Project:- Spectral occupancy measurement and analysis for Cognitive Radio application
1. Report
ON
“SPECTRAL OCCUPANCY MEASUREMENT AND ANALYSIS FOR
COGNITIVE RADIO APPLICATION ”
Submitted to
The LNMIIT
BY
Aastha Bhardwaj
Under the guidance of
Prof. Ranjan Gangopadhyay
Department of Electronics & Communication Engineering
The LNMIIT
Jaipur
2. ACKNOWLEDGEMENT
I present my heart filled gratitude to Prof. R Gangopadhyay and Dr. S
Debnath, Head of Department of Electronics & Communication
Engineering, LNMIIT for giving me an opportunity to work under
their guidance on this project of cognitive radio and encouraging me to
give my best to the project.
I would like to thank respected Phd. Scholars Anirudh Agarwal and
Aditya Singh Sengar from the bottom of my heart for their constant
guidance and support throughout the project, solving my doubts
patiently, introducing me with the equipments used and giving ideas to
implement things in a much simpler way.
Aastha Bhardwaj
3. CONTENTS
Abstract
Introduction
Measurement Methodology
Time Series
Different time series models
Working with time series modeling
Working with MATLAB
Conclusion
References
4. Abstract
Cognitive radio (CR) technology is envisaged to solve the problems in
wireless networks resulting from the limited available spectrum and the
inefficiency in the spectrum usage by exploiting the existing wireless
spectrum opportunistically. It involves spectrum sharing which consists
of spectrum sensing, spectrum decision , spectrum mobility. The 24-
hour spectrum usage pattern is studied. The objectives are to find how
the scarce radio spectrum allocated to different services is utilized and
identify the bands that could be accessed for future opportunistic use due
to their low or no active utilization, followed by prediction. While
dedicated spectrum occupancy monitoring provides vital information for
frequency planning and management, it usually cannot tell the common
properties in spectrum occupancy. As a complement, the models
approach can be used to describe and compare the occupancy situations
under similar conditions. Time series analysis has been applied for
modeling the radio spectrum occupancy.
5. INTRODUCTION
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
defined 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. It works on the principle of artificial intelligence.
The accuracy of the decisions made by artificial intelligence method are
based on the quality and quantity of the inputs to the system. These
inputs comprise of transmission parameters and environmental
parameters which are collectively known as operating parameters.
Wireless spectrum is carved up into chunks called frequency bands.
These are licensed bands, meaning that individual companies pay a
licensing fee for the exclusive right to transmit on assigned channels
within that band in a given geographic area. These licensed bands gave
rise to the concept of primary and secondary users. A user who has
higher priority or legacy rights on the usage of a specific part of the
spectrum is called primary user. A user who has a lower priority and
therefore exploits the spectrum in such a way that it does not cause
interference to primary users is known as secondary user. A channel can
6. be considered as an opportunity if it is not currently used by primary
users.
The cognitive radio networks will have to respect the policies, defined
by regulatory bodies which are based on central idea that cognitive radio
can access and share the spectrum in an opportunistic manner with
licensed users, provided that there should have no or very limited impact
on licensed user communication . Such a solution can be complicated
and impose unique challenges due to their coexistence with primary
networks, typical dynamic behavior of primary user, interference
avoidance and QoS awareness. In order to meet these challenges,
cognitive radio operates on cognitive cycle which comprises of 4 main
steps. These steps are: spectrum sensing, spectrum decision, spectrum
sharing and spectrum mobility.
Spectrum sensing is the fundamental requirement of cognitive system to
work. A cognitive user should monitor the spectrum bands to determine
the presence or absence of primary user before transmission. Spectrum
sensing is done in order to minimize the impact of secondary users on
primary users. Basically spectrum sensing techniques are classified into
three main groups: Primary transmitter detection which includes
matched filter detection, energy detection and feature detection. The
other two groups include primary receiver detection and interference
temperature management. Spectrum Sensing Techniques are classified
according to
1. Architecture :- Centralized ,Distributed
2. Spectrum allocation behavior :- Cooperative , Non Cooperative
3. Spectrum Access Technology :- Overlay Spectrum Sharing : In this,
secondary users aim to exploit temporal spectrum opportunities resulting
from the burstly traffic of primary users. A typical application is the
7. reuse of certain TV bands that are not used for TV broadcast in a
particular region.
Underlay Spectrum Sharing: In this radios coexist in the same band
with primary licenses, but are regulated to cause interference below
prescribed limits.
