Open Access is an initiative that aims to make scientific research freely available to all. To date our community has made over 100 million downloads. It’s based on principles of collaboration, unobstructed discovery, and, most importantly, scientific progression. As PhD students, we found it difficult to access the research we needed, so we decided to create a new Open Access publisher that levels the playing field for scientists across the world. How? By making research easy to access, and puts the academic needs of the researchers before the business interests of publishers.
We are a community of more than 103,000 authors and editors from 3,291 institutions spanning 160 countries, including Nobel Prize winners and some of the world’s most-cited researchers. Publishing on IntechOpen allows authors to earn citations and find new collaborators, meaning more people see your work not only from your own field of study, but from other related fields too.
The fast expansion of new wireless devices and applications over the last ten years has resulted in a massive increase in the demand for wireless radio spectrums. The existing fixed spectrum assignment strategy hinders efficient spectrum use. Much of the licensed spectrum has been underutilized due to this approach. To solve the spectrum efficiency problem, the idea of Cognitive Radio Networks (CRNs) has gained traction in recent years. Thanks to CRNs, Wi-Fi users can adjust their operational parameters best to suit their interactions with the surrounding radio environment. This paradigm aims to improve spectrum utilization through the dynamic allocation and sharing of spectrum resources. One of the most critical issues affecting the design and performance of traditional networks, such as Wireless Sensor Networks (WSNs), is localization. Location information is crucial for network design as primary users (PUs), key components of CRNs, have priority in using licensed spectrum. The challenge lies in accurately determining the location of sensor nodes within the network and determining the exact location where an event occurs. Precise information about the location of PUs is of great importance in CRNs. In this chapter, the issue of localization in CRNs is discussed in detail from an algorithmic point of view.
Department of Electrical and Electronics Engineering, Ondokuz Mayıs University, Samsun, Turkey
Sercan Demirci*
Department of Electrical and Electronics Engineering, Ondokuz Mayıs University, Samsun, Turkey
Department of Computer Engineering, Ondokuz Mayıs University, Samsun, Turkey
*Address all correspondence to: sercan.demirci@bil.omu.edu.tr
1. Introduction
Wired communication is not a practical option for most of today’s applications due to the costly communication channels, the need for regular monitoring, and the high costs of maintaining its systems. Therefore, a cost-effective and alternative strategy to the wired system is expected to be implemented in industries [1]. Long-standing user demands for wireless connectivity have recently led to the development of cellular communication technologies such as 4G/LTE or 5G and Wi-Fi upgrades such as Wi-Fi 6. Mobile innovations will be required to support existing Internet of Things (IoT) connectivity density and high-engagement applications. The increase in IoT applications significantly changes the wireless network requirements in different aspects. According to the annual internet report published by CISCO for the years 2018–2023 [2],
By 2023, IoT devices will make up about half of all networked devices, and nearly a third will be wireless.
By 2023, traffic from a 5G connection will be around three times that of a 4G connection.
The number of public Wi-Fi hotspots worldwide, which was 169 million in 2018, is predicted to reach approximately 628 million by 2023.
Table 1 shows the global wireless network metrics. From these metrics, the point and trend of wireless communication by 2023 can be better understood.
Global Internet users are expected to increase from 3.9 billion in 2018 to 5.3 billion in 2023, with a compound annual growth rate (CAGR) of 6%, as shown in Figure 1. By 2023, this will account for 66% of the world’s population penetration, having made up 51% of the world’s total in 2018.
Although there is a worldwide trend toward increasing the number of Internet users, there are regional differences as well (Table 2). The Middle East and Africa will likely experience the quickest increase, with a 10 percent compound annual growth rate expected from 2018 to 2023. However, North America will continue to have the highest adoption during the forecast period, followed by Western Europe [3].
Region
2018
2023
Global
51%
66%
Asia Pacific
52%
72%
Central and Eastern Europe
65%
78%
Latin America
60%
70%
Middle East and Africa
24%
35%
North America
90%
92%
Western Europe
82%
87%
Table 2.
Internet users as a percentage of the regional population [3].
In recent years, in parallel with the rapid developments in wireless technologies, the production and use of microelectromechanical systems have become quite widespread. In this direction, Wireless Sensor Networks (WSNs) are at the forefront of the systems that have become widespread in production and use. WSNs, which were used in the military field initially, have started to be used very widely with the decrease in their costs over time and the increase in their detection capabilities thanks to the developing sensor technology. Therefore, it can be said that WSNs are a technology that have a wide variety of applications in both civil and military fields and are developing day by day [4]. WSNs consist of many distributed nodes that monitor a specific physical phenomenon and track a particular object in the area of interest, with the capabilities of sensing, processing, and transmitting the environmental information it obtains [5]. The sensor nodes are small in size. These nodes are equipped with embedded microprocessors and radio transceivers.
In many WSN applications, the deployment of sensor nodes is done on an ad-hoc basis without careful pre-planning and engineering. The most important feature of the sensor nodes is that they can communicate and organize with each other and quickly deliver the data they obtain according to the application areas they serve to a common center. Sensor nodes transmit the data they have obtained from cooperation with each other to a central node called the “Source Node” using the multi-hop paths they are connected to. The source node is a special node that is responsible for collecting incoming data within the task area where the sensor nodes are located and sending this data to the central station for evaluation when necessary. The source node is also called by different names, such as “Monitor Node,” “Sink,” or “Base Station” in the literature [6]. Sensor nodes are generally battery-powered, and in most applications, they are expected to operate for an extended period of time without any outside interference. In these applications, it is very difficult or even impossible to replace or recharge the batteries of the sensor nodes.
The variety of possible applications of WSNs around the world is almost limitless. There are many application areas, from environmental monitoring to health applications, from positioning and monitoring to logistics and locating. WSNs are networks comprising various sensors, including acoustic, magnetic, thermal, optical, infrared, and radar. Because of this, the application domains are broad and practically infinite. Sensor network applications fall into two main categories: tracking and monitoring. Environmental monitoring applications encompass indoor and outdoor environments, energy, health and wellness, inventory location, industrial and process automation, and seismic and structural monitoring. Applications for tracking include following people, cars, animals, and objects [7]. A summary of the uses for WSNs is presented in Figure 2 [8].
The following is a summary of the primary problems and difficulties influencing a WSN’s performance and design found in Ref. [9]: Energy, harsh environmental conditions, self-management, compatibility between hardware and software, redundancy in data, freshness in data, event-driven challenges, quality of service, operating system, deployment, security, fault tolerance, localization, and synchronization are some of the concerns involved.
Localization is one of the key factors influencing a WSN’s performance and design. The issue with localization in WSNs stems from the necessity of the data perceived by the network’s sensor nodes or the location of the event. In WSNs, utilizing the data gathered by the sensors typically necessitates knowing their location. For instance, the physical locations of each sensor are necessary to locate the monitored object to detect and track it using sensor networks. Most of the time, though, the sensors are dispersed throughout the region of interest without their initial location being known. Therefore, once they are deployed, no supporting infrastructure is accessible to locate them. For this reason, some approaches should be found to determine the location of each sensor deployed over the area of interest [10]. In general, localization models can be classified separately according to their intended use, localization method, localization process, distribution area, application area, or anchor type [11]. Figure 3 summarizes the objectives of the localization models.
Figure 3.
Classification of localization models in WSNs according to their purpose [11].
