Search

Breast cancer is currently one of the leading causes of death among women worldwide. Masses are considered significant signs of the existence of malignant lesions, as they occur in most breast cancer cases. However, their detection is challenging since masses have large variation in shape, margin, size and are often indistinguishable from surrounding tissue, making the radiologist's task tedious in the case where a significant number of mammograms require fast and accurate interpretation. For these reasons, computer-aided diagnosis (CAD) systems are being developed to make the diagnostic process easier for radiologists. In these systems, segmentation and classification of breast masses in mammograms are important steps. This work aims to evaluate the performance of machine learning techniques in classifying tumors into benign and malignant. The selected techniques were applied on 1663 mammograms from the Digital Database for Screening Mammography. Of the 1663 images, 769 images correspond to malignant cases, and 894 correspond to benign cases. The efficiency of each of the considered techniques was evaluated by using four metrics, namely, the false positive rate, sensitivity, specificity, and accuracy.

Smart grid is an emerging technology that delivers intelligently to the end-users through two-way communication. However, this technology can be subject to several cyber-attacks due to this network's inherent weaknesses. One practical solution to secure smart grid networks is using an intrusion detection system (IDS). IDS improves the smart grid’s security by detecting malicious activities in the network. However, existing systems have several shortcomings, such as a low detection rate and high false alarm. For this purpose, several studies have focused on addressing these issues, using techniques, including traditional machine learning models. In this paper, we investigate the performance of three different ensemble learning techniques: bagging-based, boosting-based, and stacking-based. Their results are compared to those of three traditional machine learning techniques, namely K nearest neighbor, decision tree, and Naive Bayes. To train, evaluate, and test the proposed methods. We used the benchmark of CICDDos 2019 that consists of several DDoS attacks. Two feature selection techniques are used to identify the most important features. The performance evaluation is based on the probability of detection, probability of false alarm, probability of miss detection, and accuracy. The simulation results show that the stacking-based ensemble learning techniques outperform the other algorithms in terms of the four-evaluation metrics.

Alzheimer's disease (AD) affects fifty million people worldwide and is the sixth cause of death in the United States. However, there is no cure or treatment for patients with AD; thus, it is important to detect this disease at an early stage to improve patients' lives qualities. Several studies have been proposed to detect and differentiate between different AD groups, although most of these works only focused on differentiating between healthy people and people with Alzheimer's. These studies also did not identify the most reliable biomarkers to provide more accurate results and did not use the best hyperparameters to provide optimal results. To address these issues, we developed a model that leads to a better performance in differentiating between healthy people (cognitively normal), people with mild cognitive impairment, and people with Alzheimer’s disease. For this purpose, we combined a stacking-based ensemble learning, consisting of four traditional classifiers, with a hyperparameter tuning technique, a genetic algorithm. The model was evaluated in terms of accuracy, precision, recall, and F1-score. The simulation results show that stacking-based ensemble learning, using genetic algorithm, provides 96.7% accuracy, 96.5% recall, 97.9% precision, and 97.1% F1-score in differentiating between CN, MCI, and AD groups.

The forecasting of Wind power generation plays a critical role in the safe and stable operation of a power grid. Grid operators rely on the short-term forecasts of load and generation sources to optimize operations such as unit commitment and economic dispatch. These forecasts needs to be stable and efficient because of the low dispatchability and increasing percentage of renewable energy sources in the generation mix. We will describe the results of our performance study with different forecasting methodologies and will also propose hybrid methods for delivering consistent results with a varying dataset. The National Renewable Energy Laboratory (NREL) wind integration dataset having 5 predictor variables and a data resolution of 5 minutes is used for this analysis. Forecasting methodologies evaluated include ARIMA, RF, SVM, GLM, GAM and four additional hybrid methods. We will reveal the robust models of GLM and GLM based hybrid methods to deliver consistent forecasts of wind power generation.

