

to know the radio environment by providing information on, for example, spectral

regulatory rules and user-defined policies to which the cognitive radio should

conform; spectrum opportunities; where the radio is now and where it is heading;

the appropriate channel model to use; the expected path loss and signalto- noise

ratio (SNR); hidden nodes present in the neighborhood; usage patterns of PUs1

and/or secondary users (SUs); and interference or jamming sources. Fig. 3.13

shows REM for Cognitive Radio

Fig.3.13 REM for Cognitive Radio

REM Design

The REM plays an important role in the cognition cycle of cognitive radio, as illustrated

in Figure 3.14. Both direct observations from the radio and knowledge derived from

network support can contribute to the global and/or local REM. The radio‘s environment

awareness can be obtained from direct observation, such as spectrum sensing, and/or

from the REM. Reasoning and learning help the cognitive radio to identify the specific

radio scenario, learn from past experience and observations, and make decisions and

27

plans to meet its goals. The global REM and/or the local REM should be updated once

action is taken or scheduled by the radio to keep the REM‘s information current as given

in Fig. 3.14

Fig. 3.14 Role of REM for Cognitive Radio

The REM contains information at multiple layers, as illustrated in Figure 3..15. By

integrating various databases, the REM enables or supports cognitive functionality for

radios with different levels of intelligence. The REM helps a cognitive radio to be aware

of situations and make optimal adaptations according to its goals; for legacy or hardware

reconfigurable radios, the REM facilitates smart network operations by providing

cognitive strategies to the network radio resource management control. Just like the city

map that is informative to every traveler, no matter whether driving a car or taking the

bus, the REM is transparent to the specific radio access technology (RAT) to be

employed regardless of whether the subscriber radio is cognitive or not.

Fig. 3.15 Databases for REM

With the help of the REM, a radio can become cognitive of performance metrics, the

application, topology, and network (routing), as well as the medium access control

(MAC) and physical (PHY) layers of communication stacks under different and varying

radio environments. For example, if the radio is used on the battlefield, reliability and

security are of high importance. Therefore, special source coding, encryption, anti-

jamming channel coding, frequency planning, and routing algorithms could be employed

28

accordingly. The REM can support various network architectures: centralized,

distributed, or heterogeneous networks, or even point-to-point communications. It can

also support collaborative information processing among multiple nodes for obtaining

comprehensive awareness.

3.5 Optimization of Radio Resources (NOV 2017)

The objective space defnes the radio resources used to determine radio behavior. A radio

consumes resources while communicating, therefore depriving other radios access to

those same resources. Spectrum is the key communications resource and is a reusable

resource by sharing in space, time, and transmit power. Spectrum sharing and reuse is

accomplished through numerous techniques such as spatial distribution like cellular

infrastructures or beam-forming antennas. In both the time and frequency domains,

Dynamic Spectrum Access (DSA) technology is developing to provide intelligent

schemes that use spectrum during times when other users or primary users are silent.

Concepts such as an ultra-wideband underlay, spread spectrum, and interference

temperature are all methods that manage transmit power to allow coexistence with other

radio systems. The task is then to properly use the resources to provide appropriate

sharing among all radios while maintaining the proper level of QoS. Each user has a

different and subjective perspective on quality of service based on the radio's

performance. A user may require high data rates, low latency, or long battery life

depending on the situation for which he or she is using the radio service. Video

conferencing requires high data rates and low latency, while voice calls require low

latency but have signi_cantly relaxed requirements for average throughput. On the other

hand, checking stock prices or even email has low requirements for speed.

Each node in a network can look at resource allocation as an optimization problem with

two potential goals. First, it can attempt to optimize the use of the resources from the

perspective of maximizing its own use of resources and therefore its own ability to

communicate; this would be called a greedy approach. The other way is to look at

resource utilization from a needs perspective; that is, resources are sought only to support

the needs of the service. More resource utilization is wasteful while less harms the quality

of service. Resource allocation on either side of what is required is inefficient. Of course,

there is a third way of looking at resource allocation, and this is to look at it from a global

perspective where the utilization of resources by all nodes is taken into account. On the

whole optimization of radio resources is a Multiobjective optimization problem involving

objective functions.