Today, increasing end-user demands and the development of different communication techniques lead to the emergence of new communication technologies, such as CRNs, which require more efficient use of the spectrum [12]. CRNs sense the spectrum by dynamically changing their operating parameters, detecting idle bands, and increasing overall spectrum utilization by opportunistically using available bands. Cognitive Radio (CR) allows temporarily unused spectrum, defined as a spectrum hole or white space [13]. If the licensed user occupies the spectrum band, the CR switches to another spectrum space or remains on the same band by changing the communication power so as not to cause interference. An illustration describing this function is given in Figure 4.
These skills allow them to perform in both licensed and unlicensed bands. Wireless users with appropriate licenses, called Primary Users (PU), are allowed to communicate on the designated frequency in the licensed bands and use the channel. Secondary Users (SU) or CR users are unlicensed users who can utilize the channel without interfering with PUs’ communication ability. When the PU starts to communicate, the CR user detects the available bands, then decides which channel to move to, and finally continues to communicate over the new channel by adjusting the transceiver [15].
Localization is an essential problem for CRNs as well as for WSNs. Therefore, it is necessary to define and solve the localization problem in CRNs. In this study, the localization problem in CRNs is defined, and the algorithmic solutions presented in the literature for the solution to this problem are examined. Chapter 2 is devoted to the research methodology of the relevant studies that form the basis of this study. The problem discussed in Chapter 3 is introduced with its basic components. In the 4th chapter, the algorithmic perspectives proposed for the problem of interest in the literature are examined. In Chapter 5, the main results obtained in this study are given.
In this study, the most valid digital libraries were searched, namely ACM, IEEE, Scopus, Web of Science, and Google Scholar. The phrases used to search these databases and the number of studies found as a result are given in Table 3.
Search strings
Database
Ref.
recordAbstract: (localization” AND “wireless sensor networks” AND “cognitive radio networks” OR “primary user localization)
We consider K CRs localizing a single PU on a two-dimensional Cartesian coordinate system. We denote locations of the PU and the nth CR as lp=xpypT and lk=xkykT, respectively. As seen in Figure 5 [16], we assume that the fusion center is aware of the CRs and that the CRs that are able to hear the PU are distributed uniformly within a circle of radius R.
Figure 5.
Illustration of the PU-centric circular model of CR placement [16].
CR locations are known, and all locations remain constant during the observation period [17].
4. Algorithmic perspective to localization in CRNs
4.1 A semi range-based iterative localization algorithm
A semi-range-based localization technique is presented for SUs to estimate PU placements of PUs in CRNs [18]. This algorithm’s main concept is calculating the distances between SUs and PUs using the estimated detection probabilities obtained from the SUs’ binary detection indicators. An iterative procedure and a weighted least-squares technique are included to improve the accuracy of the suggested localization approach. Moreover, the proposed localization estimator’s lower bound on the mean-square error (also known as the Cramer-Rao Lower Bound, or CRLB) is established. The technique is then expanded to detect malicious users (MUs) in MU-related scenarios. To highlight the advantages of localization, the authors create a location-aware Medium Access Control (MAC) protocol that significantly increases throughput over conventional MAC protocols.
The authors present a semi-range-based, weighted iterative localization method using the derived detection probability. There are three steps to the algorithm. Initially, the least-squares method is used to introduce the fundamental idea of the semi-range-based approach. The mean-square error (MSE) is then minimized using a weighted technique. Lastly, the traffic volume β of PUs is estimated using an iterative method.
4.1.1 First stage: Localization algorithm
The assumption made by the authors is that SU nodes are uniformly distributed across the CRN. The core principle of any range-based algorithm centers around ranging, with PUs being the intended targets. SUs need to remain transparent to PUs, indicating that no collaboration occurs between the two entities during the localization process. This sets CRNs apart from traditional networks, as conventional algorithms, such as those based on Time Difference of Arrival (TDOA) and Received Signal Strength (RSS) are not applicable for estimating distances between nodes and PUs. To address this challenge, the authors proposed a solution. Each SU, based on the outcomes of binary sensing, initially computes the average detection probability. A shared receiver then collects these individual probabilities. Utilizing the collected sensing results and the correlation between distance and detection probability, the common receiver calculates the position of the PU.
Only nodes inside this circle can detect signals from the PU, according to the assumption that the PU creates an interference circle with a radius of ‘r’. If there are less than three nodes nearby, the algorithm breaks. Figure 6 illustrates the basic idea of this localization technique.
Figure 6.
The basic illustration of localization algorithm proposed in [18].
In this illustration, three SUs are used to pinpoint the location of a certain PU. The estimated distances between the SUs and PU are represented by three circles, each having a radius d̂i(i = 1, 2, and 3), respectively. The line going across the intersections of user i and j’s two circles is represented by the notation lij(,i,j = 1, 2, and 3;i≠j). The point where these three circles connect is where the PU is expected to be [18].
4.1.2 Second stage: Weighted linear least-squares method
The variability of distance estimation errors is influenced by the detection probability (PD). On the other hand, unweighted least squares operate under the implicit premise that the variance of the mistakes is identical. Recognizing this, the authors use the weighted least squares approach to enhance the method.
4.1.3 Third stage: Proposed iterative algorithm
The outlined algorithm relies on the awareness of β, a parameter typically unknown to SUs. To address this, an iterative approach is employed to determine β. The iterative process involves selecting an initial value for β and executing the algorithm to obtain an updated value of β, which becomes an output of the localization algorithm. This iteration persists until the predetermined stopping criterion is met.
The following definitions apply to some notations used in the algorithm: βns denotes the estimated degree of the PU’s traffic in the sth interation, P̂Dns denotes the estimated detection probability of the n-user n in the sth iteration, d̂Ns denotes the estimated distance from the n-user to the PU in the sth iteration, and C denotes the number of elements in set C. The algorithm can be found in Ref. [18].
Algorithm 1: The Iterative Localization Algorithm [18].
Set the initial value of β. Let β0=maxnpn;
Forn=1 to NDo.
Ifxn>0Then.
Node n can detect PU, put it into the set ℂ;
End
End
Fors=1 to SDo.
Forn∈ℂDo.
P̂Dns=xn/βs−1T;
d̂ns=f−1P̂Dns;
End
Estimate the position of PU θ̂s by (15).
Ifθ̂s−θ̂s−1<εThen.
Stop;
End
Forn∈ℂDo.
βns=xnfθ̂xs−an2+θ̂ys−bn2T.
End
Let β̂s=1/ℂ∑n∈ℂβns;
End
4.2 Cooperative detection of primary user emulation attacks in CRNs
Due to their unique characteristics, CRNs are susceptible to new security risks, the most serious of which is the primary user emulation (PUE) attack. Ref. [19] describes a cooperative localization strategy that targets CRNs and detects such attacks using TDOA measurements and Taylor-series calculations. In this system, the CRN base station (BS) uses data gathered by a group of CRN nodes to determine the location of a possible primary transmitter using a time-based location algorithm. When determining whether a broadcast is allowed in the case of a TV emitter, the BS considers both this estimate and the actual position of TV primary transmitters, which is presumed to be known to the CRN. It must be noted that the method is unable to resist PUE attacks that rely on primary transmitters with unknown locations, like wireless microphones.
