Al-Iraqia Journal for Scientific Engineering Research

Over the past fifteen years, phishing has emerged as the leading cybercriminal activity, resulting in the unauthorized acquisition of substantial financial resources amounting to billions of dollars. This phenomenon arises due to using novel (zero-day) and complicated tactics by phishing attackers to deceive internet users. Email is the primary approach utilized to initiate phishing attacks. This study comprehensively analyzes popular methods used in email spam tests. The present analysis comprehensively examines the key concepts, techniques, and research trends relative to spam filtering. The topic of discussion involved a general email spam filtering mechanism and the attempts of various scholars to counter spam by employing machine-learning methodologies. Our review examines the advantages and disadvantages of several machine learning methods within the context of spam filtering while addressing some of the biggest research inquiries in this domain

Recent advancements in machine learning have played a crucial role in the healthcare industry, particularly in predicting heart disease with assorted datasets. Despite the registration results indicating promising accuracy and resilience in heart disease prediction models, it is still low compared to what we expected. In this article, a new technique has been demonstrated to predict heart diseasebased on state-of-the-art machine learning with a focus mostly on Hyper Adaboost classifiers. Weevaluate the performance of several key machine learning algorithms, including Random Forest, Extra Trees, LightGBM, Decision Tree, and Hyper Adaboost, on well-known heart disease datasets. Ourresults are mixed, but they suggest that the Hyper Adaboost classifier provides stronger accuracy, with performance metrics consistently exceeding 97%, much better than similar models. This research highlights the promise of newer packages under Hyper for detecting and screening individuals with heart disease early, which can offerinsights into advancing methodologies in future studies as well as clinical use. Our technique exhibited strong predictive power, encouraging additional study with a larger and more comprehensive dataset for the validation of our mode

The use of recycled concrete aggregate (RAC) in creating new concrete has gained significant attention for environmental and financial reasons. However, the compressive strength of the product concrete is hard to predict due to many variables. In this study, the compressive strength of recycled aggregate concrete was predicted using eight well-known machine learning algorithms, including support vector machine (SVM), artificial neural network (ANN), XGboost, Tree, Random Forest, Gradient Boosting, CatBoost, and AdaBoost. Every machine learning algorithm's general methodology entails gathering and analyzing input data, training the algorithm, testing the algorithm, and producing an output. A total of 419 data samples (experimental tests) were used in training and testing all the machine learning models. The results show that the best models for estimating RAC compressive strength are Neural Network, AdaBoost, and XGBoost. The other algorithms, random forest, gradient boosting, and Catboost, performed well in predicting the compressive strength of RAC, however, tree decision and SVM performed badly. The primary evaluation metrics used in this study were Mean Squared Error (MSE) and R-squared (R²), which helped determine the accuracy and reliability of the predictive models

The significance of human activity recognition (HAR) is rising as it seeks to improve everyday life and healthcare through better technology access and efficiency. Its objective is to transform industries by enabling smart homes, improving robots, bolstering security, and improving human-computer interactions. HAR works to improve well-being, which is essential to health, wellness, and sports.

While the complexity of human behavior poses challenges, machine learning advancements offer hope for solutions. Continuous research in accurately detecting a wide array of human activities underscores the significant impact of HAR on technological development and its broad applications.

In this work, a convolution neural network CNN algorithm and random forest RF algorithms were produced for human recognition activity classification using WISDM-51 dataset that contains 18 human activities. The CNN achieved an accuracy of 89.36%, whereas the RF algorithm reached a slightly higher accuracy of 93.46%. The results suggest that the proposed algorithms offer promising potential

Cloud computing is one of the major IT and networking innovations that started in 2006; since then, a significant part of tech companies and users have adopted the cloud computing approach as it provides efficient networking and on-demand services in addition to the high scalability and flexibility that the cloud computing can offer to its users. However, the increasing demand for cloud computing has resulted in a high energy consumption rate and higher CO2 emissions. This paper will analyze and explore the adverse effects of cloud computing on the environment and climate change based on the increased energy needed to operate cloud services around the globe. The paper will also analyze the possible strategies and techniques as alternative solutions to minimize energy consumption and CO2 footprints associated with cloud services to maintain environmental suitability through the green cloud computing approach. A survey was conducted targeting 65 individuals of different backgrounds to assess and measure public awareness regarding the environmental impact of cloud computing; the survey findings showed a need for building communities and public awareness in this regard, as a considerable number of responders were not aware of the impact and relation of cloud computing on the increased energy consumptions and carbon emissions