Spectrum decision: Based on information of spectrum sensing, a
spectrum band is analyzed and best available spectrum is selected for
transmission. This allocation is focused mainly on spectrum availability,
cost of communication and quality of service requirements.
Spectrum sharing: Cognitive radio has to access and share the spectrum
with multiple other secondary or cognitive users. Spectrum sharing is to
distribute the spectrum among all cognitive and non-cognitive users such
that there should be no collisions among them. A spectrum overlay
technique is a spectrum management principle whereby a secondary user
uses a channel from a primary user only when it is not occupied.
Spectrum underlay technique is a spectrum management principle by
which signals with a very low spectral power density can coexist as a
secondary user with the primary user of the frequency band.
Spectrum Mobility: The fourth step in spectrum management and one of
the most prominent features of cognitive radio networks will be the
ability to switch to different portions of radio spectrum as soon as
spectrum left over or spectrum holes are detected. Spectrum mobility is
the technique that will enable cognitive radio networks to achieve this
goal. As licensed users or primary users have the right to their spectrum
slice thus cannot accept any interference thus in this direction the most
important and challenging issue of spectrum mobility is to avoid
interference to primary users and attain a seamless communication.
8. Spectrum Mobility or handoff process is carried out when channel
occupied by secondary users is interrupted or reclaimed by the
occurrence of primary users. As soon as the primary user appears,
secondary user has to vacate the frequency channel to avoid interference
to primary user and switch to other available free channel to resume and
finish its ongoing transmission.
Cognitive radio also involves the concept of white space and gray space.
White Space refers to the unused broadcasting frequencies in the
wireless spectrum whereas gray space sharing is one in which devices
are given access to spectrum that is already in use.
9. Measurement Methodology
For doing prediction and performing cognitive cycle operations practical
data sets are needed for which discone antenna and rf explorer.
A discone antenna is a version of a biconical antenna in which one of the
cones is replaced by a disc. It is usually mounted vertically, with the disc
at the top and the cone beneath. Omnidirectional, vertically polarized
and with gain similar to a dipole, it is exceptionally wideband, offering a
frequency range ratio of up to approximately 10:1. The radiation pattern
in the horizontal plane is quite narrow, making its sensitivity highest in
the direction of the horizon and rather less for signals coming from
relatively close by. The discone’s wideband coverage makes it attractive
in commercial, military, amateur radio and radio scanner applications.
Using discone antenna data sets for different places are collected. The
figure below shows a discone antenna.
10. A spectrum analyzer is used in analyzing the collected data set. It
measures the magnitude of an input signal versus frequency within the
full frequency range of the instrument. The primary use is to measure the
power of the spectrum of known and unknown signals. One such
spectrum analyzer RF Explorer is a remarkable diagnostic tool used for
monitoring and troubleshooting wireless systems and communications.
The MATLAB software is used to perform different operations on
collected data set. A certain threshold is fixed than power spectral
densities above this threshold is assigned 1 and below the threshold is
assigned 0. This gives the information about how much spectrum is
occupied. The occupancy is quantified as the amount of spectrum
detected above a certain received power threshold.
11. Power Spectral Density(PSD): It is a measure of a signal’s power
intensity in the frequency domain. The PSD provides a useful way to
characterize the amplitude versus frequency content of a random signal.
With the help of data sets collected 3 graphs were plotted
1. PSD vs Frequency 2. Waterfall(PSD vs Frequency vs Time) 3. Duty
Cycle vs Frequency
Duty Cycle: It is the proportion of time during which a component,
device or system is operated.
12. Time series
Need for time series modeling
Spectrum monitoring is one of four key spectrum management
functions which include spectrum planning , spectrum engineering and
spectrum authorization. It helps spectrum managers to plan and use
frequencies , avoid incompatible usage, and identify sources of harmful
interference. However detailed measurements and analyses intended to
quantify the performance of a particular band and a particular measuring
period usually cannot be extended directly to others and usually cannot
tell the common properties in spectrum occupancy. As a complement,
the models approach can be used to describe and compare the occupancy
situations under similar conditions.
What is time series
A time series is a sequence of data points made
over a continuous time interval
out of successive measurements across that interval
using equal spacing between every two consecutive measurements
with each time unit within the time interval having at most one
data point.
The purpose of time series analysis is to draw inferences from series, so,
one can infer the general occupancy situation without monitoring the
spectrum.