3.5.1 Multi-objective Optimization: Objective Functions

Multi-objective optimization has a long history in mathematics, operations

research, and economics. Multi-objective decision making (MODM) is given as

29

The equation defines n dimensions in the search space where each objective function

)(

xfn

evaluates the nth objective. The set

x defines the set of input parameters that the

algorithm has control over, and

y is the set of objectives computed by the objective

functions. Both of these may be constrained to some space, X and Y, depending on real-

world constraints like available radio resources

y or radio capabilities

x . The solutions

to multi-objective problems lie on the Pareto front, which is the set of input parameters,

x , that defines the non-dominated solutions,

y in any dimension. The following are the

radio resources which needs optimization namely Bit Error Rate(BER), Signal to Noise

Interference Ratio (SINR), Bandwidth, Spectral Efficiency, Throughput, Power,

Computational complexity and Interference as shown in Figure 3.16.

Fig. 3.16 Radio Resources for Optimization

Bit Error Rate (BER)

Dependencies

30

Knobs: transmitter power, modulation type

Meters: noise power, channel type, path loss

objectives: bandwidth

Definitions

= energy per bit to noise energy ration (EbN0)

PT = transmit power (effective isotropic radiated power (EIRP)) (dBm)

L = estimated path loss (dB)

B = bandwidth in (Hz)

M = number of symbols in the modulation's alphabet

Rs = symbol rate (sps)

N0 = noise floor (J)

Bit error rate (BER) is an important objective for all digital communications

needs. It provides a baseline for the amount of information transferred, and so

understanding it in light of the design of a waveform under certain channel conditions is

therefore necessary. Unfortunately, BER calculations depend heavily on the type of

channel and type of modulation, and so the cognitive engine must know the formula for

each modulation type the radio is capable of using and the channel types it is likely to see

during operation. The following are BER formulas of relative cognitive radio systems

31

Signal to Interference Plus Noise Ratio (SINR)

Dependencies

Knobs: transmit power

Meters: noise power, path loss

Objectives: interference, bandwidth

Definitions

PT = EIRP (dBm)

L = estimated path loss (dB)

N = noise power (dBm)

I = interference power (dBm)

The following formula is for interference power, noise power and signal to interference

noise ratio

Bandwidth (Hz)

Dependencies

Knobs: modulation type, symbol rate, pulse shape filter

Meters: none

32

Objectives: none

Definitions

k = number of bits per symbol

Rs = symbol rate (sps)

α = property of the pulse shape filter (roll-off factor in RRC or bandwidth-time

product in a Gaussian filter)

Instead of using the raised cosine Nyquist pulse shaping, a root-raised cosine filter in

both the transmitter and receiver provides a more practical implementation. Over the air,

the pulse is shaped by a single RRC filter while the second RRC filter in the receiver

shapes the received signal as though it was passed through a single raised cosine filter to

reduce the Inter Symbol Interference (ISI). Many narrowband digital modulations use RC

pulses, including the M-PSK waveforms used in this work. For these, the approximate

null-to-null bandwidth can be calculated and the roll-o_ factor of the RRC fillter is

defined as.

Spectral Efficiency (bits/Hz)

Dependencies

Knobs: modulation type, symbol rate

Meters: none

Objectives: bandwidth

Definitions

k = number of bits per symbol

Rs = symbol rate (sps)

B = bandwidth (Hz)

Spectral Efficiency represents the amount of information transferred in a given channel

and is measured in bits per second per Hertz (bps/Hz). Spectral efficiency helps shape the

decision space by biasing the solution towards a symbol rate and modulation type that

provides high bandwidth efficiency and produce better data rates for a given spectrum.

ηs = Rsk/B

Throughput

Dependencies

Knobs: modulation type, symbol rate, number of bits per packet

33

Meters: none

Objectives: bit error rate

Definitions

l = number of bits per packet (bits)

Pe = bit error rate

Rb = bit rate (bps)

Rs = symbol rate (sps)

k = number of bits per symbol

Throughput is a measure of the amount of good information received. This definition

distinguishes throughput from data rate in that data rate is simply a measure of the rate

data arrives with no consideration for transmission errors.

