بحوث سكوبس — علي كاظم خضير

Structural health monitoring (SHM) is crucial for ensuring the safety and longevity of infrastructure. Traditional SHM methods rely on periodic inspections and centralized data processing, leading to delays and inefficiencies. This chapter proposes an integrated framework combining the Internet of Things (IoT), finite element analysis (FEA), and edge machine learning (ML) to enable real-time, data-driven decision-making for SHM. IoT-based sensor networks continuously collect vibration, strain, and environmental data, which are processed using FEA models for predictive analysis. Edge ML algorithms optimize operational efficiency by executing anomaly detection routines for predictive maintenance alongside real-time applications at local edge nodes to minimize processing delays and decrease cloud dependency. Recorded simulation runs and case-based validation show that the proposed method achieves better structural anomaly detection precision along with faster response times. This framework uses IoT FEA and edge ML technology to create a scalable solution for continuous SHM that enhances both safety and maintenance scheduling for critical infrastructure while maintaining cost-effectiveness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.

A brain tumour is the result of the growth of abnormal brain cells, some of which have the potential to become malignant. Timely and accurate diagnosis of illness and implementation of treatment regimens improves patients’ quality of life and improves their life expectancy. Radiologists’ manual review of medical images is the conventional approach to diagnosing brain tumours, but it is laborious and error-prone. The article proposes a methodical approach to detect brain tumours. The suggested model is a near-ideal synthesis of nature-inspired and quantum-based algorithms, incorporating their more optimistic features. Using the quantum-based binary bat algorithm (q-BBA), the suggested model has been able to reduce dimensionality, or extraneous features. Machine learning classifiers such as SVM, Random Forest, Gaussian Naive Bayes and XGBoost were used to compute the optimality of features. After comparing QBBA’s performance with that of its conventional algorithms, it was found that, when applied to the same population, QBBA achieved better results. Having improved noise immunity and an average accuracy of 98.89%, QBBA emerges as a significant algorithm. Brain Tumour detection can be potential application of the suggested Quantum-based Binary Bat algorithm. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.

The continuous developments in vehicular communication technology have brought a significant interest in Vehicular Ad-Hoc Networks (VANETs). VANETs aim to enhance road safety, improve traffic management, and provide a suite of infotainment services to passengers. This type of network is characterized by high-speed, dynamically varying mobility, leading to increased Energy Consumption (EC), End-to-End (E2E) delay, and Routing Overhead (RO) during network communication. Various researchers have developed ways to overcome this drawback through the employment of clustering techniques in VANETs. However, utilizing clustering techniques in VANETs is critical as it requires maintaining robust communication links, optimizing resource allocation, and minimizing E2E delay. Subsequently, this paper proposes an improved Fuzzy-based Cluster Head Selection (FCHS) technique to enhance the overall performance of VANET. In VANET, the clustering is formed from Cluster Head (CH), Cluster Child (CC), and Backup-Cluster Head (BCH) along with the other network nodes. The FCHS optimizes the CH selection using a fuzzy logic algorithm based on various VANET parameters, including average distance, satisfaction degree, EC, Packet Delivery Ratio (PDR), and vehicle connectivity level. The performance of the proposed FCHS technique is simulated utilizing Network Simulator (NS) 2.35 with the Simulation of Urban MObility (SUMO) platform. The performance metrics that are considered for the result evaluation are PDR, EC, E2E delay, and RO. The overall results of the VANET is compared with two recent methods. The results show that the VANET performance with the aid of the proposed FCHS technique achieves the highest PDR, low EC, E2E delay, and RO. © 2025, Penerbit UTHM. All rights reserved.

Power systems have become more efficient, reliable, and sustainable as smart grids are increasingly integrated into them. Developing innovative security solutions is often required to counter evolving cyber threats because traditional security mechanisms lack real-time protection. The purpose of this paper is to propose a blockchain-based hybrid optimization approach for enhancing smart grid security and resilience. To improve decision-making, resource allocation, and attack mitigation, the framework incorporates artificial intelligence (AI) to detect anomalies in real-time, blockchain technology to store unrestricted data, and a hybrid optimization algorithm to make decisions in real-time. Adaptive Vulture Optimization Algorithm (AVOA) and Convolutional Neural Networks (CNN) combine to reduce computational overhead and maintain detection accuracy effectively. According to the proposed approach, which is compared to existing models, including ML-ID, HHT, and THD, it achieves superior performance, with a detection rate of 99.17%, a reduction in computation time, and enhanced scalability. These results demonstrate that AI-Blockchain hybrid frameworks are effective in protecting smart grids against emerging cyber threats, making them a reliable and scalable security solution. © 2025, Taru Publications. All rights reserved.