Efficient signal identification in wireless communication systems is critical for optimal service provision. However, the complexity of contemporary criteria and factors such as noise and fading make it hard to do so. To address this problem, convolutional neural networks (CNNs) are used to classify signals using 3G, LTE, and 5G standards. This approach involves creating a range of datasets with different Signal-to-Noise Ratios (SNR) and introducing Rayleigh fading to represent real-world environments. Two CNN architectures for dependable assessment, VGG19 and ResNet18, with robust 5-fold cross-validation, are employed. To test model resilience, the dataset includes Poisson noise and Thermal noise. Despite noise and fading in the system, VGG19 and ResNet18 show high accuracies across all standards. However, ResNet18 demonstrates relatively better performance, especially under Poisson noise conditions. Both models also have good signal detection from among noises generated by Poisson thermal or Rayleigh distribution. ResNet18 demonstrates a commendable average accuracy of 99.52%, while VGG19 Net demonstrates 97.14%. CNNs effectively identify signals amidst noise scenarios and contribute to advancing deep learning techniques in signal processing, enhancing the reliability of wireless communication systems

he importance of security and surveillance systems in today\'s society cannot be overstated. The monitoring system can send out early alerts in an emergency. This study investigates an IoT surveillance system that can detect movement, record video footage by the camera module in real-time, and count the recorded video footage. The video footage is sent to the user via Mail. In addition to sending notification messages, \"WARNING! Motion is detected, and the video is sent to Gmail for playback\" immediately tothe mobile phone using the Telegram application via Wi-Fi (Wireless Fidelity). It is worth mentioning that the microcontroller for this system is the Raspberry Pi 3 Model B. Additionally, the Passive Infrared (PIR) sensor detects any movement by measuringthe changes in Infrared (IR) radiation. The Light Dependent Resistor (LDR) sensor is also used to check whether daylight or darkness exists. Furthermore, if the movement is detected, the Light Emitting Diode (LEDs) and the Lightbulb will be lit up/down depending on the state of the Light Dependent Resistor (LDR) sensor. The buzzer will activate to provide the emergency sound, and a power bank will be used as a power source to avoid interruptions in electrical power for the surveillance system

This study evaluates the thermal performance and energy cost implications of integrating Phase Change Materials (PCM) into building walls with brick and concrete core materials, under Baghdad, Iraq's climatic conditions. A numerical simulation was conducted using MATLAB to analyze the heat transfer and temperature distribution across wall configurations, with a focus on PCM thickness ranging from 0.01 m to 0.1 m. The simulation assumed a constant indoor temperature and real outdoor boundary conditions over one year, accounting for hourly variations in heat flux, wall temperatures, and energy costs. Results demonstrated that incorporating PCM significantly reduced heat flux and stabilized the wall's internal temperature, leading to lower energy costs for heating and cooling. For brick walls, the optimal PCM thickness was determined to be 0.07 m, yielding an annual energy cost of 150 USD. For concrete walls, the optimal PCM thickness was found to be 0.05 m, with an energy cost of 120 USD. The concrete wall exhibited superior thermal performance and cost efficiency compared to the brick wall due to its higher thermal conductivity and density, which enhanced heat storage and transfer. These findings highlight the effectiveness of PCM-enhanced walls in improving energy efficiency and thermal comfort, particularly in climates with high-temperature variability. This work provides insights for optimizing wall configurations and selecting PCM thickness to achieve sustainable building designs and cost-effective energy management.

The use of unmanned aerial vehicles (UAVs) for aerial photography has become increasingly popular in various fields, including engineering, urban planning, environmental impact studies, and monitoring of transportation lines, power supplies, and other military applications. One of the main challenges in using drone imagery to model urban buildings is achieving high accuracy in generating 3D models. This paper focuses on improving the accuracy of urban building modelling using drone imagery, structure from motion, and multi-view stereo techniques used in computer vision, augmented virtual reality, geo-science, photography, and aerial photography to create 3D models from pairs of images. The study uses a UAV with a DJI Mavic 2 Pro drone 4K sophisticated camera for up to 31 minutes of flight time. Data collection time at different angles and specific heights while incorporating ground control points to enhance the accuracy of the generated point clouds. A specific location inside Al-Nahrain University was chosen as a fieldwork model, where the images captured by the drone were processed using Agisoft Metashape Pro software to create detailed 3D models for the buildings depending on the Structure form Motion (SfM) technique. The results demonstrate the efficient integration of Ground Control Points (GCP) combined with advanced processing techniques that enhance the model accuracy, achieving highly accurate GCP demonstrating error margins as low as 0.1% based on drone-derived data for urban buildings.