Research [19] shows that TDoA is the only way to find PUE attackers on IEEE 802.22 networks. RSS-based methods cannot accurately detect long-distance links, and AoA-based solutions require additional specialized equipment. GPS-based processes require attacker disclosure. TDoA measurements are inaccurate in real-world environments, and the resulting hyperboloids rarely overlap. In this situation, the location problem is transformed into an optimization problem to reduce the estimation error. Most optimization techniques use iterative techniques because they yield higher accuracy. These techniques require a base estimate that is updated at each iteration. Iterative least-squares estimators and extended Kalman-Buchy filters are two examples of this technology. The exact number of observations has a significant impact on accuracy in both scenarios. More nodes cooperate when there are fewer faults. Using the Taylor series estimator, non-iterative methods such as least squares (LS) outperform TDoA. Although Ref. [19] describes this iterative technique, it is not as complex as filter-based systems like Kalman-Bucy.
Reference [19] also outlines a method for obtaining precise TDoA measurements by synchronizing data from many anchor nodes. The ultimate resolution selected is simple and may be encapsulated in the subsequent five stages:
During the localization process, the BS first asks all CRN anchor nodes to start recording the primary signal, and then it starts recording.
The BS provides a marker signal through the CRN operational antenna during recording, so each anchor node adds this marker to its recording the moment it receives it. Each anchor node can be detected since the marker signal is specific.
Records and markers are sent from each anchor node to the BS. Since the BS knows the exact location of all anchor nodes, it can calculate the elapsed time between marker transmission and reception. Therefore, recordings can be synchronized by BS. Assuming that the BS is located at (0, 0), Eq. (1) can be used to describe the elapsed time, where i represents the anchor node that received the marker, (x, y) represents its location, and vp represents the propagation velocity.
The BS calculates the delay between each anchor node’s recording and its recording and generates a set of TDoA measurements after synchronizing the recordings.
BS calculates the emitter position estimate using the least squares method.
∆ti=xi2+yi2vpE1
4.3 Primary user localization using Bayesian compressive sensing and path-loss exponent estimation
Distance-based received signal strength has been the preferred tool for measuring distance in location processes, although it is inherently unreliable. Most previous publications used a path loss propagation model with a fixed path loss exponent value. However, the root loss exponent for each channel will be different in a real environment. Moreover, as the number of channels increases, more RSS measurements are required, and more energy is consumed due to the complexity of the wireless channel. In order to address this issue, Ref. [20] uses a calibrated path loss exponent in conjunction with a Bayesian compression detection technique to enhance the performance of the PU localization method.
In Ref. [20], the CRN is assumed to consist of SU and PU. Figure 7 shows a general system model with PU at xiyi, the cluster head (CH), and several SUs randomly distributed at xjyj, where j equals from 1 to number of SU. Assume that during the localization process, the positions of PUs, CHs, and SUs are fixed and that the locations of the SUs are known. The localization of one PU is the primary concern since different PUs can be localized, correspondingly, in the same way, that distinct PUs use discontinuous frequency ranges.
Figure 7.
System model of the presented approach in Ref. [20].
Typically, the receive node’s RSS can be represented in Eq. 2 by the log-normal shadowing path-loss model as follows [20]:
PLddB=PLd0dB+10γlog10(dd0)+χdBE2
where PLddB=PtdBm−PrdBm is the path loss at distance d from the primary transmitter. PLd0 is the path loss at a standart distance d0, PLd0 is a fixed quantity and can be found using the free space model with d0 set to 1 m in the small grid and 10 m in the large grid (≥1000 m) and γ=2. χ is a random variable with a zero-mean Gaussian distribution and variance σn2 on a dB scale. γ is the path loss exponent that differs by environment.
In order to enhance the precision of their suggested localization algorithm, Ref. [20] looked into a method of updating the path loss index using calibration. A minor error in the path loss index will cause a significant error in the distance measurement since the path loss model depends on the power law of the path loss index. CRN assumes that the geographic coordinates of all SUs are known. This study used the Angle of Arrivals (AoA) method as a calibration technique instead of the average value to determine the optimal temporal and spatial consistency of the root loss index around the SU. The main flowchart of the proposed localization strategy for CRN is shown in Figure 8. This technique is divided into four phases: initialization, discovery, path loss calibration, and primary user location.
Figure 8.
Overall flow of the proposed localization scheme in Ref. [20].
4.4 Primary user localization algorithm based on compressive sensing
A novel compressive sensing-based primary user localization technique (PU-CSL) in CRNs is proposed in Ref. [21] to identify source signals in allowed frequency ranges more precisely. PU-CSL shows improved locating accuracy for in-depth analysis of the link between the source signal and SUs compared to existing centroid-finding techniques. Energy detection is used at each SU to determine the energy fingerprint of the source signal. After that, compressive sensing (CS) is used to establish weights for the centroid coordinates and determine the degree of correlation between the source signal and the SUs. The source position is eventually estimated using a weighted centroid technique.
The data in Figure 9 are presented assuming that N SUs are randomly distributed throughout the CRN. The source signal gathers the received energy at each receiver and creates an energy fingerprint, which is thought of as its unique identity label based on the correlation between the energy of the received source signal and the position of each receiver. SUs gather the source signal’s energy inside the sensing band during the course of a single sensing period. The energy fingerprint of the source signal node is represented by the following if Ei is defined as the energy collected at the ith SU from the source signal node, where i = 1, 2, …, N, then the source signal node’s energy fingerprint is represented by Eq. (3) [21]:
Figure 9.
The relation between primary user (PU) source signal and secondary users (SUs) [21].
E=E1…Ei…ENTE3
Similarly, the authors define Eik as energy from the ith SU to the kth SU (the theoretical energy value between two nodes can be calculated using the signal transmission model). Then energy fingerprints between the ith cognitive node and all SUs are defined as mi=(Ei1,…,Eik,…,EiNT, where the value of Eik can be obtained using channel transmission model or actual measurement. Collaborative spectrum-sensing nodes in distributed CRNs can use reporting channels to obtain sensing information from neighboring nodes (e.g., energy values required by the algorithm in this study). The two methods used to derive energy values are the channel transmission model and measured values. The authors use the software platform to simulate the method and compute the energy value using the channel transmission model. The measured energy levels are critical when applying the technique in real-world wireless environments.
All of the SUs’ energy fingerprints in CRN are combined in the sparse transformation matrix M [21]:
M=m1…mi…mNTE4
The sparse transforming basis M allows for the representation of the energy fingerprint E [21]:
E=∑i=1Nmiμi=MμE5
where μi represents the degree of correlation between the ith SUs and source signal node, μ=μ1…μi…μNT. As SUs get closer to the source signal node, their signal energy fingerprints begin to converge. In other words, the degree of correlation mi increases as the distance between the Sus, and the source signal node decreases; otherwise, μi decreases or is even zero. The main components of the energy fingerprint E can also be explained by the energy fingerprints of a few close SUs. A tiny percentage of SUs have correlation coefficients higher than zero or extremely near zero. It indicates that the correlation vector μ’s degree is sparse.
Taking into account the CS information operator, the measurement vector y is expressed as [21]:
y=ΦE=ΦMμ=A′μE6
where Φ is the grid transfer connection between each SU and PU source signal, and y is the measurement vector of each SU’s RSS. Basic CS can reconstitute the normalized correlation coefficient μ.
Then the weighted value ωi of each SU can be obtained as follows [21]:
ωi=μi∑i=1NμiE7
Assume SixSix is the coordinate of the ith SU, then the two-dimensional location of source signal node PxPy can be estimated by Weighted Centroid Localization (WCL) [21].