13. Different time series models
Autoregressive Model(AR)
In statistics and signal processing, an autoregressive (AR) model is a
representation of a type of random process . The autoregressive model
specifies that the output variable depends linearly on its own previous
values and on a stochastic term (an imperfectly predictable term); thus
the model is in the form of a stochastic difference equation. the AR
model is not always stationary as it may contain a unit root.
In mathematics and statistics, a stationary process (or strict(ly) stationary
process or strong(ly) stationary process) is a stochastic process whose
joint probability distribution does not change when shifted in time.
Consequently, parameters such as the mean and variance, if they are
present, also do not change over time and do not follow any trends.
Stationarity is used as a tool in time series analysis, where the raw data
is often transformed to become stationary.
The notation AR(p) indicates an autoregressive model of order p.
Moving-average Model(MA)
In time series analysis, the moving-average (MA) model is a common
approach for modeling univariate time series. The moving-average
model specifies that the output variable depends linearly on its own
previous stochastic term and on a stochastic term (an imperfectly
predictable term). Contrary to the AR model, the MA model is always
stationary.
The notation MA(q) refers to the moving average model of order q.
14. Autoregressive–moving-average model
In the statistical analysis of time series, autoregressive–moving-average
(ARMA) models provide a parsimonious description of a (weakly)
stationary stochastic process in terms of two polynomials, one for the
auto-regression and the second for the moving average. Given a time
series of data (Xt), the ARMA model is a tool for understanding and,
perhaps, predicting future values in this series. The model consists of
two parts, an autoregressive (AR) part and a moving average (MA) part.
The model is usually then referred to as the ARMA(p, q) model where p
is the order of the autoregressive part and q is the order of the moving
average part.
A stationary process 𝑋𝑡 is defined to be an ARMA(p, q) if for every t,
∅( 𝐵) 𝑋𝑡 = 𝜃( 𝐵) 𝑍𝑡
where 𝑍𝑡 ~N(0,𝜎2
) normal distribution with zero mean and variance 𝜎2
,
B is the backward shift operator
𝐵 𝑗
𝑋𝑡 = 𝑋𝑡−𝑗
and φ(B), θ(B) are the pth degree autoregressive (AR) and qth degree
moving average (MA) polynomials respectively.
∅( 𝐵) = 1 − ∅1 𝐵 −....−∅ 𝑝 𝐵 𝑝
𝜃( 𝐵) = 1 + 𝜃1 𝐵 +....+𝜃 𝑞 𝐵 𝑞
.
15. Autoregressive integrated moving average
In time series analysis, an autoregressive integrated moving average
(ARIMA) model is a generalization of an autoregressive moving average
(ARMA) model. These models are fitted to time series data either to
better understand the data or to predict future points in the series
(forecasting). They are applied in some cases where data show evidence
of non-stationarity, where an initial differencing step (corresponding to
the "integrated" part of the model) can be applied to reduce the non-
stationarity.
Non-seasonal ARIMA models are generally denoted ARIMA(p,d,q)
where parameters p, d, and q are non-negative integers, p is the order of
the autoregressive model, d is the degree of differencing, and q is the
order of the moving-average model. Seasonal ARIMA models are
usually denoted ARIMA(p,d,q)(P,D,Q)m, where m refers to the number
of periods in each season, and the uppercase P,D,Q refer to the
autoregressive, differencing, and moving average terms for the seasonal
part of the ARIMA model. Seasonal ARIMA models allow for
randomness in the seasonal pattern from one cycle to the next. However
it can be quite complicated.
Working with time series modeling
Time series modeling involves the use of autocorrelation and partial
autocorrelation function. It starts by first finding whether the model is
stationary or not. If the tendency for the autocorrelation function did not
die out quickly as shown in the figure below this might suggest non-
16. stationarity. So the difference operation should be applied to the series.
After a time series has been stationarized by differencing, the next step
in fitting the model is to determine whether AR or MA terms. If the
autocorrelation function displays a sharp cutoff while the partial
autocorrelation function decays more slowly like the one shown in the
figure given below. It is said that the stationarized occupancy series
displays a MA signature, meaning that the autocorrelation pattern can be
explained more easily by adding MA terms than by adding AR terms
17.
18. If the partial autocorrelation function displays a sharp cutoff while the
autocorrelation function decays more slowly like the one shown in the
figure given below. It is said that the stationarized occupancy series
displays a AR signature, meaning that the autocorrelation pattern can be
explained more easily by adding AR terms than by adding MA terms.