Social Engineering outweighs any other security threats as it has proven to be one of the easiest, cheapest, and most potent and highly successful ways for criminals to achieve their ends. We have witness how globally governments, organizations and institutions down to every individual have embraced the state-of-the-art technological advances and interconnectedness brought about by networks and telecommunications. We all together experience the ease and comfort of these modern living; however, many malicious actors also identified these technological vehicles as a means to selfishly benefit themselves. Cybercriminals have found humans to be an easy prey to victimize and fall for their intent. Hence, we have presented the significance of mitigating this type of attack in order to give precaution on the danger of what social engineering can ensue. Following the quote of Sun Tzu "Know Your Enemy", the initial step to plan and perform for an effective defense strategy is similar to the military operations, which is to do a reconnaissance Synthesis analysis using the rapid synthesis approaches to systematically identify the primary studies and have followed an orderly series of steps in order to accomplish our exploration. Applying the same technique, we have investigated the current existing social engineering attacks and countermeasures and performed a distillation of some selected studies regarding attacks in social engineering.

Unmanned aerial vehicle (UAV) networks offer enormous potential in civil, commercial, and military applications. As network sizes become large, the communication of UAV networks poses serious cybersecurity challenges. One solution for providing a secure, scalable communication mechanism is to integrate blockchain in the peer-to-peer UAV networks. Blockchain provides a way for multiple entities to communicate securely in a decentralized and cooperative manner. Because blockchain is primarily used for preventing double-spending in cryptocurrency, it lacks methods to secure communications in a network. Specifically, this paper proposes a novel consensus-building mechanism for securing communications in a UAV network, integrating blockchain with public key cryptography method Elliptic Curve Diffie-Hellman and one-time pad encryption method.

Breast cancer is the most common form of cancer among women and the leading cause of female deaths from cancer worldwide. Microcalcifications, small crystals of calcium apatites, are considered the first sign of breast cancer. Since microcalcifications are small and have different shapes and low contrast, they can be easily missed or misinterpreted by radiologists. For these reasons, automatic image processing systems are being developed to make the diagnostic process easier for radiologists. In this work, we present a computer-based automated approach for segmentation and classification of breast microcalcifications in mammograms using discrete wavelet transform and random forest (DWT-RF). The proposed approach was tested on 966 images (322 benign, 322, malignant, and 322 normal) from the Digital Database for Screening Mammography. The results indicate that DWT-RF achieves a sensitivity of 93%, a specificity of 97%, a false positive rate of 3%, an accuracy of 95%, and an area under the ROC curve of 0.92, which are comparable in terms of accuracy to state-of-the-art methods and other existing classifiers.

Alzheimer's disease (AD) is a neurodegenerative brain disorder and the fifth leading cause of death among people aged 65 and older. Based on recent research, it was found that in addition to cognitive tests, quantitative biomarkers can be useful indicators for monitoring the progress from Mild Cognitive Impairment (MCI) to Alzheimer's disease. Hence, identifying the most relevant biomarkers and cognitive tests can lead to a more reliable and accurate diagnosis of AD. Therefore, this study aims to identify the most pertinent cognitive tests and biomarkers, features, to detect Alzheimer's disease. This aim is achieved by using six conventional feature selection methods. In addition, we used a feature combination approach to find the best subset of the features that can lead to the highest accuracy in differentiating between healthy subjects, early mild cognitive impairment (EMCI), and AD patients. Unlike conventional feature selection methods that select the Clinical Dementia Rating Scale Sum of Boxes (CDRSB) as a unique feature, the proposed feature combination method selects this CDRSB as well as the Middle temporal gyrus (MidTemp). The results show that this combination gives the highest accuracy in differentiating between cognitively normal (CN), EMCI, and AD groups.