PxPy=∑i=1NωiSixSixE8
Equations (3-8) can be found in Ref. [21]. The source node localization flowchart of the PU-CSL algorithm for one sensing period is shown in Figure 10.
Figure 10.
The node localization flow chart of the primary user localization algorithm based on compressive sensing (PU-CSL) [21].
4.5 Energy-efficient and mobile-aided cooperative localization in CRNs
The work’s authors, published in Ref. [22], suggest a new method for precise PU positioning using mobile-aided cooperative localization called the Energy-Efficient Cooperative Localization Algorithm (EE-CLA). The main objective of EE-CLA is to strike a balance between positional precision and power consumption constraints by selectively activating a suitable number of CRs in collaboration with a mobile CR manager. This is crucial because cooperative communication among CRs tends to consume a significant amount of energy. Additionally, the researchers employ a location-aware CR (LaCR) routing technique to investigate the impact of precise location data on overall performance.
Extensive simulations are run to assess the performance of the proposed EE-CLA and the LaCR procedure. The findings show that in terms of energy efficiency, the proposed EE-CLA performs better than its more straightforward predecessor, the CLA. Moreover, by leveraging the localization data obtained from the EE-CLA, the LaCR protocol achieves noteworthy enhancements in the end-to-end performance of CRs. It mitigates the risk of collision with primary users.
This research employs a mobile CR manager to reduce energy consumption and communication overhead. Through a cooperative approach, the CR manager selectively activates the appropriate number of CRs in local areas, striking a balance between achieving a high probability of PU localization and adhering to power consumption limitations. Additionally, due to its mobility, the mobile CR manager serves as an anchor node, providing spatial diversity and collaborating with permanent CRs to accurately estimate the position of the PU, thereby pinpointing its location precisely. Several fixed local CRs equipped with GPS systems provide RSSI reports to the mobile CR manager. These reports are then utilized to estimate the distance between the mobile CR manager and the local CRs, following Eq. (10), which determines the channel parameters through the process outlined in Section III-C. Subsequently, the position of the mobile CR manager is determined by incorporating the distance calculations with a velocity algorithm [23]. The velocity approach is predicated on a linear prediction model of the speed and prior positions of the mobile CR manager acquired in a noisy channel using the min-max algorithm.
d̂i,j=10RSSId0−RSSIi,j10×α̂i,jE9
The cooperative localization scheme, which operates on an opportunistic wake-up model, controls the on–off states of CRs by adjusting the wake-up probability of each CR in various local regions. Additionally, it guarantees sufficient active CRs to achieve the desired possibility of PU localization [24]. Within this group of CRs, the mobile CR manager employs a cooperative algorithm to identify the optimal set of local CRs within each individual local region for estimating the position of the PU. This process is further elaborated in the subsequent explanation.
Minimizing the localization error yields the ideal set of local CRs that work together to estimate the position of the PU. In the study by Ref. [22], messages are sent among the local CRs rather than the PU working in tandem with them. To find the optimal collection of local CRs for PU positioning, the location estimation of the CR nearest to the PU is considered. Geometrically speaking, the ideal collection of local CRs for placing the closest CR will also be the ideal collection for identifying the PU. While the locations of other nodes within the transmission range are known, it is thought that the nearest CR is located somewhere unknown. Any triplet of local CRs can be used to estimate the position of the nearest CR, but only one triplet will get the best results. Equation (10), which takes the positions of the local CRs and their distances as inputs, applies the min-max localization strategy to accomplish this. The best local CR set is defined as the set of local CRs with the least localization mistake when it comes to placing the closest CR. The local CR set may include local CRs with fixed positions and/or varying positions of the mobile CR manager along its path, it should be mentioned. After that, PU placement is done using the best local CR set.
The optimal local CR cluster and an enhanced min-max estimator are used to localize the PU. The mobile CR manager is given many random paths to choose from by repeatedly executing the random waypoint (RWP) mobility model. Each iteration estimates the PU position using the min-max estimator with the optimal local CR set. At the conclusion of the iterations, various PU position estimates are acquired. Some of these estimates are discarded using an outlier technique, namely the closeness level. The average of the remaining PU position estimations is used to determine the final PU position estimate.
This study focuses on the energy-detection-based technique, which gathers RSSI readings for localization in order to estimate energy consumption. Active local CRs communicate with one another within the local region throughout the data collecting phase.
The summary provided here outlines the procedure using its pseudocode. In particular, the EE-CLA localization and energy consumption procedures are represented by Algorithms 2 and 3.
Algorithm 2: EE-CLA
Data acquisition (RSSI) and PU detection PD
PD=Qd×Pbusy←RSSIi,j,∀i,jCRindex
if PU is detected then
find p̂MCR position
find local region Lr̂p̂MCR←p̂MCR,r̂ local region radius
else
repeat detection process
end if
forl=1:L,L number of local region do
pwl←Ppû=PD×Preq
go to Algorithm 3
estimate, d̂MCR,d̂CRi,d̂CRj from PU
find Best ser for localization P̂MCRP̂CRiP̂CRjd̂MCRd̂CRid̂CRj
P̂PUi←P̂MCRP̂CRiP̂CRjd̂MCRd̂CRid̂CRj
find iterativelyP̂PU=∑i=1kwi×P̂iPU/∑i=1kwi
end for
Estimate the position of PU θ̂s by (15)
Algorithm 3: Energy consumption
pwl,∀l,q are the local region and LCR index
E=∑l=1L∑q=1QEq×pwl
update
4.6 Energy-efficient localization algorithm with improved accuracy in CRNs
A unique, energy-efficient CRN localization mechanism is presented in Ref. [25]. The suggested approach seeks to minimize the total power consumption of the CRNs by giving SUs the best possible transmission range. The placements of PUs and SUs determine the best transmission range since energy efficiency is an essential consideration in the construction of CRN nodes. By using this method, a spectrally efficient CRN can use the spectrum opportunistically while causing little to no interference to the PUs. The algorithm also determines the ideal places for SUs whose locations are known to maximize the CRN’s total localization accuracy and achieve the optimal transmission range. This allows for the localization of the CRN users with the lowest energy consumption for a given target root mean square error. The simulation results show the optimal transmission range for varying numbers of network users, demonstrating the suggested technique’s efficacy.
Using both mere connectivity between PUs and SUs as well as calculated distances between SUs, the algorithm makes use of a Hybrid Connectivity and Estimated Distance (HCED) model. Figure 11 displays the suggested algorithm’s flow chart. The proximity information is denoted by H=ψij2i,j=1K which is the shortest path routes between every user in the network, where ψij is the estimated shortest path route between users i and j. There are two steps in the algorithm that go sequentially. Initially, the most advantageous places are taken by the SUs that are known to exist. The algorithm then determines the ideal transmission range for each SU in the network based on their ideal locations, intending to minimize the overall power usage.
Figure 11.
Flow chart of the proposed algorithm in Ref. [25].
The proposed algorithm’s localization error performance is displayed in Figure 12. In a square region, measuring 30 m × 30 m, there are 50 users in the network, comprising two PUs and 48 SUs. Of these, four SUs out of the 48 SUs possess position information beforehand. The graphic illustrates how SUs with known locations are positioned at each corner of the square region, the ideal places determined by GDOP≤Γ. The threshold value of Γ in the simulations is set to 6. The localization error between each user’s estimated and red lines shows the actual position within the network. At an 8-meter transmission range, the root mean square error (RMSE) for the specified network is around 0.004 meters.
Figure 12.