The probability of a packet error, or the packet error rate, is shown in equation

Power

Dependencies

Knobs: Transmit power

Meters: none

Objectives: none

Definitions

PT = transmit power (dBm)

There are two ways to look at power as a resource. The first way is to think about power

in terms of how the radio transmitter uses the external power in the spectrum. In this

manner of speaking, power is a shared resource by all radio nodes, so radios should strive

to reduce their transmission power. This objective balances efforts to reduce BER or

maximize SINR. The transmitted power used here refers to power of the signal sent to the

antenna. In all of the calculations here, though, EIRP is an assumptions such as when

calculating the received power for the BER equations. In calculations, 0 dB gain antenna

is assumed. The assumption is based on the lack of the antenna as a knob or even a

parameter in the current analysis. In a more developed system that either has a static

antenna gain or a smart antenna capable of doing beam-forming, the antenna gain would

be used here to add to the transmit power as well as in the BER equations to calculated

the EIRP.

34

Computational Complexity

Dependencies

Knobs: modulation type, symbol rate

Meters: none

Objectives: none

Definitions

k = number of bits per symbol

Rs = symbol rate (sps)

The second way to analyze power is to measure it in terms of power consumption

by a radio. Each waveform consumes a certain amount of power relative to the processes

required to transmit and receive information correctly. For example, noncoherent

reception requires less processing power than a coherent receiver, which performs the

frequency and phase correction, and faster symbol rates require faster processing speed,

and therefore more power. The total power consumed includes all aspects of the

transmitter and receiver of a waveform, including the transmitter power.

Interference

Dependencies

Knobs: frequency

Meters: interference map

Objectives: bandwidth

Definitions

fc = center frequency of waveform (Hz)

I(f) = interference power at frequency f from interference map (mW)

B = bandwidth (Hz)

The calculation of interference power is different than SINR from an objective

perspective: SINR helps the cognitive engine decide if it is good for the waveform to

transmit on this frequency. The interference objective looks at the use of the spectrum

from the external perspective to see how much overlap exists between competing signals

for the same spectrum. Focusing on this objective biases the cognitive engine away from

using a waveform that conflicts with another user for the sake of the resources and not the

capabilities of the waveform. The interference power is given as

3.5.2 Multi-objective Analysis

Multiobjective analysis is a method of performing analysis in a large factor which is

results in optimal results.The most straight-forward method of selection is to build a

single utility function that combines the objectives into one number. The algorithm can

then easily rank the solutions and select the solution that maximizes (or minimizes) the

35

utility function. Utility functions are a core research area in economics and operations

research. The most basic utility function is the weighted-sum approach, shown in

equation where U is an overall metric of performance. The weights wi, applied to each

function,

)( xfi

are a weighting of importance, or preference, of the objective.

Another popular method of evaluating performance in a multi-objective problem space is

using population-based analysis and Pareto-ranking. The Pareto-ranking analysis takes a

set, or population, of possible solutions to a multi-objective problem and looks to see

which members are non-dominated; that is, which members of the population outperform

others in all dimensions. In Pareto-ranking, each potential solution is ranked relative to

other solutions. In a search or optimization algorithm, the idea is to push for better and

better solutions until they lie on the optimal Pareto front. This set of solutions represents

a trade-off space among all objectives. The final step of the algorithm is to select the

solution that best represents the desired trade-off, which is done through some subjective

or weighted analysis process to find the proper trade-off space.

3.5.3 Genetic Algorithm for Optimization of Radio Resources

In their most simplistic form, genetic algorithms (GA) are single-objective search and

optimization algorithms. Common to all GAs is the chromosome definition: how the data

are represented; the genetic operations of crossover and mutation; the selection

mechanism for choosing the chromosomes that will survive from generation to

generation; and the evaluation function used to determine the fitness of a chromosome. A

genetic algorithm encodes a set of input parameters that represent possible solutions into

a chromosome. The evaluation stage develops a ranking metric of chromosome fitness

for each individual, which then determines their survival to the next generation.