This paper discusses how visualization and machine learning models can be effectively used to track and forecast trap counts of Culex Tarsalis, female mosquitoes responsible for spreading the West Nile Virus (WNV). This paper applies four different machine learning models namely, Support vector machines (SVM), Regression tree (RT), Partial Least Square Regression (PLSR), and a hybrid combination of SVM-PLSR to multi-year WNV data sets. Precisely, historical data sets ranging from 2005–2015 was used to predict trap counts for 2016. This paper also discusses a tree-based data visualization technique for displaying historical trap counts. The visualization model is designed to focus on identifying trends in the behavior of Culex tarsalis by tracking parameters such as meteorological data, dead birds, WNV cases, human cases, and deaths. The preliminary results indicate that SVM model outperforms interms of accuracy than other machine learning models. Keywords: Culex Tarsalis, Partial Least Squares Regression (PLSR), Support Vector Machines(SVM), SVM-PLSR, West Nile Virus (WNV), Vector Control.

Unmanned Aerial Systems (UAS) are vulnerable to a number of cyber-attacks. To navigate and avoid collisions, they rely on accurate information from GPS and Automatic Dependent Surveillance-Broadcast devices. Signals of these devices are can be jammed, spoofed, deleted, and modified. Identifying the location of an attacker can detect and mitigate cyber-attacks. This paper proposes a novel approach to identifying the angle of arrival of a signal. The approach consists of using a spinning radome with an attenuating, parasitic, section. The attenuating section spins around to create a periodic amplitude modulation on the received signal while not breaking the communication link. The amplitude of the signal drops when the attenuating section is in-line with the transmitter attacker. When the received signal is the lowest, the angle of the radome attenuator corresponds to the angle of arrival of the signal of interest. The results show that by using two measurements of the angle of arrival at different locations, one can calculate the attacker/transmitter's location with good precision. Two radomes were used for measurements at various locations.

Breast cancer is the most common cancer among women and the second leading cause of female cancer mortality following lung cancer. Breast cancer also affects men but with a low incidence rate. However, the survival rate for men is much lower than women's. Early screening to identify breast abnormalities can increase the survival rate. Since the causes of this disease are not all known, primary prevention seems impossible. Therefore, early detection is the major key to surviving this disease. If breast cancer is detected earlier, there are more treatment options and a better chance for survival. The first sign of numerous breast cancer cases is the presence of microcalcifications in mammograms, which are small crystals of calcium apatites ranging from 0.1 mm to 0.5 mm. In this paper, we propose a multi-objective optimization approach to determine the optimal boundary between the signal subspace and noise subspace in TR-MUSIC algorithm.

Spectrum sensing plays an important role in enabling cognitive radio technology for the up-and-coming generation of wireless communication systems. Over the last decade, several sensing methods have been proposed, including energy detection, cyclostationary feature, and matched filter. However, these techniques present several limitations. Energy detection performs poorly under low signal-to-noise ratio, cyclostationary features are complex, and matched filter requires some prior knowledge about the primary user signal. In addition, all of these techniques require setting a threshold which needs the prior knowledge of the noise distribution. Thus, the reliability of spectrum sensing is still an open issue in wireless communication research. In this paper, we propose a spectrum sensing method based on a machine learning theory for cognitive radio networks. The spectrum sensing problem is rigorously modeled and out of which a large-scale comprehensive dataset is built. This dataset is then used to train, validate, and test several machine learning techniques, including random forest, support vector machine with different kernels, decision tree, Naive Bayes, K-nearest neighbors, and logistic regression. The models were extensively tested and evaluated using metrics such as the probabilities of detection, false alarm, and miss-detection as well as the accuracy of the classification. The simulation results show that the random forest model outperforms all the other machine learning methods.