44 Sus with unknown location, four Sus with optimal location, and two Pus randomly placed in 30 × 30 square area [25].
4.7 A decision-making approach for detecting the primary user emulation attack in cognitive radio networks
A study on a unique algorithm that accurately estimates the PU position based on RSSI by using a trilateration technique and a Bayesian model was carried out in Ref. [26]. The algorithm aims to identify whether the broadcaster is a trustworthy party or a malevolent user to stop possible malicious activity and preserve CRN’s correct operation. Three cost matrices that specify the required productivity levels, security, and balancing schemes are the foundation of this decision-making process. The suggested approach integrates the decision-maker and the Bayesian model together with an uncertainty area that takes into account the conditional distance parameters. This method reduces the possibility of selecting the wrong spot for the PUE by enabling a low-level risk assessment.
The methodology proposed in Ref. [26] focuses on a fixed position within the protected area of the PU. To determine the location of the PUE, the PU utilizes the trilateration technique, which involves three nodes with RSSI. However, the accuracy of PUE attacker localization is affected by errors in the RSSI measurement caused by the signal propagation environment, particularly in uncertain areas. A probabilistic strategy based on Bayesian decision-making is proposed to deal with this problem. This method aims to predict and reject PUE attacks by taking into account the effect of distance on RSSI under various distributions of edge location errors.
The detection technique used in methodlogy is shown in Figure 13. First, the Security, Balanced, or Productivity detection models are chosen to initialize the anchor nodes. A detection mechanism is triggered upon the detection of the PU signal. This process applies the cost matrix parameters, estimates the signal source coordinates, and then uses the Bayesian model treatment. Ultimately, based on these procedures, a determination is made concerning the source of the signal—a genuine PU or an attacker.
The following succinctly describes the proposed PUE attack detection technique in Ref. [26]:
Algorithm 4: PUE attack detection
Choose the decision model (Security, productivity, balanced);
Call procedure localization
ifdN,O≤r1then
Broadcast real PU exit;
else ifdN,O≥r2then
Alert PUE attack and exit;
else ifr1<dN,O<r2then
Apply cost matrix parameters (14);
Apply Bayesian theory (15);
Find the TP;
if PU coordinates > TP coordinates then
Alert PUE attack and exit;
else
Broadcast real PU exit;
end if
end if
procedure localization
Calculate the distance O-anchor nodes using (1)
Estimate the coordinates of PU/PUE with trilateration by solving the system of Eqs. (4a)-(6a)
Set coordinates of PU/PUE
end procedure
In summary, the method presented in Ref. [26] introduces a novel localization technique that identifies PUE emulation attacks by integrating trilateration and RSSI techniques with Bayesian decision theory. This approach aims to determine the decision cost by considering conditional risk, enabling the assessment of the actual risk associated with retaining or relinquishing the PU channel.
4.8 Primary user localization and its error analysis in 5G cognitive radio networks
The motivation for the study conducted by Ref. [27] stemmed from the increasing prevalence of directional sectorized antennas in various applications. These antennas offer the advantage of longer transmission distances and improved performance by focusing the transmission beam toward a specific direction. Consequently, it is essential to consider these characteristics of directional antennas in localization research, which predominantly focuses on omnidirectional wireless communication [28]. Despite the sectorization of cells in cellular networks, it is generally assumed that the antennas used for localization purposes are omnidirectional.
Two major categories can be used to classify the research on PU localization in CRNs based on the transmission capabilities of Pus and Sus:
Omni-directional: The localization of omnidirectional CRNs has been extensively studied. These techniques, which include fingerprinting, trilateration, and multidimensional scaling-based estimation methods, are used to determine the location of a PU [25, 29, 30, 31, 32, 33, 34].
Directional: Many recent studies [17, 35, 36, 37] have shown that using directional antennas can effectively lower localization errors. Specifically, the impact of directional antennas on the localization performance of sectorized cellular networks has been studied by writers in Ref. [37]. PU localization in CRNs with sectorized directional antennas at the Sus was investigated in Ref. [27]. The error performance of the suggested two-dimensional PU localization approach was further examined. Additionally, the traditional Centroid Localization (CL) methodology and the method suggested in Ref. [27] were contrasted in terms of root mean square error (RMSE). The findings show that the suggested strategy is robust and free of center bias.
A two-dimensional CRN arrangement with N Sus and a PU placed in L × L m2 was considered, as shown in Figure 14. The two-dimensional location of i-th SU is denoted by li=xsiysi, and the location of the PU is denoted by lp=xpyp. A central unit was taken into consideration, which gathers data from the SU regarding the existence and absence of the PU and approximates the PU’s location.
Figure 14.
The PU location is represented by the black triangle in the CRN system model, while the SUs are represented by the red circles. Green sectors on each of the three sector antennas on each SU are capable of detecting the presence of the PU [26].
A square area measuring L m on each side was chosen, and the entire region was split up into A × A grids with a resolution of L/A m each. Each of the uniformly dispersed SUs has a beamwidth of 2π3. and three sector antennas. Every SU has a uniform distribution and a random orientation. A summary of the suggested localization technique in Ref. [27] is as follows:
To determine whether PU signals are present in a given channel, the SUs in CRNs perform spectrum sensing. The signal that the SUs get is usually noise if the channel is idle; however, if the channel is busy, they receive both the PU signal and the noise. In such a situation, the existence or absence of the PU signal is often determined using an energy detection model.
The direction of the signal from the PU is then determined by the SUs spanning the angle range −φ0φ0. To put it another way, after an i-th SU receives some power from a PU, it estimates the direction of the signal φik, which lies between −φ0φ0 depending on its bore-sight angle φik in the k-th sector. Figure 15 provides more illustration of this phenomenon.
Select the grid points with location la=xaya for each SU i and assign a value vi to each grid point, where vi = 1 if the grid angle is between −φ0φ0 or else vi = 0.
Choose the grids with the most significant RSS values by aggregating the RSS data from the N SUs. Here, two distinct scoring schemes are chosen in the following manner:
Figure 15.
Illustration of finding the direction of a PU signal [27].
fvi=vivi×piE10
The total number of grid points, K, are chosen for the PU location estimate based on the scoring techniques mentioned above.
Lastly, the PU position is estimated using a grid-based centroid localization approach and is provided as follows:
l̂p=∑a=1Kwala∑a=1KwaE11
where wa=pi−pminpmax−pmin,pmin=minpi, and pmax=maxpi.
In summary, Ref. [27] has proposed a grid system-based PU localization approach. Sectorized antennas are installed on the SUs, and different scoring functions are used to locate the PU by looking at the grids with the highest scores. The proposed method’s mean square error was also calculated for theoretical analysis.
4.9 Efficient-spectrum management based on the localization of primary user position toward 5G
The CRN described in Ref. [38] uses an interweave strategy in which the primary and secondary systems share a licensed spectrum band for primary use. With this method, unlicensed users ensure no interference is made to the current primary network while using spectrum-sharing techniques to investigate and take advantage of spectral holes. The crucial first step in accomplishing this is the spectrum-sensing system, in which the SU finds empty bands that licensed (primary) users subsequently reuse. In order to prevent harmful interference to other SUs and preserve the quality of service (QoS) for the PU, SUs must adequately identify the signal generated by the PU from other SU signals to assure proper spectrum usage [39, 40].
