per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-102327998articleA. FarhadianAlifrd49@gmail.com1Mostafa Bastambastam@umz.ac.ir2E. Ataeiataei@umz.ac.ir3M. Babagolimehdi.babagoli@email.kntu.ac.ir4Faculty of Com. Eng., University of Mazandaran, Babolsar, IranFaculty of Com. Eng., University of Mazandaran, Babolsar, IranFaculty of Com. Eng., University of Mazandaran, Babolsar, IranFaculty of Com. Eng., University of Mazandaran, Babolsar, Iran<p style="direction: ltr;">The expansion of cloud services has highlighted the necessity of virtualization methods for optimal use of hardware resources. While virtual machines were traditionally the main solution for virtualization, the emergence of containers has enabled the elimination of additional operating systems and reduced resource overhead. Technologies such as Docker, Podman, and LXC have gained widespread adoption in this domain. Concurrently, web servers like Nginx and Apache have been optimized for compatibility with these technologies. This paper evaluates the performance of these two web servers across different container platforms under various resource and concurrency conditions. The experiments indicate that the choice of container depends significantly on the web server type and the available resources. In resource-constrained environments, LXC shows better performance for Apache. Conversely, under higher resource availability, Docker yields superior results for running Nginx. The findings of this research can guide better decision-making when selecting the optimal combination of container technology and web server based on infrastructural requirements.</p>http://ijece.org/Article/49036Virtualization containers Docker Apache Nginx LXC.per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-1023299110articleMarzieh sepahvandmarziehsepahvand60@gmail.com1Mojtaba Salehisalehi_mojtaba@ymail.com2Dept.t of Comp. Eng., Khorramabad Branch, Islamic Azad University, Khorramabad, IranDept.t of Comp. Eng., Khorramabad Branch, Islamic Azad University, Khorramabad, Iran<p>The software testing process is very time-consuming and expensive and accounts for almost half of the cost of software production. The main issue in the test data generation process is determining the program's input data in such a way that it meets the specified test criteria. In this research, the structural method has been used to automate the process of generating test data, focusing on the criterion of covering all finite paths. In the structural method, the problem becomes a search problem, and metaheuristic algorithms can be used to solve it. The proposed method is a hybrid algorithm in which the q-learning algorithm is used as a local search method within the structure of the Coati search algorithm. The results of the tests have shown that this method for generating test data is faster than many metaheuristic algorithms and can provide better coverage with fewer evaluations. On average, our proposed algorithm shows about 25-30% improvement in coverage compared to other algorithms, which makes it significantly more effective than other algorithms. This shows that our algorithm achieves superiority over other compared algorithms due to its more efficient and optimal path coverage approach.</p>http://ijece.org/Article/46471software test test data generation structural test metaheuristic algorithms learning Q algorithm. per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-10232111120articleM. R. Sharafi Hoyavadamohamadreza_sharafi@yahoo.com1Mohammad Reza Mollahosseinimr.mollahoseini@iau.ac.ir2Vahid Ayatollahitafti vahid.ayat@gmail.com3Dept. of Comp. Eng., Meybod Branch, Islamic Azad University, Meybod, IranDept. of Comp. Eng., Meybod Branch, Islamic Azad University, Meybod, IranDepartment of Computer, Taft Branch, Islamic Azad University, Taft, Iran<div class="markdown-body antialiased typing-container"> <p>The Internet of Things poses significant challenges in data processing and storage due to the large volume of data generated, including latency, location awareness, and real-time mobility support. Edge computing is recognized as an effective solution to these challenges. This paper examines various secure offloading methods based on collaborative learning in edge computing environments using software-defined networking and analyzes four optimization methods: SDN, SA+GA, OLB-LBMM, and Round-Robin. The main objective of this research is to improve performance and security in the data offloading process while addressing existing challenges. The SDN method provides a flexible framework for managing resources and data in IoT networks, demonstrating better performance than other methods. By reducing latency and optimizing resource allocation, it enhances user satisfaction and increases revenue for cloud service providers. Additionally, the SA+GA and OLB-LBMM algorithms offer improvements in efficiency and security, although they face challenges related to latency and computational complexity. The results indicate that collaborative learning combined with SDN can significantly enhance secure data offloading and enable dynamic network resource management. This research can serve as a foundation for future studies aimed at optimizing data offloading processes in edge computing environments.