Cognitive radio sensor networks (CRSNs) are characterized by their ability to adapt to dynamic spectrum availability. Computation power in cognitive radio network (CRN) is essential for efficient spectrum utilization and seamless connectivity among nodes in a sensor-based health monitoring system. The proposed algorithm, Cognitive Adaptive Metaheuristic Optimization Algorithm (CAMOA) is a Hybrid Particle Swarm-Tabu Search Optimization (HPSTSO) technique that integrates two optimization methods: Particle Swarm Optimization (PSO) and Tabu Search (TS). It dynamically selects the best resource allocation algorithm based on the healthcare network requirements of quality of service and channel conditions, traffic load, and network topology to enhance the performance of communication, and contribute to optimal allocation, and maximize resource productivity in resource-constrained CRSNs. The evaluation metrics are recorded and exported into a newly created dataset named Cognitive Adaptive Metaheuristic Optimization Dataset (CAMOD.csv), which is used to train a machine learning model as Multilayer Perceptron (MLP) neural network—to provide the best prediction of spectrum sensing of secondary users with consideration resource utilization for each cognitive radio sensor node. Results of HPSTSO showed that the average of processing time compared to existing approaches is 30.0675 seconds, packet delivery ratio is 99.08%, channel utilization is 99.166%, probability of channel collision is 0.0508, total network utilization is 7.0604 KB and total resource utilization is 23.0871%. In addition, the MLP accuracy of RadioML2016.10B dataset compared to existing approaches is 98.7%, Precision is 99.98%, Recall is 95.76% and F1 Score is 97.82%. Furthermore, MLP accuracy, precision, recall and F1 score of the CAMOD dataset are 99.45%, 1.0, 98.38%, and 99.18% respectively. Copyright: © 2025 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/).

It tricked users into disclosing sensitive information through web page phishing. A traditional phishing detection method involves analyzing web pages and user behaviour using machine learning models, which require access to raw data, raising privacy concerns. Using homomorphic encryption and multiparty computation, this paper protects privacy in phishing detection. Security and privacy are enhanced by encrypting sensitive user data during the detection process. With this approach, Extreme Learning Machine (ELM) classification is combined with TF-IDF feature selection to detect phishing while maintaining optimal computational efficiency. According to our experimental results, this approach provides an accuracy of 93.5% and has low computational and communication costs, making it a viable solution for secure real-time phishing detection. © 2025, Taru Publications. All rights reserved.

Spam email filtering has recently become the most important task helping in maintaining secure and efficient communication systems. As spam emails lead to security leach, reduced productivity, and increased storage costs, this paper is intended to present a proactive approach to spam email classification, leveraging the advanced techniques to increase detection accuracy and efficiency. The proposed work consists of the three steps. The preprocessing step introduces MinHash which provides a small signature matrix for fast approximation based on a k-shingle technique that generates overlapping sequences of k word, effectively capturing the context and nuances of the spam email text. The second step uses the advanced techniques of machine learning (ML) Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), Multi-Layer Perceptron (MLP), Logistic Regression, and K-Nearest Neighbors (KNN), and Long Short-Term Memory (LSTM) for deep learning (DL) to classify ham and spam emails. The outcomes illustrate that combining the k-shingle, MinHash with advanced text for feeding ML and DL results in high accuracy rate compared with the other works where the SVM classifiers achieves accuracy rate of 98.95% highlighting its effectiveness in distinguishing between ham and spam emails. Other ML shows competitive performance, With MLP 98.25%, RF 95.6%, Logistic regression 98%, DT 93.3%, and lowest accuracy with KNN 70.1%. DL satisfies a high accuracy rate up to 96.1%. This work contributes to the development of a scalable and reliable solution for spam filtering in modern communication systems. ©2025 The authors.

Wireless sensor networks (WSNs) have had many problems up until now because they are open, adaptable, and limited in resources. These problems have included privacy, effectiveness, and consumption of energy. Sensitive information should always be transmitted over wireless networks with extreme caution because public communications on these networks are sometimes unreliable. Although hierarchical routing methods may handle many applications, there are difficult problems with cluster head (CH) selection and network overload distribution. The secure low energy adaptive clustering hierarchy (SLEACH) protocol Cryptographic n-RSA method (SLEACH-n-RSA) is introduced in this work to improve network longevity, reduce energy consumption, and guarantee high security. The initial step of the SLEACH-n-RSA protocol is to use the improved LEACH protocol, which is based on the estimated remaining energy (ERE) and depleted energy (DE) for setting the threshold function value that will decide who will be the CH and how the cluster will form. In the second step, the suggested n-RSA encryption algorithm has been used to ensure the confidentiality of the transmitted data. The performance analysis of the proposed SLEACH-n-RSA protocol shows better performance results when compared with other currently used protocols in terms of network lifetime, packet delivery ratio, energy consumption, and execution time. The experimental results show that the proposed protocol outperforms other existing protocols. © 2024, Penerbit UTHM. All rights reserved.