ABSTRACT
The two major users of a Cognitive Radio (CR) system are the Primary User (PU) and the Secondary User (SU). A secondary spectrum user cannot transmit in a channel before sensing and knowing the spectrum occupancy state as this may cause interference. This poses a major challenge because these operations ought to be performed in each time slot and thereby causing a substantial delay before the user gains access to the spectrum, leading to inefficient utilisation. Therefore, a channel predictive system will mitigate this problem. In this work, a machine learning model for spectrum occupancy prediction was developed. Power Spectrum Density (PSD) data were collected for 24 hours in Minna, Niger State and FCT Abuja both in Nigeria with 3  measurement  sites  per  location  within the  VHF  band  (30-300  MHz). Exploratory Data Analysis (EDA) using power density plots was used to reduce the dimensionality of the dataset so that the data can be fit for machine learning. The power density plots reveal 12 distinct groupings or frequency sub-bands for the entire dataset. A Back-Propagation Neural Network (BPNN) model was developed to predict the spectrum occupancy using time-series data which was converted into a feature vector that was captured as time instances of the occupancy of all the frequency sub- bands. This serves as the input vector into the feedforward neural network. Twenty- four different input parameters, which capture hourly spectrum occupancy, were used with only one output  that  predicts  the  spectral  occupancy.  Comparison of the prediction results with the actual results obtained was done. The weight of the neural network initially generated randomly was improved using the Auto-Regressive (AR) model whose order is based on the time dimension of the feature vector. The coefficients of the AR model were obtained from the synaptic weights and adaptive coefficients of the nonlinear sigmoid activation function of one hidden layer with a ten-neuron Real-Valued Neural Network (RVNN). A linear activation function was used in the output layer. To obtain the AR coefficients, the training data and the corresponding expected occupancy (estimated from the raw data) are passed to the neural network alongside the number of neurons in the hidden layer. Based on the training data and the corresponding output data, the neural network model trains itself to come up with the best weights which can generally be used by the AR model for unseen data. After computing the weights, the performance is first tested on the entire training data, on the validation dataset, and the test dataset. Overall, the results for Minna dataset, in band 1 (30-47 MHz), reveal a highest actual spectral occupancy of 40.59% with a prediction accuracy of 99.06% while band 7 (137.05-144 MHz) has a lowest occupancy of 25.24% and a prediction accuracy of 99.31%. The corresponding results for the Abuja dataset, in band 1 (30-47 MHz), show a highest actual spectral occupancy of 39.11% with a prediction accuracy of 98.59% while band 11 (230.05- 267 MHz) has a lowest occupancy of 22.13% in with a prediction accuracy of 99.40% were obtained for Abuja dataset. Clearly, band 1 had the highest spectral occupancy values in both locations and therefore should be avoided for Cognitive Radio (CR) deployment. The performance of the Neural Network prediction model reveals accuracy of 91.51% to an unseen test dataset, an accuracy of 99.02% on the training dataset and an accuracy 91.63% to the validation dataset.
CHAPTER ONE
1.0 INTRODUCTION
1.1 Background to the Study
The demand for wireless equipment is high. Similarly, the usage of data and multimedia applications has resulted in a massive increase in demand for faster data rates. This has resulted in a greater need for radio spectrum (Oluwafemi et al., 2021). The radio spectrum is the wheel on which wireless communication systems are driven. The high demand for radio spectrum can be observed in the auctions that have been conducted in nations where large sums have been paid for spectrum licenses (Kyeremateng-Boateng et al., 2020; Anabi et al., 2016; Mishra et al., 2012; Doerr et al., 2008; Dame, 2001). Due to the restricted natural frequency usage, there is a spectrum deficit. Because the entire radio spectrum has been assigned to various wireless services, some newly emerging wireless technologies are unable to operate (Gupta and Jha, 2015). The spectrum’s opportunity cost rises as a result. Because of inefficient static spectrum distribution policies, spectrum scarcity is not real but artificial (Peha, 2009).