Aviation communities are nowadays deploying automatic dependent surveillance-broadcast (ADS-B) systems on aircraft to have more accurate and reliable air traffic control. This system broadcasts some of the aircraft telemetry data in the form of plaintext messages over unencrypted datalinks, which makes this system vulnerable to several cyber-attacks including jamming. To mitigate these attacks, the first step to take is to detect them. A few techniques have been proposed in the literature, but these techniques suffer from several limitations. This paper presents a comprehensive study highlighting the efficiency of machine learning based classifiers towards jamming attacks detection. Several supervised machine learning algorithms including support vector machine, k-nearest neighbor, artificial neural network, and decision tree are used and their performance is compared. Features such as bit error rate, bad packet ratio, and energy statistic are utilized to train these models and to distinguish the received jamming signals from legitimate ones. The results show that the two-hidden-layer neural network with 15 neurons outperforms all the other algorithms.

Compressive sensing is a rapidly growing research area that is being applied in many fields, including mathematics, aerospace, electrical engineering, computer science, and biomedical engineering. This concept proclaims that sparse signals can be recovered from a set of few measurements. It involves two processes, encoding and decoding. The encoding process deals with the question how one should design the linear measurement process and the decoding deals with what algorithm can successfully recover the original sparse signal. To date, both processes are still open problems, particularly how to construct explicit measurement matrices that lead to a successful recovery process. Therefore, in this paper, we review the main properties that can be used to design suitable measurement matrices and propose an explicit construction of measurement matrix based on the Chebyshev polynomial and Vandermonde matrix. The performance of this matrix is evaluated, and its efficiency is compared to those of the existing matrices using several evaluation metrics, namely the recovery error, processing and recovery time, and phase transition diagrams. This matrix is also evaluated in the context of wideband spectrum sensing using metrics such as the probabilities of detection and false alarm. The results show that the proposed matrix outperforms the performances of Gaussian and Toeplitz based models.

Glioblastoma is the most common and primary type of brain cancer. A surgical intervention is the first treatment as it improves the prognosis of the patient. Unfortunately, this type of tumor is aggressive, and difficult to remove completely, which makes the resection surgery more challenging. Magnetic resonance imaging (MRI) is the most used screening technique for brain cancer diagnosis and surgery planning as it provides detailed information about the tumor's location and size. In this paper, we propose a system for the reconstruction of three-dimensional brain models containing a glioblastoma tumors using the Microsoft headset HoloLens. The developed mixed-reality system projects and overlaps the 3D brain model from the MRI images onto the patient's head during the surgery. This system has the potential to simplify the surgery, reduce the surgery high-risk, increase the resection precision, and help remove the tumor tissue as much as possible.

With the rapid proliferation of mobile users, the spectrum scarcity has become one of the issues that have to be addressed. Cognitive Radio technology addresses this problem by allowing an opportunistic use of the spectrum bands. In cognitive radio networks, unlicensed users can use licensed channels without causing harmful interference to licensed users. However, cognitive radio networks can be subject to different security threats which can cause severe performance degradation. One of the main attacks on these networks is the primary user emulation in which a malicious node emulates the characteristics of the primary user signals. In this paper, we propose a detection technique of this attack based on the RSS-based localization with the maximum likelihood estimation. The simulation results show that the proposed technique outperforms the RSS-based localization method in detecting the primary user emulation attacker.

Breast cancer is one of the most prevalent diseases among female populations worldwide. Because the cause and prevention remain unknown, early detection is considered the only way to increase the survival rate. X-ray imaging is currently the most reliable technique for detecting abnormalities in the breast and it is highly recommended for middle-aged women (40- 60) who are at higher risk of developing the disease. However, the efficiency of X-ray is dependent on radiologists’ expertise as image interpretation is challenging and can lead to false positives and false negatives. Image processing algorithms are highly useful for providing reliable diagnosis that can help radiologists make a fast, reliable medical interpretation. Several techniques for isolating tumors from digital mammograms have been proposed, including histogram-based methods, region-based algorithms, and edge detection approaches. However, these techniques are not effective, as they use fixed threshold levels to isolate suspicious areas from the non-uniform image backgrounds. In this paper, we propose an approach based on Invasive Weed optimization (IWO) and Smallest Univalue Segment Assimilating Nucleus (SUSAN). The IWO algorithm determines the optimal threshold for the extraction of the suspicious regions in mammograms. The selected threshold is then used for the detection of dense abnormalities using the SUSAN algorithm. The results show that this technique is more accurate in detecting suspicious areas in breasts, and particularly dense breasts, than the existing techniques.