19.
20. Typically, the goodness of fit of a statistical model to a set of data is
judged by comparing the observed values with the corresponding
predicted values obtained from the fitted model.
Working with MATLAB
The data set of different places which is obtained with the help of
discone antenna is now operated on in MATLAB software. First all the
values of different days are concatenated and then a particular threshold
value is set, occupancy data is obtained. For applying time series
modeling we need model order(values of p, q, d) which is calculated for
all the 112 channels by using autocorrelation function and partial
21. autocorrelation function. After all this, prediction work starts which
involves the use of econometrics toolbox and financial toolbox of
MATLAB. Prediction is done for different percentage of data sets. 70%
data is used for training and 30% for testing . In another case 70% data
is used for training and the same is used for testing. This was also done
for 50%.
22. CONCLUSION
For prediction and analysis several operations were performed on the
collected data set. There were 112 channels in every data set and on
calculating autocorrelation function and partial autocorrelation function
different model orders were noticed for different channels. The data sets
were grouped for different time intervals i.e values in ten minutes of
interval were grouped together and occupancy data set was formed.
Prediction was easily performed in the case of data set where values in
one minute of time interval were grouped and training-testing was done
for different data as predicted vector was obtained by running the code
once for all 112 channels. On the other hand in the case of data set
where values in ten or fifteen minute of time interval were grouped and
training-testing was done on same values then the code was run
individually for all the channels. However the first approach was more
time consuming. Cognitive radio (CR) is a promising technology that
can alleviate the spectrum shortage problem by enabling unlicensed
users equipped with CRs to coexist with incumbent users in licensed
spectrum bands while causing no interference to incumbent
communications.
23. REFERENCES
Hybridization of intelligent techniques and ARIMA models for
time series prediction O.Valenzuelaa, I. Rojasb,∗, F. Rojasb, H.
Pomaresb, L.J. Herrerab,A. Guillenb, L. Marqueza, M. Pasadasa
Spectrum Occupancy Statistics and Time Series Models for
Cognitive Radio Zhe Wang & Sana Salous
Spectrum Occupancy Measurements and Analysis in Beijing
Jiantao Xue*,Zhiyong Feng,Ping Zhang
Spectrum Occupancy Survey In HULL-UK For Cognitive Radio
Applications: Measurement & Analysis Meftah Mehdawi, N.
Riley, K. Paulson, A. Fanan, M. Ammar
A Survey of Cognitive Radio Access to TV White Spaces Maziar
Nekovee1,2 1BT Innovate and Design, Polaris 134, Adastral Park,
Martlesham, Suffolk IP5 3RE, UK 2Centre for Computational
Science, University College London, 20 Gordon Street, London
WC1H 0AJ, UK
Cognitive Radio Networks Mobile Communication Networks
(RCSE)
ENERGY DETECTION TECHNIQUE FOR SPECTRUM
SENSING IN COGNITIVE RADIO: A SURVEY, Mahmood A.
24. Abdulsattar and Zahir A. Hussein ,Department of Electrical
Engineering, University of Baghdad, Baghdad, Iraq
A Survey on MAC Protocols for Cognitive Radio Networks,
Claudia Cormio, Kaushik R. Chowdhury
A Review on Spectrum Mobility for Cognitive Radio Networks,
Anuj Thakur1, Ratish Kumar2
Spectrum-Aware Mobility Management in Cognitive Radio
Cellular Networks, Won-Yeol Lee ; Georgia Institute of
Technology, Atlanta ; Ian F. Akyildiz
http://www.radio-electronics.com/info/rf-technology-
design/ofdm/ofdm-basics-tutorial.php
http://www.telecomabc.com/s/spectrum-underlay.html
A Comparison Between the Centralized and Distributed
Approaches for Spectrum Management, Gbenga Salami ; Centre
for Communication Systems Research, University of Surrey, GU2
7XH, Guildford, United Kingdom ; Olasunkanmi Durowoju ;
Alireza Attar ; Oliver Holland
Cognitive Radio Networks ,X. Hong ; King's College, London,
United Kingdom ; Z. Chen ; C-X. Wang ; S. A. Vorobyov more
authors
Compressed Sensing for Wideband Cognitive Radios, Zhi Tian ;
Dept. of Electrical & Computer Engineering, Michigan
Technological University, Houghton, MI 49931 USA ; Georgios
B. Giannakis