The material mentioned above does, however, have some limitations because it needs to take into account how dynamic the environment is in a CRN. The critical users in a typical environment are in flux, shifting positions quickly at any given time. It is difficult for SUs to identify idle bands because of the volatility of the spectrum holes. SUs must do the difficult task of determining which PU is using the spectrum during spectrum detection since the licensed user may suddenly reenter the band at any moment. If the SU continues to utilize the opportunistic band, this could lead to crashes. As a result, inefficient spectrum management results in less-than-ideal use.
Locating the locations of dispersed PUs within the network coverage region is one potential solution to this problem. The primary idea of a study cited as Ref. [38] centers on simultaneously determining the spectrum gaps and the PUs’ locations using a tracking procedure. The goal is to efficiently use the resources in the licensed spectrum to satisfy the QoS demands of the growing number of intelligent and connected devices.
This research study suggests using Kriging interpolation and Kalman Filter (KF) localization techniques to get beyond the aforementioned restrictions as well as difficulties with PU localization and opportunistic spectrum access for SUs in the examined environment.
The dynamic environment depicted in Figure 16 is what the secondary network’s SUs strive to reuse—that is, the spectrum allotted to the primary network. The region of interest is shown by the enclosed square area on the left side of the picture. The SUs are sensing devices that sense their surroundings and report in real time to the cognitive engine or radio resource management function.
Figure 16.
Primary network and secondary network are sharing the same spectrum in dynamic environment with sensors distributed in the coverage area [38].
The PUs in a CRN have free reign over the spectrum. On the other hand, the SUs take advantage of the licensed spectrum by accessing it opportunistically, depending on the PU’s state, so as not to impede communication inside the core network. By measuring the RSS from any place within the dynamic environment, the sensor nodes depicted in Figure 16 play a critical role in supporting the SUs [41]. This enables effective detection and monitoring of the PU’s status, facilitating efficient access to opportune spectrum resources. Additionally, the RSS measurements are influenced by the transmit powers of both PUs and SUs, ultimately revealing the system’s state.
The first stage is estimating a Received Power Estimation Map (REM) that shows the predefined power level received at every pixel in the coverage region of the network. After the REM is estimated, it is processed to determine which pixels have higher power levels than others. This last procedure aids in locating possible PU spots. The KF has been introduced by the authors in order to enhance the position estimate of the non-stationary PU in Ref. [38]‘s model. In a dynamic two-dimensional environment, the KF is used to swiftly estimate the expected position of the moving PU. The KF is thought to be the best method for obtaining precise location estimation. To ensure accurate estimation, the KF scans the principal network point by point, as shown in Figure 17a and b. There are two key distances in this process:
that between the PU and a considered SU.
that between primary TX and the considered node.
Figure 17.
KF tracking algorithm for PU position estimation proposed in Ref. [38]. a) Block diagram of the proposed scheme, b) Iterative process to calculate the estimate position [38].
The primary objective of network scanning is to monitor the position of the PU to determine the distance between the PU and a specific SU. This estimation of separation between the PU and the considered node is achieved by utilizing the equation of free space path loss [42]. To simplify the analysis, it is assumed that the PU’s state within the coverage area can be characterized by its velocity and position in a two-dimensional Cartesian plane. Furthermore, in the scenario presented in Ref. [38], it is assumed that the initial position of the PUs is known.
In summary, for effective spectrum management in dynamic CRNs, the KF and Kriging interpolation algorithms were presented in Ref. [38]. The assessment of the PU position became essential since the objective is to opportunistically use the primary system’s existing spectrum under an efficient spectrum sharing technique. In Ref. [38], KF and Kriging interpolation both perform well in the case under consideration when compared to RSSI-based techniques.
4.10 Interference avoidance in cognitive radio networks using cooperative localization
The authors in Ref. [43] expanded on their prior research [44] by implementing cooperative localization to prevent interference. They utilized a log-normal shadowing path loss model to estimate the distance between an SU and a PU in the interference link. This estimated distance was then utilized to adjust the transmission power to avoid interference. The simulation results demonstrated significant improvements in performance due to Algorithm 5 using the proposed approach. The following summarizes this article’s main goals and contributions.
Working together, the primary and secondary networks reduce interference and improve the effective utilization of the spectrum. Analysis of the interference relationship between a PU and an SU is done using a log-normal shadowing model. The major and secondary networks are then able to cooperate more easily thanks to this approach. More specifically, it helps determine the separation between a PU and an SU.
A comprehensive and resilient cooperative localization framework is introduced. A PU employs the framework to determine its own position in collaboration with neighboring SUs.
For the purpose of assisting cooperative localization, the SUs regularly broadcast beacon messages. Using the data transmitted by these beacon messages and the collaborative localization paradigm, a PU can ascertain its location. This positioning information is then used to improve the accuracy of distance calculation between a PU and an SU.
To avoid interfering with the PU, the SU modifies its transmission power in accordance with the distance data between the two devices.
A series of extensive simulation experiments is carried out to assess the effectiveness of different components of the suggested approach. The evaluation specifically focuses on channel availability, capacity, bandwidth use, and estimation of distance and position. The findings demonstrate that the algorithm attains significant performance improvements due to the implementation of Algorithm 5.
4.10.1 Cooperative localization
All of the SUs in the CRN are aware of their own position data. Moreover, through cooperative localization, a PU and an SU can work together to estimate a position. For this process to be easier, every SU sends out a beacon message, as seen in Figure 18. In this beacon message, ts denotes the time stamp, (xq, yq) represents the SU q position, and Ptq stands for the transmission power. Upon receiving the beacon message, the PU ascertains the RSS. This RSS value is subsequently employed to estimate the path loss and determine the distance.
For more details on the position estimation process, it is assumed that the PU p receives beacon messages from n number of SUs. The PU is able to determine the positions of these SUs from the beacon messages, which are given by (x1, y1), (x2, y2), …, (xq, yq), …, (xn, yn).
4.10.2 Interference avoidance
The proposed IA algorithm in Ref. [43] is explained in this section. One part of the algorithm is executed by the SU q and the other part is executed by the PU p.
An information beacon message, as depicted in Figure 2, is transmitted by each SU periodically. The PU p measures the RSS when the beacon message is received. The path loss can then be calculated using the RSS information. Let the RSS of the SU q transmitted interference signal at the PU p be represented by Pqp. The path loss is then given by
LpdB=PtqdB.−PqpdBE12
After estimating the path loss with the help of RSS information, the distance is also ascertained as follows:
logdpq=110ηLpdB−Lp¯−XσE13
dpq=10110ηLpdB−Lp¯−Xσ.E14
The PU p estimates this distance dpq using the RSS value. Using the position data for each, one can also estimate the distance between the SU q and the PU p. As stated earlier, each SU already knows its location using GNSS or a localization algorithm.
Interference should be avoided as a result of the PU p and the SU q exchanging a pair of messages. If there is shadow fading, the process might need to be repeated. In the latter scenario, a periodic beacon message is transmitted by the SU q. The SU q changes its transmission power once more as the PU p responds with a response message. Until the interference drops below the threshold and no more messages can be sent between the PU p and SU q, the process is repeated. Algorithms 5 and 6 provide a summary of the algorithms that the PU p and SU q executed.