Early spectrum survey in the United States of America (USA) found a shocking result of spectrum underutilisation, corroborating the assertion of inefficient spectrum management (McHenry, 2005). According to the study, a considerable amount of the available spectrum is unused. Due to time, frequency, and geographic location, 15 percent to 85 percent of the allocated spectrum remains inactive (Seflek and Yaldiz 2019; Akyildiz et al., 2006). These studies reveal that the existing spectrum allocation mechanism is faulty and incapable of meeting the increasing demand for spectrum for future cellular services. The present system of spectrum management is very rigid, with major licensed users having entire or absolute control over the radio frequency spectrum (Matheson and Morris, 2012).
1.1.1 The Radio Spectrum
Spectrum  refers to the frequency  range  of Electromagnetic (EM)  waves  or radio frequencies utilised for communication. As shown in Equation 1.1, the frequency f of EM waves, measured in Hertz (Hz) or cycles per second, is inversely proportional to the wavelength.
where is the speed of light in vacuum and λ is wavelength in meters EM waves have the capability of transporting energy through space at the speed of light. Frequency, wavelength and amplitude are the parameters for defining the EM waves. Wavelengths are shorter for high-frequency waves and longer for low- frequency waves. The wavelength also reflects the wave’s ability to move through space.  Waves  with  a lower  frequency  travel  further  than  waves  with  a higher frequency. It is worth noting that radio waves are usually characterised in terms of frequency than wavelength. The radio frequency spectrum spans frequencies between 30 Hz and 300 GHz. It covers a range of radio waves, which are a subset of EM waves. Transmitters generate these waves which are received by the antennas or aerials.
Spectrum  is  a  valuable  finite  resource  that  is becoming  insufficient  day  by  day because of the growth in deployment of newly emerging wireless systems. To support this growth, technology developers and regulators  are examining and considering better techniques to increase spectrum efficiency to avoid spectrum crisis. Consequently, several investigations have been conducted around the world to assess and estimate the effectiveness of the fixed spectrum allocation technique currently in use. This involves conducting spectrum occupancy measurements in various locations with the aim of evaluating and selecting suitable spectrum bands for the application of Cognitive Radio (Martian et al., 2013). Results obtained from such measurements reveal very low utilisation of the frequency spectrum. However, in Nigeria, the availability of such spectrum utilisation information remains limited (Ajiboye et al., 2019) and this serves as the motivation for this research.
The radio spectrum is used by broadband services, radio and television transmission, mobile phones, radar, two-way radios, fixed lines, and satellite communications. As shown in Table 1.1, radio spectrum is divided into frequency bands based on frequency range.
Table 1.1: Radio Frequency Bands
Diverse telecommunication technologies might theoretically be found in any part of the radio spectrum. The size of the bandwidth required is determined by the amount of data carried by a signal. Simply said, bandwidth refers to the frequency range that a signal occupies in the spectrum. In the TV Band, the allocated bandwidth for the television  transmission  channel  is  8  MHz.  The  bandwidth  capacity  of lower frequencies is less than that of higher frequencies. This means that communications with a lot of information, like mobile phones, television, and internet, need higher frequency bands, whereas basic audio radio signals can work with low-frequency waves. Radio signals have propagation features that are associated with the frequency during transmission.
1.1.2 Cognitive Radio Network
Cognitive Radio (CR) is a smart radio that is aware of its surroundings and can be used to implement Dynamic Spectrum Access (DSA) (Piran et al., 2020; Ajiboye et al., 2013). Figure 1.1 shows the network architecture of CR, consisting of two primary networks and a secondary CR network. A legitimate user that is licensed to operate in the primary network (or licensed network) within a specific band is referred to as a Primary User (PU). All PU activities are regulated through a base station if primary networks have  infrastructure. In  terms of functioning, PUs have precedence  over unlicensed users. To be able to operate in the licensed spectrum band, it will need additional  capabilities.  The activities  of  secondary  users  are  controlled  by  CR Networks (CRN), which have their own infrastructure. Finally, spectrum brokers may be used by CR networks to allocate spectrum resources among different CR networks (Ileri et al., 2005).