For civil and military surveillance applications, a remote through-wall radar offers more practical solution compared to the conventional wired methods. For example, it helps emergency and military personnel to quickly detect tiny movements of limbs, breathing, or heartrate in a complex environment such as a collapsed building. This paper describes a low cost through-wall radar for monitoring vital signs. Three test cases along with their results are presented. Testing highlights the challenges of isolating human movement through various materials along with effectiveness of the proposed system in identifying human movements.

Breast cancer is the most common type of cancer among women and it is the major cause of female cancer-related deaths worldwide. Microcalcifications, which are small crystals of calcium apatites, are considered the first sign of breast cancer in more than half of all breast cancer cases. Since these calcium apatites are very small and may be easily overlooked by the radiologists, automatic image processing systems can help radiologists in early diagnosis of breast cancer. In these systems, detection and segmentation of breast microcalcifications from the background tissue are important steps. The purpose of this work is to evaluate the performance of several breast microcalcifications segmentation techniques and select the best technique using a multicriteria decision making approach. The approaches were applied on 630 mammograms from the Digital Database for Screening Mammography. Of the 630 images, 315 images correspond to malignant cases and 315 correspond to benign cases. The efficiency of the considered techniques was evaluated by using five metrics, namely a similarity index, the extra overlap fraction, sensitivity, specificity, and accuracy.

Smart grid is an emerging and promising technology. It uses the power of information technologies to deliver intelligently the electrical power to customers, and it allows the integration of the green technology to meet the environmental requirements. Unfortunately, information technologies have its inherent vulnerabilities and weaknesses that expose the smart grid to a wide variety of security risks. The Intrusion detection system (IDS) plays an important role in securing smart grid networks and detecting malicious activity, yet it suffers from several limitations. Many research papers have been published to address these issues using several algorithms and techniques. Therefore, a detailed comparison between these algorithms is needed. This paper presents an overview of four data mining algorithms used by IDS in Smart Grid. An evaluation of performance of these algorithms is conducted based on several metrics including the probability of detection, probability of false alarm, probability of miss detection, efficiency, and processing time. Results show that Random Forest outperforms the other three algorithms in detecting attacks with higher probability of detection, lower probability of false alarm, lower probability of miss detection, and higher accuracy., Comment: 6 pages, 6 Figures

Unmanned Aerial Systems (UAS) have received a huge interest in military and civil applications. Applications of UAS are dependent on successful communications of these systems with different entities in their networks. A UAS network can include the UAS, ground control station, navigation satellite system, and automatic dependent surveillance-broadcast (ADS-B) receiver. Through these entities, a UAS is vulnerable to different cyber-attacks such as GPS spoofing. In this attack, a malicious user transmits fake signals to the GPS receiver on the UAS. The fake signals can mislead not only the aircraft but also air traffic controllers, leading to serious problems. These problems range from aircraft hijacking to collisions and human casualties. This paper proposes a supervised machine learning method based on the artificial neural network to detect GPS spoofing signals. Different features such as pseudo range, Doppler shift, and signal-to-noise ratio (SNR) are used to perform the classification of GPS signals. We examine and compare the efficiency of one-and two-hidden-layer neural networks with various numbers of hidden neurons. The results show that our proposed method provides a high probability of detection and a low probability of false alarm.

Collision avoidance for unmanned aerial vehicles requires dealing with a high-level of uncertainty. Classifying the existing object before dealing with it allows predicting the object’s position with a reduced uncertainty level. In this paper we propose a 3-D graphical representation for indoor objects’ occupancies based on a probabilistic Bayesian model.