The development in optical communications has made the possibility of communication by lighting rooms possible through Li-Fi technology, which is a technology for transmitting data via light. Maintaining the performance of the network is one of its most important components in the shadow of external light, whether sunlight, industrial light, or any other light source, and this effects the efficiency of the system, data rate, and BER. In this research, the effect of a white light will be discussed on the Li-Fi system, using multiple modulation patterns and getting the good performance of the system design in Optisystem and Matlab for three types of modulation and power distribution.

Free Space Optics (FSO) is a potential optical technique poised to complement traditional wireless communication. Using optical signals for point-to-point transmission, FSO offers advantages such as abundant unregulated bandwidth, high data rates, enhanced security, and lower costs than microwave links. However, attenuation due to weather conditions, particularly rain, poses a significant challenge to both FSO and RF transmission. This study proposes a Hybrid FSO/58GHz system model to address rain-induced attenuation, implementing the ITU-R FSO prediction model in Kuala Lumpur to evaluate performance in Malaysia. Utilising three years of rain intensity data from UTM, Kuala Lumpur, the study computes rain attenuation for an FSO link with a 780 nm wavelength based on the ITU-R P.1814 model. Specific attenuations for various RF frequencies are also estimated to identify a suitable frequency for the hybrid system. The study analyses the impacts of channel attenuations on both FSO and RF links to enhance system availability and data rate. Key performance metrics such as received power (Pr), signal-to-noise ratio (SNR), and bit error rate (BER) are assessed for a 1 km link. The results demonstrate the successful implementation of the FSO link based on the ITU-R model in Malaysia, identifying 58GHz as the optimal RF frequency for the hybrid system and improving performance under heavy rain conditions.

A numerical investigation has been conducted on heat transport in a domain when an equilateral triangular cylinder of (15 mm) is present. A flow analysis is conducted using a Reynolds number (Re) ranging from 100 to 500 on a cylinder encircling a laminar regime. The flowing medium, air, is thought to have a constant Prandtl number. A two-dimensional method is used to simplify the problem. The governing equations are solved using momentum, energy, and continuity equations. Studying the effect of changing the position of the cylinder within the computational domain on several parameters such as temperature distribution, pressure, velocity, streamline, surface temperatures, Nusselt number (Nu), friction factor, and the drag and lift coefficient (CD and CL). The results of the current study showed that the location of the cylinder affects the characteristics of airflow, where the first location gave the best improvement of heat transfer compared to other locations; also, the surface temperature gradually decreases by increasing the Reynolds number and increases by changing the location towards the axis of fluid flow and the friction factor decreased significantly by increasing the air velocity around the cylinder. The percentage difference of the temperature distribution rate of the surface for a computational domain of the cylinder loc

Cogging torque disadvantages permanent magnet DC motors (PMDC), creating vibration and audible noises. This research investigates the factors affecting the cogging torque of a permanent magnet-brushed DC motor used to operate automobile windshield wipers by implementing some design changes by using the finite element analysis (FEA) based on Maxwell 2D software. These factors include changing the air gap length, slot opening, embrace factor, magnet edge inset (pole arc offset), skewing angle, and magnet type to study the amount of reduction and increase in cogging torque and its effect on the motor's efficiency. We conclude from the results of the FEM simulation that when changing the parameters of the factors, we will notice a clear effect on cogging torque and motor efficiency.

This paper explores Light Fidelity (Li-Fi) technology, which uses light-emitting diodes (LEDs) to send data, providing an alternative to traditional WiFi that uses radio signals. Li-Fi takes advantage of the unused 300 Terahertz optical spectrum, helping to reduce overcrowding in radio frequency spectrums used by other technologies. The paper discusses the implementation of Li-Fi, compares it with WiFi, and describes some challenges and limitations. The goal is to provide clear insights into how Li-Fi performs, the applications, and the potential issues, showing how it could pave the way for wireless communication for various applications such as vehicles, hospitals, and smart lighting systems.

Face recognition technology is a biometric method used to identify or verify individuals by their facial characteristics. It is widely used in commercial and law enforcement applications, such as surveillance systems, passport verification, security systems, and human-machine interaction. However, face recognition challenges such as noise, image deterioration, corruption, and external elements. To improve face recognition accuracy in noisy environments, noise reduction techniques, and robust representation learning methods are needed. This study attempts to clarify on Active Appearance Models, Viola-Jones, and Convolutional Neural Networks algorithms which have been used in literature studies, as well, provide the issues by organizing the abundance of articles and information in this field to highlight current research trends, and provides an outline of their advantages and disadvantages.