The heterogeneous wireless architecture enables CR Networks to provide unlicensed users large bandwidth through the technique of dynamic spectrum sharing. The CR users sense the channels and decide on the best channel that is free for secondary communication. While transmitting as Secondary User (SU), if a PU is detected, then the SU must vacate the channel for the PU since the PU is a priority user. Spectrum sensing, spectrum decision, spectrum sharing and spectrum mobility are all types of spectrum management activities and can all be used to create spectrum dynamics (Gupta and Kumar, 2019; Akyildiz et al., 2008). CRN’s spectrum management provides various services to solve difficulties such as avoiding harmful interference to authorised users, quality of service (QoS) provisioning, and smooth secondary communication (Fakhrudeen and Alani, 2017).
1.1.3 Spectrum Sensing
Transmitter detection, cooperative detection, and interference detection are the three types of spectrum sensing techniques (Nasser et al., 2021). The transmitter detection technique is a frequently employed method used in the detection of PU. In order to detect primary transmitter signals, three procedures or strategies are used, which are are Energy detection, feature detection and match filter detection (Sivagurunathan et al., 2021; Cabric et al., 2004). An SU ensures that the channel is free by monitoring the bands that are not being utilised and thereby detecting spectrum holes for opportunistic use. In doing so the computational complexity of the network increases and this has an impact on the performance (Vartiainen et al., 2016). To avoid severe interference to primary users, effective spectrum hole detection is a critical step in CR spectrum management.
1.1.4 Spectrum Decision
Spectrum decision is a critical feature of the Cognitive Radio Network (CRN). It is the ability to choose the best channel for the application out of all available channels in order to meet the QoS requirement (Joshi et al., 2013). The choice of the spectrum is associated with channel features and the operation of the primary users. Policies made within and outside the country about channel allocation are very definite hence, spectrum decision is key. Two steps are involved in spectrum decision:
1. Each accessible spectrum band is characterised.
2. A suitable channel is selected based on characterisation.
The channel is characterised using statistical data from the primary network and the activity of CR users.
1.1.5 Spectrum Sharing
In the CRN, spectrum sharing comprises the coordination of transmission from CR users (Zareei et al., 2017). Spectrum sharing can be achieved when functionality in Medium Access Control (MAC) protocol is engaged. Spectrum sharing in CRN is complicated by the coexistence of CR users and licensed users, as well as the large range of spectrum available for operation. The architecture, spectrum allocation behaviour and spectrum access technique are used to classify spectrum sharing and scope, as well as the related problems.
1.1.6 Spectrum Mobility
A CR user scans for the most vacant channel for signal transmission and immediately a PU appears, the CR user vacates to another channel ensuring noninterference (Oyewobi and Hancke, 2017; Ajiboye and Adediran, 2012). This phenomenon is called spectrum hand-off or spectrum mobility. To allow for faster switching without reducing connection performance, this frequency-agile operation necessitates a change to a network-protected parameter. The handoff time is an important factor in spectrum mobility management. The available channels for spectrum mobility change over time, making it difficult to maintain QoS. Furthermore, the migration of CR users causes a difficulty in which the same channel is continuously assigned to a new location.
1.2 Statement of the Research Problem
The majority of spectrum measurement campaigns done across the globe show that there is gross underutilisation of the radio spectrum temporally and spatially. However, since geographical locations are unique and have varied spectral activities, measurement results for a location cannot be directly applied to another geographical location. This is due to the impact of geographical and social characteristics on spectrum usage.