</p> </div> <div class="chat-index-message-content">&nbsp;</div> <div class="message-metadata-button-container buttons-left-top">&nbsp;</div> <div class="message-metadata-button-container buttons-right-top">&nbsp;</div>http://ijece.org/Article/48486Software-defined network federated learning edge computing Internet of Things.per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-10232121131articleMaral Kolahkajmaralkolahkaj@gmail.com1Marjan Motiee Zadehmiss.motiee@gmail.com2Comp. Eng. Dept., Susangerd Branch, Islamic Azad University, Susangerd, IranComp. Dept., Ahvaz Branch, Islamic Azad University, Ahvaz, Iran<p style="direction: ltr;">Heart disease is one of the leading causes of mortality worldwide, and its early diagnosis is of great importance. Existing feature selection methods for heart disease diagnosis are typically limited to using a single algorithm, which may lead to the selection of redundant features or the omission of important ones, consequently reducing classification accuracy. In this paper, a novel hybrid method for feature selection is proposed, which identifies more efficient and relevant features by employing a soft integration of the results from multiple feature selection algorithms. To enhance the accuracy and speed of diagnosis, an Extreme Learning Machine (ELM) classifier with a wavelet kernel is utilized, where its parameters are optimized using a modified version of the Shuffled Frog-Leaping Algorithm (SFLA). The improved algorithm incorporates a dynamic weighting mechanism and is combined with a Genetic Algorithm (GA), contributing to improved classification accuracy and speed. To demonstrate the robustness and generalizability of the proposed method, it is tested on three well-known UCI datasets. Evaluation results show that the proposed model achieves an accuracy of 93.3%. These findings highlight the high capability and generalization power of the proposed method in heart disease diagnosis.</p>http://ijece.org/Article/33191Feature selection shuffled frog-leaping algorithm heart disease diagnosis ELM classification wavelet kernel.per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-10232132138articleZahra Amirizahra.amiri@mazust.ac.ir1Zahra Mortezaiezm.mortezaie@gmail.com2Department of Computer Engineering, Faculty of Electrical and Computer Engineering, University of Science and Technology of Mazandaran, Behshahr, IranDepartment of Computer Sciences, Faculty of Mathematics and Computer Sciences, Hakim Sabzevari University, Sabzevar, Iran<p style="text-align: left;"><span style="font-family: 'times new roman', times, serif;"><strong>Breast cancer</strong> is considered one of the major concerns in global health, and it is divided into two types: benign and malignant. The malignant type poses a higher risk due to its faster metastasis. Therefore, there is a critical need for fast and accurate detection. Despite the expertise of radiologists, errors due to incorrect interpretation often lead to misdiagnoses. To address this issue, this paper proposes an intelligent system for analyzing mammography images, which includes preprocessing, feature extraction, and classification stages. In this system, the image quality is first improved using preprocessing techniques like Contrast Limited Adaptive Histogram Equalization (CLAHE), and then the region corresponding to the cancerous mass is extracted using Otsu&rsquo;s thresholding segmentation method. Additionally, key features for distinguishing between benign and malignant tumors are extracted using two pre-trained Convolutional Neural Network (CNN) models, namely ResNet50 and InceptionV3. Finally, the extracted features are analyzed using a Support Vector Machine (SVM) classifier to predict the tumor types. The result of this work is an improvement in diagnostic accuracy and early breast cancer detection, which reduces human error and the current challenges in interpreting mammography images</span></p>http://ijece.org/Article/48640Computer Vision Breast Cancer Deep Learning Image Enhancementper Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-10232139145articleK. Mehrgankomail.mehrgan1994@gmail.com1A. Ebrahimi moghadama.ebrahimi@um.ac.ir2M. Khademi Dorohkhademi@um.ac.ir3Dept. of Elec. Eng., Ferdowsi University of Mashhad, Mashhad, IranDept. of Elec. Eng., Ferdowsi University of Mashhad, Mashhad, IranDept. of Elec. Eng., Ferdowsi University of Mashhad, Mashhad, Iran<p style="direction: ltr;">Low-resolution text images often lead to significant errors in Optical Character Recognition (OCR), negatively impacting the performance of automated text recognition systems. Text image super-resolution (SR) is a critical step for improving OCR accuracy, particularly when dealing with inputs of very low resolution. While conventional SR methods succeed in enhancing general image quality, they often struggle to preserve the fine-grained details and structural integrity of characters. In this paper, we propose a novel text super-resolution method that leverages intelligent feedback; by intentionally weakening the OCR loss, our approach imposes stricter reconstruction constraints on the SR network. This unique approach specifically guides the network to generate images that faithfully preserve character structures. The modified loss function compels the SR network to reconstruct fine details lost in the low-resolution input, thereby leading to a significant improvement in downstream OCR accuracy. Experimental results demonstrate that our method not only enhances visual clarity but also boosts the accuracy of subsequent OCR systems by approximately 10% compared to the original low-resolution images. This novel approach represents an effective step toward optimizing the pipeline for text recognition from low-resolution inputs.</p>http://ijece.org/Article/48961Super-resolution text Image recognition intentional loss weakening intelligent feedback.