Due to the rapid progression of Human-Robot Collaboration (HRC), ensuring safe interactions between humans and robots, contact detecting systems must be dependable and efficient. In this research, various models are tested using a contact detection dataset that includes non-contact motions, intentional interactions, and accidental collisions among others. K-Nearest Neighbors (KNN), Bagging, and Long Short-Term Memory (LSTM) networks are evaluated on their ability to classify different types of contacts. According to the findings of the experiment, it is clear that KNN and Bagging are reasonably accurate, but LSTM has surpassed both by achieving higher accuracy levels besides being better at handling temporal dependencies which are inherent in sensor data collected from dynamic human-robot interactions. The results have shown that when it comes to such kind of contact detection datasets, long short-term memory (LSTM) and other deep learning models are superior to other methods. These results show that HRC systems can be made safer and more effective by using more sophisticated neural networks. This research helps connect theory with practice by providing a foundation for the creation of collaborative robots that are not only intelligent but also safe.

This research simulates the temperature dynamics of the phase change material PCM using the new approach to calculate heat capacity, supported by advanced, efficient explicit numerical methods. The study examines many scenarios for building wall geometries and boundary conditions by controlling the heat loads and ensuring comfortable interior temperatures. Paraffin wax, selected for its distinct melting temperatures and latent heat capacities, is used as the PCM. The study consistently demonstrates the effectiveness of PCMs in decreasing the heat transfer indoors of the building, regardless of the wall material. This research helps to understand the PCMs' behaviour using the Effective Heat Capacity model, offering valuable insights for energy-efficient building design and highlighting the critical role of selecting suitable PCMs in construction.

The rapid advancement of face image manipulation (FIM) algorithms and the proliferation of their user-friendly applications underscore an urgent need for manipulation detection techniques. These methods should be capable of revealing modifications in face images and substantiating their authenticity. Recently, the term "DeepFakes" and their detection techniques have attracted the attention of the research community. In addition, pay attention to the most recent techniques for detecting facial image manipulation that utilize watermarks. It's crucial to note that each of these techniques comes with its own set of limitations. This research aims to critically evaluate recent developments in face image authentication (FIA) methods, considering the widespread and user-friendly applications of FIM algorithms and their rapid growth. This study focuses on the urgent necessity for effective manipulation detection methods that can reliably identify and verify modifications in facial images, particularly with the rise of sophisticated DeepFakes. It explores two primary detection approaches: deep learning (DL) techniques, which leverage large datasets to detect subtle manipulations, and watermarking-based methods, which embed verification data into images to safeguard authenticity. The findings showcase the positive aspects and the limitations of these methods. DL techniques are powerful in detecting complex alterations but require substantial computational resources and data for training.

Conversely, watermarking offers a proactive solution for verifying image integrity but may be vulnerable to advanced manipulation tactics and can impact image quality. The review emphasizes the importance of ongoing innovation, advocating for hybrid approaches that integrate the benefits of both DL and watermarking to overcome their shortcomings. This paper serves as a crucial reference for researchers, presenting a detailed overview of current trends, challenges, and future directions in face image manipulation detection (FIMD), underscoring the need for continuous development to keep pace with advancing manipulation technologies.

In recent years, Microgrid has become more attractive. A combination of renewable energy resources with load at the distribution level appears preferable. However, such a combination is challenging to work correctly. Power system indices such as voltage and frequency appear difficult to be within an allowable limit. This research will use a STATCOM compensator to increase the voltage stability of the power system containing wind and solar energy generation systems. The STATCOM based modular multi-level converter is designed to increase the output voltage quality and reduce filter requirements. The proportional and integral (PI) and integral sliding mode (ISM) controller is designed in a MATLAB/SIMULINK environment. The results demonstrate the effectiveness of the STATCOM compensator in this hybrid system with both PI and ISM controllers.

This study aims to determine the location and track of sensor nodes in indoor spaces. The challenge of significant estimation errors in target position brought on by erratic noise in received signal strength indicator (RSSI) readings is a major area of current research focus, especially in interior conditions. In place of the traditional RSSI-based approach, this study suggested a hybrid technology called Nomadic People Optimizer-Generalized Regression Neural Network (NPO-GRNN) to increase the sensor nodes\\\' capacity to estimate location and target tracking with more accuracy. The RSSI values can be used by the GRNN method as start data to determine and trace the target node\\\'s location. The spread constant (σ) is a crucial part of the GRNN architecture. To choose the spread constant (σ), an insecure and sometimes unreliable method by trial and error is employed. The ideal GRNN spread constant is found using the NPO approach. To get around these problems and improve L & T tracking precision without the need for additional equipment, the hybrid NPO-GRNN method was employed, and these coordinates were refined using a Kalman filter to increase accuracy. Impressive results were obtained by the tracking algorithm NPO-GRNN-UKF hybrid, which performed better than the traditional LNSM approach. By comparing the suggested approach to the traditional RSSI, a significant 98.4% and 98.1% for targets 1 & 2, respectively, gain can be achieved.