An opportunistic secondary usage of the spectrum requires repeated scanning of the bands and determining their occupancy in spectrum sensing. A secondary user therefore cannot transmit in a channel before sensing and knowing the occupancy state, as this may cause interference. This poses a major challenge because these operations ought to be performed in each time slot and thereby causing substantial delays before the user gains access to the spectrum, leading to inefficient utilisation. Therefore, a system that can intelligently forecast the state of the channel for future time slots can minimise the delay and also the energy consumed during spectrum sensing and the decision-making phase. The development of a prediction model for spectrum occupancy is what this study seeks to address.
1.3 Aim and Objectives of the Research
The research aims to develop a spectrum occupancy prediction model for Cognitive Radio (CR) systems. Towards achieving this aim, the following objectives are highlighted:
1. To Measure and Analyse spectrum data of TV bands with a view of quantifying the occupancy ratio.
2.  To  develop  a  Neural  Network-based  Model  for  Spectrum  Occupancy Prediction.
3.  To  analyse  and  validate  the  performance  of  the  developed  Spectrum Occupancy Model.
1.4 Scope of the Study
This research focuses on TV bands. There are numerous reasons for focusing on the TV band for secondary usage. VHF sub-band (30-300 MHz) exists in some portions of the TV. The band has great potential in rural areas that are difficult to be reached due to remoteness. Optical fiber techniques and other types of technologies are not economically viable in such rural settings, therefore the TV band and its associated technology are most suitable (Kimani and Langat, 2017). Also, the VHF band has its relevance in developing countries where the telecommunications network is not yet deployed. This can serve as an economical way of making up for the unavailability of telecommunications infrastructure (Kimani and Langat, 2017).
Furthermore, the VHF band has a good propagation characteristics. According to Liang et al. (2008), the IEEE 802.22 standard is very popular for remote communication and employs CR technology. The standard ensures opportunistic use of  the VHF  in  TV  spectrum  bands.  The  Institute of  Electrical  and  Electronics Engineers (IEEE) 802.22 standard does not need a dedicated spectrum and this has led to a reduction in the expenses for its deployment. The advantage of network coverage makes IEEE 802.22 technology appropriate for use in rural and remote places. Finally, due to their relatively static network design, open standards, and public availability of information on TV transmitters, TV broadcast systems are extremely simple. A major advantage of Cognitive Radio is that it has potential of learning from past experiences. Therefore, models can be developed for the prediction of spectrum usage and the channel state accurately. Prediction-based spectrum sensing is a better alternative of spectrum sensing. The ability to predict spectrum holes will translate to a high- performance accuracy.
1.5 Significance of the Study
Primary User (PU) and Secondary User (SU) spectrum hole prediction should be done seamlessly and without creating interference by SU. Sensing the whole wideband spectrum each time a transmission is to be initiated is time-consuming and will ultimately lead to the draining of SU’s power. Through a prediction mechanism, access to the spectrum that will help in restricting the sensing of channels to portions of the spectrum that are less busy through learning is desirable and this will greatly improve the ease of use and administration of the spectrum.
1.6 Thesis Layout
The introductory chapter, Chapter One, contains the following sections: Background to the Study, Statement of Problem, Aim, and Objectives, Scope, and Significance of the Study. Chapter Two contains Literature Review while Chapter Three highlights the proposed Research Methodology adopted. In Chapter Four, Results are presented and discussed while in Chapter Five Conclusions and Recommendations are drawn.
This material content is developed to serve as a GUIDE for students to conduct academic research
DEVELOPMENT OF A SPECTRUM OCCUPANCY PREDICTION MODEL FOR COGNITIVE RADIO SYSTEMS
A1Project Hub Support Team Are Always (24/7) Online To Help You With Your Project
Chat Us on WhatsApp » 09063590000
DO YOU NEED CLARIFICATION? CALL OUR HELP DESK:
09063590000 (Country Code: +234)
YOU CAN REACH OUR SUPPORT TEAM VIA MAIL: [email protected]
09063590000 (Country Code: +234)