Visible light imagery is one of the most common data products of remote sensing satellites and is of particular importance for Earth-based disaster response and management, national defense, and numerous Earth science studies. However, size, mass, and power constraints impose severe limitations on payload designs, especially imaging systems whose fields of view and resolutions are dictated primarily by their physical dimensions. In this paper, the capabilities of multiple platforms are investigated for use in small satellites. Based on power consumption, processing speed, interface adaptability, and reliability in extreme conditions, three platforms were chosen for further investigation: the BeagleBone Black, the Raspberry Pi 1 B+, and the BeagleBoard-xM. This paper focuses primarily on planned development of the Raspberry Pi 1 into a functioning imaging payload system.

One of the disadvantages of the Automatic Dependent Surveillance-Broadcast (ADS-B) communications protocol is its vulnerability to a number of security attacks since it broadcasts information of the flying aircraft carrying the device over unencrypted datalink. As the U.S. Federal Aviation Administration has mandated the use of ADS-B by 2020, solid, reliable security solutions are needed to secure ADS-B data exchange. The first step in this regard is investigating ADS-B attacks and their impacts on air traffic. Therefore, in this paper, we describe a preliminary work performed to study some effects of an ADS-B attack category, message injection, using Hardware in the Loop testing platform. Using a set of flight plans developed to generate aircraft encounters, results show that this attack can distract pilots and ground controllers, cause air traffic disturbance, and increase the risk of aircraft collisions.

Unmanned Aerial Systems (UAS) are swiftly becoming a large sector for research and development. However, the Federal Aviation Administration (FAA) has set forth many regulations that UAS must comply with before they are able to be integrated into the National Airspace System (NAS). Among these regulations is for all UAS to have an ability analogous to a human pilot's ability to “see-and-avoid.” This system must be capable of detecting both large and small obstacles, such as terrain and intruding aircraft. A Detect-and-Avoid (DAA) algorithm is currently in development at the University of North Dakota (UND) in order to comply with these regulations. However, while this DAA algorithm is capable of detecting and avoiding cooperative targets through an Automatic Dependent Surveillance-Broadcast (ADS-B) system, UAS are also required to detect and avoid uncooperative obstacles. This paper describes a proof-of-concept solution to this problem by integrating a small RADAR into UND's current DAA algorithm.

Compressive sensing has been proposed as an alternative solution for signal acquisition to directly acquire compressible signals at a rate lower than the Nyquist rate. Compressive sensing techniques can be classified into two categories: one-bit compressive sensing and multi-bit compressive sensing. One-bit compressive sensing performs by acquiring only the sign of each measurement while multi-bit compressive sensing performs by acquiring multiple bits. In this paper, we analyze the performance of each category. We compare their efficiencies using metrics that cover the most aspects of the compressive sensing performance in cognitive radio networks. Simulation results show that one-bit compressive sensing can outperform the multi-bit compressive sensing in terms of speed, robustness to noise, complexity, and reconstruction success rate.

Security in cognitive radio networks is considered as an important problem that is attracting a lot of interest from researchers. One of the main security threats is the Primary User Emulation (PUE) attack, which aims to gain illicit access to the licensed channels. A PUE can mimic the same signal as the real primary user (PU) which requires secondary users (SUs) to free immediately the channel. This attack can result in service degradation, deny of service (DoS), and a considerable impact on the cognitive network. To identify the attacker, nodes need to precisely locate the source of the PU and identify the authenticity of the PU. In the literature, most of localization of unknown signal sources are based on ranging schemes, which measure the distance between the blind node and the anchors. These anchors are static with known positions and are located near the signal source for accurate position detection. The challenge is to have limited number of anchor nodes that can detect the location of the signal. In this paper, we propose a technique based on particle swarm optimization algorithm and the received signal strength indicator (RSSI) for the PU/PUE position detection to increase the detection accuracy and decrease the probability of false alarms.

