Bilad Alrafidain Journal for Engineering Science and Technology (BAJEST)

Images play a fundamental role in digital media, and altering digital images can present a significant risk since it contributes to disseminating false information. The rapid advancement of technology in digital image forensics has significantly improved the quality of forged images to the extent that many forgeries are now indistinguishable. Digital image authenticity and reliability are becoming more significant as evidence. Some people invalidate photos by adding or removing sections. Therefore, image forgery detection and localization are crucial. Image manipulation techniques have made this a major computer vision issue. Images can be obtained from many origins and may appear in multiple formats. Consequently, passive techniques for detecting image forgeries are generally favored, which do not necessitate prior knowledge about an image. The prevalent forms of passive image forgery detection encompass the identification of copy-move and image-splicing forgeries. Recently, deep learning techniques have become prevalent in image manipulation detection. These techniques demonstrated superior accuracy to traditional approaches due to their ability to extract features from images effectively. This study comprehensively examines deep learning methodologies utilized in detecting copy-move forgery, and it has mostly focused on studies conducted in recent years, and data sets commonly used in detecting copy-move forgery have been mentioned. This abbreviated survey concluded that combining conventional image processing methods with pre-trained CNN approaches could leverage the strength of both, exhibit significant efficiency, and decrease the requirement for labeled image datasets. Furthermore, utilizing ensemble techniques to integrate multiple approaches improves overall forgery detection performance.

Low-quality fingerprint identification has played a significant role in the field of biometric identification. Especially for those who are experts in the use of sophisticated biometric technology in jail security and criminal identification in recent years. Many factors affect the accuracy of low-quality fingerprint identification, such as the quality of the fingerprint image, the accusation device, the extraction tools, etc. Many approaches have been proposed in this field to improve low-quality fingerprint identification. In this paper, we discuss the main factors of the low-quality fingerprint, the main approach of the human biometric data, as well as the main factors of the low-quality of the fingerprint data. Also, we list the state of the art, most recent machine learning and deep learning approaches that have been used recently in this field. As the end, this field of research is still in progress to improve the quality of the identification progress

Sleep is the cornerstone of overall health, and the process of sleep staging involves classifying sleep data into specific stages. Key signals such as EEG, EOG, and EMG are useful in analysing and categorizing sleep data, but it is a complex and time-consuming task. This paper focuses on designing a hybrid deep learning model to accurately classify sleep data using the Sleep Heart Study (SHHS) dataset. Considering that sleep signals show similar temporal patterns to the time series data, we also use transform encoders to extract essential features and facilitate the discrimination of sleep stages. By leveraging the power of transform encoders to capture crucial temporal patterns, we have successfully enhanced the classification of sleep data into five stages. Through comprehensive evaluation using various criteria, we measure the performance of our model and compare it with cutting-edge methods. The results show a significant accuracy of 0.883 and 0.836 for accuracy and Cohen's kappa, respectively, confirming the effectiveness of our approach. The results also highlight the robustness and efficiency of our approach in accurately diagnosing sleep states, ultimately contributing to the advancement of sleep analysis and overall health monitoring

Recent decades have seen significant progress in the growing intersection between artificial intelligence and the way we understand and manage emotions, commonly referred to as emotional intelligence. Although understanding and predicting emotions is one of the greatest human abilities, it has now become an important research area in computer science and information technology. Emotion recognition is a fundamental task in affective computing and a key development area for improving human computer interaction. Early efforts to perform emotion recognition relied primarily on traditional techniques of signal analysis and were unable to provide definitive results. The limitations of conventional signal analysis methods in emotion recognition, the requirement for effective and scalable deep learning models, and the incorporation of these models into practical applications are the main issues this study attempts to address. Deep learning models have become vital due to the breakthroughs in artificial intelligence over the past decade. To this end, in this study, we trained two models, Vision Transformer (ViT) and ResNet50V2, to recognize emotions from image datasets. Among these models, ViT stood out with an impressive accuracy of 78% and an AUC of 0.86 for emotion detection and identification on the FER2013 dataset, demonstrating its strength in understanding facial expressions. This work establishes to develop scalable, efficient, and adaptable models that integrate seamlessly into real-world applications while continuing to push the boundaries of human-computer interaction. This work enhances the ongoing advancements in efficient computing by showcasing the effectiveness of Vision Transformer in emotion detection and offering insights into its use in various interactive systems. In the future, researchers can investigate sophisticated data augmentation approaches. Including other modalities, such as speech, physiological cues, and facial expressions, aids in capturing emotions in greater detail.