Algorithm 5: The Primary User
procedure POSITION ESTIMATION()
Receive beacon message
ts,xqyq,Ptq← Beacon message
Pqp←RSS
LpdB←PtqdB−PqpdB
dpq←10110ηLpdB−Lo¯−Xσ
if (n≥3) then
Formulate Az = R
Derive Hz = T
Determine z from Hz = T
xy←z
else
xy←−1−1
end if
PU message ←Lp,dpq,xy
Transmit PU message
end procedure
Algorithm 6: The Secondary User
procedure TRANSMITPOWER()
Transmit beacon message
Receive PU message
Lp,dpq,xy← PU message
ifxy≠−1−1then
dxy←xq−x2+yq−y2
d←dxy
else
dxy←−1
d←dpq
end if
if (d≤R) then
PtqndB←LpdB+1−θIptdB
end if
if (Ptqn≥Ptq) then
while (Ptqn≥Ptq) and (d≤R) do
σ←σ−1
LpdB←Lo¯+10ηlogd+Xσ
PtqndB←LpdB+1−θIptdB
R←10110ηLpmaxdB−Lo¯−Xσ
end while
end if
end procedure
In summary, in order to prevent interference, PU-assisted localization has been utilized in Ref. [43]. The path loss model in conjunction with localization is utilized to calculate the separation between the PUs and SUs. Algorithm 5 to the PU uses SU transmission power adjustment once the distance has been estimated.
4.11 Detection of primary user emulation attack using the differential evolution algorithm in cognitive radio networks
The main contributions of Ref. [45] are outlined as follows:
A highly effective technique is proposed to reduce the Primary User Emulation Attack (PUEA) efficiently.
The Differential Evolution (DE) algorithm is combined with a TDOA-based localization approach.
The intervention during the spectrum-sensing process is minimized, resulting in increased bandwidth utilization.
The fitness function is optimized to estimate the attacker’s position accurately.
The correct detection probability is enhanced.
The false detection probability is reduced.
The approach presented in Ref. [45] consists of a CRN comprising a Cognitive Radio Base Station (CRBS) as its central point, along with randomly deployed N SUs. Additionally, there is a PU setup consisting of a PU transmitter and multiple fixed receivers. Figure 19 provides a visual representation of this configuration. The CRBS possesses knowledge of the positions of both the PU transmitter and the SUs. The distance between the PU transmitter and the CRN ranges from 30 km to 100 km. The PU is positioned outside the CRN at a specific location.
The primary objective of Ref. [45] is to identify the attacker (PUEA) within or outside the CRN with utmost precision. This is achieved by effectively recognizing the intruder when it is close to the PU. The approach aims to reduce the localization positioning inaccuracy while minimizing the number of coordinating clients needed and the time to locate the attacker.
4.11.1 Mathematical model for TDOA localization technique
A framework outlining the process of TDOA localization is presented below:
An unidentified source transmits a modified signal.
Three or more receivers around the source capture the transmitted signal.
To determine the precise location of the source, the time each receiver captures the signal is calculated.
The variation in signal arrival time at each receiver is crucial, as it enables the calculation of the differences in distance between the receivers.
The time difference is computed using hyperbolic lines.
The intersection points of these hyperbolas indicate the target’s position as determined by TDOA.
However, due to potential inaccuracies in the calculated TDOA values, the intersection points of the hyperbolas may not always accurately pinpoint the target. To address this, optimization algorithms are employed in conjunction with TDOA to locate the target accurately.
The following is a description of the computed model for the TDOA localization method:
(a, b) represents the exact position of the unknown sender. The sender can be an authorized user or attacker.
(a0, b0) represents the position of CRBS.
(ai, bi) represents the position of the ith SU.
ri indicates the interspace of the unknown sender with the ith SU.
r0 indicates the interspace of CRBS with the unknown sender.
ri,0 represents the real distance difference of the PUEA with the ith SU, where CRBS is the static node.
Additionally, Firefly Optimization Algorithm was used as the optimization algorithm in Ref. [45]. A flash of light is used to search for food and is considered the fitness function while an issue is being optimized.
4.11.2 Detection of PUEAs using the DE-based localization algorithm
Using cross-correlation at several nodes, the TDOA approach determines the difference in signal arrival. Specifically, the TDOA uses two hyperbolic curves that meet at a predetermined point to determine the transmitter’s position. Next, this position is contrasted with the PU location in terms of the PU error analysis. The DE algorithm is applied to improve the accuracy of the location identification process. The DE helps to achieve a high fitness value for the optimization job by minimizing potential errors through iterative optimization of the search space, as shown in Figure 20.
Figure 20.
Workflow of TDO- based localization techniques using Differential Evolution [45].
The primary objective of the proposed algorithm is to accurately and precisely detect the PU Error Analysis. To achieve this, three key steps must be followed:
The received signals and the PU and SU signals’ properties have to match.
The TDOA measurements, obtained using the DE algorithm, should correspond to an appropriate distance.
To distinguish between the PU and the PU Error Analysis, the measured position is then compared to the location of the PU transmitter.
In Ref. [45], to address the PUEA, a localization strategy based on DE and TDOA is suggested. One PU, fifty random SUs, a PUEA, and a CRBS comprise the suggested CRN. The PUEA can be found inside or outside the network and is fixed. Because of how the network is set up, both the attacker and the SUs know where the PU is. When sensing the spectrum, all SUs send sensing data to the CRBS. To find the TDOA measurements, the CRBS gathers all the data and uses cross-correlation. The location of the sender—whether a PU or a PUEA—is then ascertained using these variables. The DE algorithm is used to provide precise estimation and to reduce the MSE so that convergence occurs faster.
A vital resource for wireless communications is the radio spectrum. The CR method reduces spectrum scarcity and increases the efficiency of spectrum utilization. For the CR to function at its best, the PU’s operations must be stopped. An extensive introduction to the subject of localization techniques for CRs is provided in this article. In particular, the distinctions between cognitive network localization and classical network localization are explored, and the impact of different parameters on the efficiency of localization algorithms in CRs is looked into. Meanwhile, some current localization algorithms do not account for the non-cooperative character of PU; future research paths in CRNs may focus on developing localization algorithms that do not incorporate valuable information from the PU.