Quantification of uncertainty due to wind-energy production becomes more and more crucial as the penetration of wind into smart grid increases. System operators (TSOs) and planners would be interested to see how wind production varies over different look-ahead hours and estimate the probability of those variations under several uncertain conditions. As wind is a stochastic source of generation, this paper provides a state-of-the-art literature review on the uncertainties related to wind-energy dispatch.

Mobile advertisements are critical for both mobile users and businesses as people spend more time on mobile devices than on PCs. However, how to send relevant advertisements and avoid unnecessary ones to specific mobile users is always a challenge. For example, a concert-goer may like to visit restaurants or parks before the concert and may not like the advertisements of grocery stores or farmers' markets. This research tries to overcome the challenge by using the methods of location-aware mining. Furthermore, privacy is always a great concern for location-based advertising (LBA) users because their location information has to be shared in order to use the LBA. This research also takes the concern into serious consideration, so the user privacy will not be compromised. Preliminary experiment results show the proposed methods are effective and user-privacy is rigorously preserved.

Primary user emulation (PUE) attack is one of the main threats affecting cognitive radio (CR) networks. The PUE can forge the same signal as the real primary user (PU) in order to use the licensed channel and cause deny of service (DoS). Therefore, it is important to locate the position of the PUE in order to stop and avoid any further attack. Several techniques have been proposed for localization, including the received signal strength indication RSSI, Triangulation, and Physical Network Layer Coding. However, the area surrounding the real PU is always affected by uncertainty. This uncertainty can be described as a lost (cost) function and conditional probability to be taken into consideration while proclaiming if a PU/PUE is the real PU or not. In this paper, we proposed a combination of a Bayesian model and trilateration technique. In the first part a trilateration technique is used to have a good approximation of the PUE position making use of the RSSI between the anchor nodes and the PU/PUE. In the second part, a Bayesian decision theory is used to claim the legitimacy of the PU based on the lost function and the conditional probability to help to determine the existence of the PUE attacker in the uncertainty area.

Spectrum sensing is an important process in cognitive radio. A number of sensing techniques that have been proposed suffer from high processing time, hardware cost and computational complexity. To address these problems, compressive sensing has been proposed to decrease the processing time and expedite the scanning process of the radio spectrum. Selection of a suitable sparse recovery algorithm is necessary to achieve this goal. A number of sparse recovery algorithms have been proposed. This paper surveys the sparse recovery algorithms, classify them into categories, and compares their performances. For the comparison, we used several metrics such as recovery error, recovery time, covariance, and phase transition diagram. The results show that techniques under Greedy category are faster, techniques of Convex and Relaxation category perform better in term of recovery error, and Bayesian based techniques are observed to have an advantageous balance of small recovery error and a short recovery time., CCWC 2017 Las Vegas, USA

Estimation of the signal-to-noise ratio (SNR) has become an integral part of wireless communication systems, particularly in cognitive radio systems. The knowledge of the SNR at any time is essential because it has a significant influence on the performance of the system. Approximating this parameter can help better calculate the occupancy level of different channels of the radio spectrum which is an essential part in decision making process of cognitive radio systems. Recently, a novel SNR estimation approach based on the eigenvalues of the covariance matrix of the received samples was proposed in the literature. This method is highly dependent on a number of parameters including number of input samples, number of eigenvalues, and Marchenko-Pastur distribution size. In the process of SNR estimation, these parameters are chosen based on some factors such as available hardware, channel condition, and the application for which SNR is estimated. In this paper, we analyze the effect of each of the mentioned parameters on the SNR estimation method and show that they need to be optimized. We propose the use of particle swarm optimization (PSO) algorithm in the eigenvalue-based SNR estimation technique to optimize these parameters. The results of the proposed method are compared with those of the original SNR estimation method. The results validate the improvement achieved by our technique compared to the original technique.