With the exponential growth of the Internet of Things, efficient data aggregation is necessary to tackle the issues of network congestion, energy consumption, and latency. This paper presents an adaptive data aggregation algorithm that applies clustering, in-network processing, and optimized routing. The proposed algorithm outperforms traditional methods such as LEACH and PEGASIS with 50 ms latency, 0.5 J/node energy efficiency, and 350 kbps throughput. Key features include dynamic clustering, energy optimization, and real-time adaptation, which offer high scalability and are fit for use in smart cities, healthcare, and industrial IoT. However, future challenges like interoperability and security will still be taken into account on the ground. This paper advances scalable, efficient, and secure IoT systems towards technological advancement.

Mobile Instant Messaging (IM) applications are now considered indispensable in the global communication process; however, they are still exposed to security and privacy threats because of the use of centralized authentication methods. In traditional systems, the user credentials and other important necessary details are stored at the centralized servers which makes it easy for hackers to hack, unauthorized people to gain access, and violate data. The problem of the privacy and security threats in the traditional IM authentication system is solved in this research by developing a new Decentralized IM Authentication model based on the blockchain and SSI technologies and utilizing DIDs to manage data sovereignty. This work aims to propose a decentralized authentication solution that will not require external identification or a central server for the users’ data storage while providing storage immutability and secure end-to-end encryption for all communications. Based on the above analysis, the following methodology is proposed: blockchain for identity authentication, SSI for user-controlled identity, and DIDs for identification. The system was tested using an example IM application called Secure Chat, wherein several security threats, including intrusion, eavesdropping, and identity theft, were emulated. That is why the results obtained state that the decentralized model effectively addresses these threats: the time for the authentication process is 1-2 sec, and the system can support up to 100,000 concurrent users, having a moderate decline in performance. In contrast, decentralized authentication showed a 30% improvement in security and a 40% decrease in the risk of civil identity fraud more to the centralized models. Furthermore, the decentralized approach is fully within privacy regulations such as GDPR, which makes its data protection more effective and in control of its users. The paper makes a significant research contribution by offering an extensive assessment of the proposed decentralized IM application authentication model and revealing its strengths and weaknesses compared to the existing centralized solutions in terms of security, privacy, and sovereignty of data, as well as indicating some emerging difficulties connected with scalability and usability of the approach.

The machine foundation sometimes carries the vibration load developed from the machine run and the seismic load simultaneously. This issue is not covered enough in previous research, especially in the case of laboratory models. Therefore, this paper presents a laboratory model for the response of singly and doubly skirted foundations subjected to seismic loads in sandy soil. The soil was prepared to attain a relative density of 70% in a steel container with a dimension of (50*50*55) cm using the rain technique. Tests were conducted under the dynamic response with a frequency of machine of 30 Hz. A shake table is used; it is exposed to the Halabja earthquake with an intensity of (0.1 g). In addition, three directions of the machine concerning the axis of an applied load (0ᵒ,45ᵒ,90ᵒ) are selected. The ratio of the skirt depth to the machine foundation width (Ds/B) (single and double skirt) was (0,0.25,0.5). The results showed that high displacement can be obtained when the direction of rotation of the machine’s operation is parallel to the direction of the earthquake, also reduced when the angle is 90ᵒ compared with 00. The best case for reducing lateral and vertical displacement is a double skirt at (Ds/B) =0.5 reached (15% and 13%) comparing to single and without skirt respectively

Buildings nearby with different mass and heights, making it dangerous, cause the pounding to happen when two structures collide as a result of lateral movements caused by external forces. The results of an experimental and numerical analysis of pounding between two adjacent buildings are presented in this paper. To investigate two neighboring steel models, five-story and eight-story models, with different gap distances (zero-2-3) centimeters between the two steel models under ground motion during the El-Centro earthquake. To investigate how the buildings respond to pounding in experimental work using a shaking table and numerical analysis by using ABAQUS's finite element modeling software. The experimental study's findings suggest that pounding may affect a structure's response to an earthquake, especially at zero-gap. Display the results of the maximum displacement and acceleration for various gap distance reductions in value at (2-3) cm there’s no pounding. The difference in displacement and acceleration outcomes between theory and experiment El-Centro earthquake at zero-gap, that’s displacement, and acceleration at five-story for the five-story model is about (1%, 14%), and for the eight-story model is about (3%, 4%).