Optimization progresses through finding genes that provide higher fitness for the

chromosome in which it is found. The fitness calculation is often done through some

absolute metric such as cost, weight, or value by which the algorithm can rank the

success of an individual. Selection is the technique by which more fit individuals are

selected for survival to reproduce for the next generation while less fit chromosomes are

killed off. An algorithm terminates when it reaches a desired level of fitness in the

population, a single member exceeds a desired fitness, the fitness plateaus for a certain

number of generations, or through a simple criteria based on a maximum number of

generations. The algorithm then takes the most fit individual of the last generation as the

solution.

3.6 Artificial Intelligence Techniques in Wireless Communications

(NOV 2016, APRIL 2017)

36

A cognitive radio uses Artificial Intelligence (AI) to adapt and optimize the performance

of a radio platform, specifically an SDR. Aritificial Intelligence techniques for Cognitive

Radios are Artificial Neural Network (ANN), Hidden Markov Model(HMM), Fuzzy

Logic, Evolutionary Algorithms, and Case Based Reasoning

Artificial Neural Network

Neural networks are among the oldest form of AI in computer science. They have been

developed earlier but recent advances, both hardware and software, enable their use in

more applications. Of particular importance to cognitive radios, neural networks provide

a means for signal and modulation detection and classification.

Neural networks are signal processing elements that perform simple operations on data.

However, the collection of artificial neurons and clever learning algorithms allow

networks to build and adapt to represent and process data in interesting ways. In signal

classification, they take multiple noisy input items and provide highly accurate (when

built correctly) answers to the type of modulation represented.

The first artificial neural network is presented by neurophysiologist W.Mcculloch and

logician W.Pits in 1943 for the study of human brain. The idea of ANN is applied to

computational models. Modeled on nerve plexus, ANN is set of non-linear functions with

adjustable parameters to give a desired output. Different types of ANNs are separated by

their network configurations and training methods allowing for a multitude of

applications. All are comprised of neurons interconnected to form a network. Each

artificial neuron usually produces a single output value by accumulating inputs from

other neurons. There are many types of ANN and applicable to CR which are as follows.

Multilayer Perceptron Networks – Comprises of layers of neurons each being a linear

combination of previous layers outputs.

Radial Basis Functions – It has build in distance criterion with respect to a centre a radial

nonlinear function in its hidden layer.

BackPropagation Networks – It uses backpropation algorithm and its derivatives for

operating on set on inputs in the input layer.

37

Fig.3.17 Artificial Neural Network

A neural network shown in Fig. 3.17 is formed when a collection of individual nodes is

organized together in a multilayer fashion, as illustrated in Figure. The neural network

has a set of input points. The values sensed or input through these points is propagated

forward, through the middle layer (also referred to as the hidden layer), to a set of output

points. The output points activated by the forward-propagation are then compared with

the actual value (i.e., anticipated versus actual). If there is a match, the path followed to

arrive at the output point is followed through backpropagation, and the intervening nodes

and paths are reinforced for that particular output point. Learning within a neural network

requires feedback that allows the network to compare the expected output value

associated with a set of input data against the conclusion reached by the neural network.

This forms the back-propagation illustrated in Figure 3.17 . As the vectors of input values

are applied to the system, the data propagates through the intermediate layers to the

output. The expected output is mapped onto the output vectors. Those elements in an

output vector whose value matches the applied expected value are reinforced by

backpropagation. Thus, the intermediate layers that contributed to the propagation of data

resulting in the correct (i.e., expected) output values are reinforced. Reinforcement may

be performed through increasing weights, of the nodes involved in the propagation from

the input data to the correct output or conclusion.

This reinforcement of neural links increases the probability that the same input values

will result in the same propagation to the correct output values. Those output values that

did not match the expected output value are weakened, thereby decreasing the probability

that they would be applied again given the same set of conditions.

38

Hidden Markov Models (HMM)

Hidden Markov Model was first introduced in 1960s. It is a convenient and

mathematically tractable statistical model to describe and analyze the dynamic behavior

of a complex random phenomenon that can be modeled as a Markov process with

observable and unobservable states. HMM generates sequences of observation symbols

by making transitions from state to state one symbol per transition. However, the states

are hidden, and only the output is observable. HMM can be built for a specific system to

explain and characterize the occurrence of the observed symbols or patterns.