References
1.Sonai V, Bharathi I. An algorithmic approach to the node selection problem in industrial wireless sensor networks. In: Wireless Sensor Networks-Design, Deployment and Applications. London, UK: IntechOpen; 2021
2.Cisco. Cisco Annual Internet Report (2018–2023): White Paper [Internet]. 2020 [cited 2024 FEB 24]. Available from: https://www.cisco.com/c/en/us/solutions/collateral/executive-perspectives/annual-internetreport/white-paper-c11-741490.html
3.Cisco Annual Internet Report (2018–2023) White Paper. Updated: March 2020. Available from: https://www.cisco.com/c/en/us/solutions/collateral/executiveperspectives/annual-internet-report/white-paper-c11-741490.html [Accessed: 3 Jan 2024]
4.Akyildiz IF, Su W, Sankarasubramaniam Y, Cayirci E. Wireless sensor networks: A survey. Computer Networks. 2002;38(4):393-422
5.Misra S, Misra SC, Woungang I, editors. Guide to Wireless Mesh Networks (Vol. 9, No. 9). London, UK: Springer; 2009
6.Tiliute DE. Battery management in wireless sensor networks. Elektronika ir elektrotechnika. 2007;76(4):9-12
8.Zhang Z, Mehmood A, Shu L, Huo Z, Zhang Y, Mukherjee M. A survey on fault diagnosis in wireless sensor networks. IEEE Access. 2018;6:11349-11364
9.Bala T, Bhatia V, Kumawat S, Jaglan V. A survey: Issues and challenges in wireless sensor network. International Journal of Engineering &Technology. 2018;7(2):53-55
10.Xiang Ji. Localization Algorithms For Wireless Sensor Network Systems, A Thesis in Computer Science and Engineering by Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy, The Pennsylvania State University The Graduate School Department of Computer Science and Engineering (PhD thesis). Pennsylvania; 2004
11.Alomari A, Phillips W, Aslam N, Comeau F. Dynamic fuzzy-logic based path planning for mobility-assisted localization in wireless sensor networks. Sensors. 2017;17(8):1904
12.Mumey B, Zhao X, Tang J, Wolff R. Transmission scheduling for routing paths in cognitive radio mesh networks. In: Sensor Mesh and Ad Hoc Communications and Networks (SECON), 2010 7th Annual IEEE Communications Society Conference. Boston, MA, USA: IEEE; 2010. pp. 1-8
13.Haykin S. Cognitive radio: Brain-empowered wireless communications. IEEE Journal on Selected Areas in Communications. 2006;23(2):201-220
14.Akyildiz IF, Lee WY, Vuran MC, Mohanty S. NeXt generation/ dynamic Spectrum access/cognitive radio wireless networks: A survey. Computer Networks. 2006;50(13):2127-2159
15.Akan OB, Karli OB, Ergul O. Cognitive radio sensor networks. Network: The Magazine of Global Internetworking. 2009;23(4):34-40
16.Werner J, Wang J, Hakkarainen A, Valkama M, Cabric D. Primary user localization in cognitive radio networks using sectorized antennas. In: 2013 10th Annual Conference on Wireless on-Demand Network Systems and Services (WONS). Banff, AB, Canada: IEEE; 2013. pp. 155-161
17.Wang J, Chen J, Cabric D. Cramer-Rao bounds for joint RSS/DoA-based primary-user localization in cognitive radio networks. IEEE Transactions on Wireless Communications. 2013;12(3):1363-1375
18.Ma Z, Chen W, Letaief KB, Cao Z. A semi range-based iterative localization algorithm for cognitive radio networks. IEEE Transactions on Vehicular Technology. 2009;59(2):704-717
19.León O, Hernández-Serrano J, Soriano M. Cooperative detection of primary user emulation attacks in CRNs. Computer Networks. 2012;56(14):3374-3384
20.Anh H, Koo I. Primary user localization using Bayesian compressive sensing and path-loss exponent estimation for cognitive radio networks. KSII Transactions on Internet & Information Systems. 2013;7:10
21.Ye F et al. Primary user localization algorithm based on compressive sensing in cognitive radio networks. Algorithms. 2016;9(2):25
22.Kianoush S, Vizziello A, Gamba P. Energy-efficient and mobile-aided cooperative localization in cognitive radio networks. IEEE Transactions on Vehicular Technology. 2015;65(5):3450-3461
23.Kianoush S et al. Low-complexity localization and tracking in hybrid wireless sensor networks. International Scholarly Research Notices. 2012;2012:1-7. Article ID 430169
24.Xia F et al. Energy-efficient opportunistic localization with indoor wireless sensor networks. Computer Science and Information Systems. 2011;8(4):973-990
25.Saeed N, Nam H. Energy efficient localization algorithm with improved accuracy in cognitive radio networks. IEEE Communications Letters. 2017;21:9
26.Fihri F, Wassim, et al. A decision-making approach for detecting the primary user emulation attack in cognitive radio networks. International Journal of Communication Systems. 2019;32(15):e4026
27.Saeed N, Nam H, Al-Naffouri TY, Alouini MS. Primary user localization and its error analysis in 5G cognitive radio networks. Sensors. 2019;19(9):2035
28.Türkyilmaz O, Alagöz F, Gür G, Tugcu T. Environment-aware location estimation in cellular networks. EURASIP Journal on Advances in Signal Processing. 2008;2008:1-9
29.Saeed N, Nam H. Robust multidimensional scaling for cognitive radio network localization. IEEE Transactions on Vehicular Technology. 2014;64(9):4056-4062
30.Saeed N, Nam H. Cluster based multidimensional scaling for irregular cognitive radio networks localization. IEEE Transactions on Signal Processing. 2016;64(10):2649-2659
31.Saeed N, Haris M, Haq MIU. Jointly locating the primary and secondary users in cognitive radio networks. In: 2017 International Conference on Communication, Computing and Digital Systems (C-CODE). Islamabad, Pakistan: IEEE; 2017
32.Bulusu N, Heidemann J, Estrin D. GPS-less low-cost outdoor localization for very small devices. IEEE Personal Communications. 2000;7(5):28-34
33.Mariani A et al. Cooperative weighted centroid localization for cognitive radio networks. In: 2012 International Symposium on Communications and Information Technologies (ISCIT). Gold Coast, QLD, Australia: IEEE; 2012
34.Nam H et al. Primary user localisation and uplink resource allocation in orthogonal frequency division multiple access cognitive radio systems. IET Communications. 2015;9(8):1131-1137
35.Wang J, Chen J, Cabric D. Stansfield localization algorithm: Theoretical analysis and distributed implementation. IEEE Wireless Communications Letters. 2013;2(3):327-330
36.Penna F, Cabric D. Cooperative DoA-only localization of primary users in cognitive radio networks. EURASIP Journal on Wireless Communications and Networking. 2013;2013:1-14
37.Büyükçorak S, Kurt GK, Yongaçoğlu A. Received signal strength based localization in sectorized cellular networks. In: 2016 23rd International Conference on Telecommunications (ICT). Thessaloniki, Greece: IEEE; 2016
38.Gashema G, Lee J-M, Kim D-S. Efficient-spectrum management based on localisation of primary user position towards 5G. IET Communications. 2020;14(20):3567-3577
39.Kim Dong Seong, Jae Min Lee. Spectral Efficiency Improvement Based on Small-Cell Deployments toward 5G Communications Technologies. Gyeongsangbuk-do, Republic of Korea: IEIE Transactions on Smart Processing and Computing; 2019
40.Le TT, Kim D-S. An efficient throughput improvement through bandwidth awareness in cognitive radio networks. Journal of Communications and Networks. 2014;16(2):146-154
41.Bouraoui R, Besbes H. Cooperative spectrum sensing for cognitive radio networks: Fusion rules performance analysis. In: 2016 International Wireless Communications and Mobile Computing Conference (IWCMC). Paphos, Cyprus: IEEE; 2016
42.Fihri WF et al. Bayesian decision model with trilateration for primary user emulation attack localization in cognitive radio networks. In: 2017 International Symposium on Networks, Computers and Communications (ISNCC). Marrakech, Morocco: IEEE; 2017
43.Farooq-i-Azam M, Ni Q, Dong M, Pervaiz H, Zarakovitis C, Alsolami F. Interference avoidance in cognitive radio networks using cooperative localization. IEEE Systems Journal. 2022;16(4):5120-5130
44.Azam MFI, Ni Q, Dong M, Pervaiz HB. Cooperative localization based interference avoidance in cognitive radio networks. In: ICC 2021-IEEE International Conference on Communications. Montreal, QC, Canada: IEEE; 2021. pp. 1-6
45.Batool R, Bibi N, Muhammad N, Alhazmi S. Detection of primary user emulation attack using the differential evolution algorithm in cognitive radio networks. Applied Sciences. 2022;13(1):571
Written By
Dogan Yildiz and Sercan Demirci
Submitted: 23 February 2024Reviewed: 18 October 2024Published: 08 January 2025
Open access peer-reviewed chapter