Cognitive radio is a new technology that aims to solve the scarcity and underutilization of the radio spectrum. It also aims to achieve its goals with a high quality of service. Hence, channel quality estimation metrics are used to help the cognitive radio to readjust its parameters and enhance its quality of service. One of these metrics is the probability of outage. This metric depends on either the level of signal to interference plus noise ratio (SINR), or the channel capacity and data rate. However, uncertainty affects these two variables, which in turns affects the probability of outage. Therefore, a method that deals with uncertainty is necessary. In this paper, we propose a model based on a Bayesian network. This method qualitatively and quantitatively relates the variables affecting SINR and outage probability by a conditional probabilistic model. The results of the proposed Bayesian model show the effectiveness in handling uncertainty.

This paper discusses a model that predicts trap counts of Culex tarsalis, a female mosquito that is responsible for West Nile Virus (WNV) using machine-learning algorithms. Culex mosquitoes are the main transmission vectors for WNV infections. In this research, a Partial Least Square Regression (PLSR) has been deployed to predict mosquito trap counts of Culex tarsalis using historical meteorological and trap count data from 2005–2015. The associations between 10 years of mosquito capture data and the time lagged environmental quantities trap counts, rainfall, temperature, precipitation, and relative humidity were used to generate a predictive model for the population dynamics of this vector species. Statistical measure of Mean Absolute Error (MAE) is compared with other existing actual collected trap counts to analyze accuracy the predictive models. The paper also details the development of a user-friendly web-interface containing interactive web pages that allow users to visualize the North Dakota mosquito population, weather pattern, and WNV incidence data. The interface utilizes multi-layered Google Maps developed through Google Fusion Tables. An understanding of historical data and weather variables is essential for providing sufficient lead time to predict WNV occurrence, and for implementing disease control and prevention strategies such as spray period and hiring of seasonal mosquito workers. Further, an approach similar to the proposed approach of this paper, which involves the integration of data mining and data visualization techniques, brings novelty to vector control initiatives.

With the increase in utilization of portable devices and ever-growing demand for greater data rates in wireless transmission, an increasing demand for spectrum channels was observed since last decade. Currently, radio spectrum channels are assigned for quite long time periods to licensed subscribers who may not constantly use these bands, which leads to an inefficient use of the radio spectrum. The concept of cognitive radio technology was suggested to deal with the problem of spectrum scarcity caused by the underutilization of the radio spectrum. A cognitive radio system is able to sense its functional and geographical surrounding and adjust its operation. Therefore, cognitive radio operation has to be considered with intelligent sensing and smart decision-making methods. The aim of the paper is to examine and evaluate the three most fundamental spectrum sensing techniques i.e. energy detection-based, autocorrelation-based, and matched filter-based sensing. Simulation platforms were developed for each of the approaches using GNU radio and Python language.

Interference is one of the critical factors that affects the performance of cognitive radio networks. In these networks, secondary users are allowed to use the primary user channel with the condition that they cause no interference to it. Interference power received at the primary user is impacted by a number of parameters, including nodes transmit powers, distance between the primary user and the other nodes, path loss, and shadowing. A number of techniques have been proposed to model the interference power. However, these techniques do not consider the uncertainty associated with these parameters. Therefore, a model that deals with the uncertainty affecting the aggregate interference power is needed. In this paper, a Bayesian model is proposed to probabilistically describe how a number of factors affect the aggregate power of interference.

The scarcity of spectrum radio and the immense growth of mobile applications that demand real-time data communications have raised several challenges in the level of medium access control (MAC) design which impose the necessity for a real review and enhancement of MAC protocols. The MAC layer has gained new capabilities with cognitive radio (CR) and opportunistic spectrum access. It is one of the challenging problems in cognitive radio systems. In this paper, we present a survey of some decentralized MAC protocols for CR networks with analysis and comparative assessment of CR-MAC protocols using some Key Performance Indicators.