In some situations Hidden Markov Models (HMM) are considered as artificial

intelligence. A HMM is a processing tool that uses past data to help predict future

actions. HMM is useful in communications and cognitive radios. The best model for

studying HMMs work is obtained from Rabiner's tutorial. Channel modeling has

extensively used Markov models in research. Probably the most famous is the two-state

Gilbert-Elliot model that describes a channel as in either a good state or bad. When in one

state, there is a probability of either staying in that state or moving to the other state. The

channel properties determine the type of transition probabilities.

Researchers have developed other, more extensive models and the idea of developing

such a model lends itself to cognitive radios. Rieser and Thomas looked into using

HMM's in channel models using genetic algorithms as the training method over the

Baum-Welch algorithm in order to develop compact channel models based on

information gathered in a live system to represent the current channel statistics. The idea

was to use the HMMs as a sensor to understand the channel behavior in a cognitive

engine, though the research was not taken much farther in this direction. Mohammad's

work used HMMs for a similar purpose, but was able to develop classification schemes in

order to use the models for decision making in a cellular network. The ability of

developing is to calculate a similarity distance between HMMs provides promise for

future implementation in a cognitive radio system, especially in the context of the

environmental modeling used in a case-based system.

Fuzzy Logic

Fuzzy logic is a famous technique that started during the early development of artificial

intelligence. Because it deals extensively with uncertainty in decision making and

analysis, it has great potential for application to cognitive radio. However, only a little

work has so far been published in the field, notably by Baldo and Zorzi. Their

implementation suggests some interesting applications, and the discussion points out

larger uses than the specific application of adapting the TCP layer. A problematic aspect

of this work is the amount of domain-specific rules required. All implementations of AI

require domain information, but fuzzy logic must establish a rule related to the specific

situation in which it is used and recalls some of the limitations of expert systems, though

still far more flexible and powerful. Fuzzy logic has potential in either specific problem

solving areas or as a subset or part of a cognitive radio.

39

Evolutionary Algorithms

Evolutionary algorithms uses Genetic Algorithms (GA) where Christian Rieser and

Thomas pioneered the use of genetic algorithms early in cognitive radio. The basic

principles are that the large search space involved in optimizing a radio are more complex

than many search and optimization algorithms can handle. Among those algorithms that

are suited to the task, evolutionary, specifically genetic, algorithms offer a significant

amount of power and flexibility. Cognitive radios are likely to face dynamic

environments and situations as well as radio upgrades due to advancing technology, so

genetic algorithms are particularly applicable.

Newman has also contributed significantly to the use of genetic algorithms for cognitive

radios. The main issues involved in successful genetic algorithm behavior is the selection

of the fitness, or objective, function(s). Newman's work has developed a single, linear

objective function to combine the objectives of BER minimization, power minimization,

and throughput maximization. Mahonen work uses genetic algorithms for cognitive radio

. The topic of their research discusses the use of a cognitive resource manager (CRM) to

select an algorithm from a toolbox of algorithms to solve a particular problem. The wor

specifically points out the use of genetic algorithms for multidimensional problem

analysis, which is multi-objective optimization analysis.

Case-Based Reasoning

The final traditional AI technique is Case-Based Reasoning (CBR). CBR systems use

past knowledge to learn and improve future actions. In these systems, a case-base stores

actions and receives inputs from a sensor. Those inputs help find the action in the case-

base that best fits the information received by the sensor. As mentioned previously, an

optimization routine could, instead of designing a new waveform, select a waveform

from a predefined list. CBR is a method used to make the associations. Although this

may sound like an expert system, CBR systems generally provide learning and feedback

to continuously and autonomously improve their performance. As information is received

and actions taken, the results can help the system improve its response the next time.

Another method is that it develops a similar idea in the experiments they run using

previous knowledge to seed the next run of the genetic algorithm. The cognitive radio

remembers solutions found for one particular problem to apply to the next problem to

initialize the population with known successful chromosomes. The population seeding

resembles the case-based decision theory work. Their seeding concept uses a factor to

calculate the expected change in the environment between runs of the genetic algorithm

to provide context for how successful a new chromosome might be with respect to the

new environment.