This paper discusses recent advancements in theft surveillance techniques, including tracking multiple objects and estimating human movements. These technologies help with real-time monitoring, understanding behaviors, and ensuring public safety. The major computational challenges of the techniques have been looked at from different angles, including balancing accuracy with efficiency, occlusion management, and scalability in different environments. It also identifies areas that require further research and proposes potential future directions, such as incorporating diverse data types and developing compact and adaptable models. This work is meant to be a helpful tool for researchers and practitioners, giving them a look at the present status of these technologies and steering the creation of new smart systems

Bridge design and structural evaluation should consider thermal load, particularly for statically indeterminate and cable-supported bridges. However, there is limited study and test data on the thermal impact of pre-stress concrete girder, focusing on models developed through numerical analysis and field measurements. Temperature distributions on concrete bridges are nonlinear, causing stress distributions to self-equilibrate, to discuss the stresses, deflections, and moments caused by temperature variations. Three-dimensional finite element ANSYA R3 V19 finite element software is used to analyze pre-stress concrete box girder bridge specimens with trapezoidal cross sections. A parametric analysis was conducted to examine the effect of thermal loading on the behavior of prestressed concrete box girders. Six cases were considered to study the thermal behavior of the box girder. In each of these six cases, thermal loading was applied to three different regions of the box girder. The results showed that girder 3 and girder 6 with bottom thermal had the lowest deflected shape value of 0.84 mm when the thermal load was applied to the bottom flange of the prestressed concrete box girder, with reduced stresses observed in girder 2 and girder3 with top thermal, where the longitudinal stresses in the Z direction were 6.3 ????????????.

Although bridges are designed to last for a hundred years, they are usually used for much longer periods of time. Thermal effects occur on a daily, seasonal, or annual basis, depending on the environmental conditions. However, there are limited studies and test data on the thermal impact of concrete box girders, focusing on models developed through numerical analysis and field measurements. Temperature distributions on concrete bridges are nonlinear, causing stress distributions to self-equilibrate; to discuss the stresses, deflections, and moments caused by temperature variations, the three-dimensional finite element ANSYA program is used to analyze concrete box girder bridge specimens with trapezoidal cross sections. In this paper, three cases of concrete box girders are studied. A parametric analysis is carried out to investigate the effect of thermal loading on the behavior of concrete girders. Three instances of thermal load are studied: the thermal load applied at the upper, bottom, and upper bottom flange of the girder. The numerical results are compared with experimental results, and thermal loading is applied to three box girders in different regions. The numerical results showed that the minimum stress in the longitudinal direction (Z) is 1.28 MPa when the thermal load is applied to both the upper and bottom flanges of the box girder.

Crude oil, often referred to as "black gold," is an essential component of the global economy due to its significant contribution to the world's energy requirements. Numerous countries, regardless of their development status, rely significantly on its importation for transportation, heating, power generation, industrial production, agricultural output, and other applications. Energy security for all stakeholders, particularly oil-importing nations, will invariably rely on the sustained flow and provision of crude oil at competitive prices. The prices of crude oil substantially influence governmental decision-making. Consequently, variations in oil prices can incite social unrest and economic turmoil. Consequently, there is an increasing demand for methodologies that can accurately and efficiently forecast the future behavior of oil prices, considering their significant influence on the economy. To accurately predict oil prices, this study will make use of machine learning techniques such as Support Vector Machine (SVM), K-Nearest Neighbors (K-NN), Random Forest (RF), and Decision Tree (DT). Furthermore. long short-term memory network (LSTM) method as deep learning model is also used. The incapacity of recurrent neural networks (RNNs) to recognize long-range dependencies between distant data instances was a major drawback that LSTM was created to overcome. Despite its simplicity, the LSTM framework has remarkable performance across a wide range of applications. The LSTM models achieve a 99.99% prediction accuracy, outperforming other methods.

The Domain Name System (DNS) is a fundamental component of the Internet's infrastructure and has become a frequent target of major cyberattacks, such as DNS Spoofing, DNS Tunneling, and DNS Amplification Attacks. Among these, DNS Amplification Attacks are the most dangerous, as they exploit misconfigured DNS servers to amplify traffic and overwhelm the target with massive amounts of data. These attacks are particularly challenging for conventional detection techniques to analyze and mitigate. This research proposes an enhanced real-time DNS threat detection model based on the Random Forest algorithm. By utilizing attributes such as query type, packet size, and response time, the model achieves a 98% accuracy rate in distinguishing between normal and anomalous traffic. Additionally, false positives are reduced to 5%, and the response time is improved by 120 milliseconds compared to previously implemented solutions. The success of these network classification models consistently demonstrates the effectiveness of ensemble methods, particularly in addressing DNS threats. Future work will focus on advancing detection systems by developing hybrid models and incorporating signal processing techniques that leverage real-time analysis. This approach aims to ensure that newly emerging cyber threats are effectively identified and mitigated.