مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Robust Hybrid Deep Learning for IoT Unknown Intrusion Detection Under Data ScarcityRobust Hybrid Deep Learning for IoT Unknown Intrusion Detection Under Data Scarcity115enAliMaroosiDepartment of Computer Engineering, University of Torbat Heydarieh, Torbat Heydarieh, IranAmir Hossein HojatiniaDepartment of Computer Engineering, University of Torbat Heydarieh, Torbat Heydarieh, Iran ArashDeldariUniversity: University of Torbat Heydarieh202585The rapid expansion of the Internet of Things (IoT) has significantly heightened the need for robust intrusion detection systems capable of identifying previously unseen cyber threats. Traditional approaches often struggle with novel attack patterns, leading to decreased detection rates and increased system vulnerability. To address this critical limitation, we propose an innovative and highly effective framework that combines multi-source transfer learning with autoencoders to detect unlabeled and unknown attack types with exceptional precision. Unlike prior methods that rely on single-source transfer learning or basic feature fusion techniques, our advanced approach introduces two groundbreaking techniques: the Concurrent Feature Fusion Model (CoFFM) and the Cascading Feature Fusion Model (CaFFM). These models, along with an enhanced Unified Feature Fusion Model (UFFM), leverage autoencoders to significantly improve adaptability across diverse feature domains, ensuring superior performance in dynamic threat environments. Experimental results on benchmark datasets demonstrate that CoFFM achieves an outstanding accuracy rate of 98.13%, surpassing both non-transfer learning methods (92%) and the best single-source transfer learning approaches (94%). Furthermore, CoFFM exhibits remarkable efficiency under challenging conditions, achieving a substantial 12.24% performance gain over baseline methods even when trained on only 10% of the available data through random sampling. This highlights the model's exceptional robustness in data-scarce scenarios, making it a highly reliable solution for real-world IoT security applications. The success of our framework underscores the potential of multi-source transfer learning combined with autoencoder-based feature fusion in advancing the field of intrusion detection.

http://jist.ir/en/Article/Download/51082

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Re-CAC: A Re-Engineered Call Admission Control for LTE Downlink Networks Using Stepwise Bandwidth Degradation ConceptRe-CAC: A Re-Engineered Call Admission Control for LTE Downlink Networks Using Stepwise Bandwidth Degradation Concept1627enVitalisOnyekeUniversity of Nigeria, Nsukka, Enugu State, NigeriaUdora Nwawelu University of Nigeria, Nsukka, Enugu State, Nigeria0000-0002-1615-4122Bonaventure EkengwuUniversity of Nigeria, Nsukka, Enugu State, NigeriaNnaemekaAsiegbuUniversity of Nigeria, Nsukka, Enugu State, NigeriaBenjaminEzurikeAlex Ekwueme Federal University, Ndufu-Alike, Ebonyi State, Nigeria0000-0002-1311-5355DumtochukwuOyekaUniversity of Nigeria, Nsukka, Enugu State, NigeriaChukwudi ChukwudozieUniversity of Nigeria, Nsukka, Enugu State, NigeriaChimdaluOkideUniversity of Nigeria, Nsukka, Enugu State, Nigeria2024924<a name="_Hlk176418633"></a><span style="font-size: 10.0pt; font-family: 'Times New Roman',serif; mso-fareast-font-family: 'MS Mincho'; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA; mso-bidi-font-weight: bold;">This work presents a Re-engineered Call Admission Control (Re-CAC) scheme for long term evolution (LTE) downlink networks. The scheme is timely as communication networks become increasingly heterogeneous with ever increasing number of subscribers with different Quality of Service (QoS) requests. Bandwidth degradation is an effective concept that some Call Admission Control (CAC) schemes have employed to provide an improved QoS to the admitted RT calls. However, it has led to noticeable resource wastage due to inappropriate degradation method employed. As promising panacea, stepwise bandwidth degradation is employed in this work. This contribution allows sequential bandwidth degradation in stepwise manner. The work demonstrated through extensive simulations in MATLAB the effectiveness of the proposed concept on the basis of throughput, call blocking probability (CBP), call dropping probability (CDP), and spectral efficiency metrics. <span style="font-size: 10.0pt; font-family: 'Times New Roman',serif; mso-fareast-font-family: 'MS Mincho'; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA; mso-bidi-font-weight: bold;">The results show that Re-CAC scheme achieved an average throughput of 0.1657 Mbps and 0.0932 Mbps of RT and NRT calls, respectively; 0.0837 and 0.0650 CBP of RT and NRT calls, correspondingly; respective 0.0733 and 0.0763 CDP of RT and NRT calls; and spectral efficiency of 0.0331 bps/Hz and 0.0191 bps/Hz of RT and NRT calls, respectively. <a name="_Hlk176419990"></a>The Re-CAC scheme is benchmarked with quality of service-aware CAC (QA-CAC), adaptive CAC (ACAC), and enhanced adaptive CAC (EA-CAC) schemes. The superiority of Re-CAC scheme over the benchmark CAC schemes in handling RT services is demonstrated and this was achieved without sacrificing the performance of NRT calls

http://jist.ir/en/Article/Download/48095

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Modeling SOLOMO Marketing Based on Technological Development in the Tourism IndustryModeling SOLOMO Marketing Based on Technological Development in the Tourism Industry2839enMeysamBayatDepartment of Management, Ro.C., Islamic Azad University, Roudhen, Iran.Elham Fazeli VeisariTonekabon branch, Islamic Azad University, Tonekabon, IranMohammad JavadTaghipourianDepartment of Management, Cha.C., Islamic Azad University, Chalus, Iran202525The objective of this research is to identify the components and develop the internal relationships of the SOLOMO marketing components in the tourism industry. The statistical population of the research in the qualitative part includes 15 experts in the field of tourism, using the purposive sampling method, and in the quantitative part includes 420 active participants in the tourism sector. In the qualitative part, the techniques of word association, sentence completion, and dream exercises were used in the form of in-depth interviews and using the MAXQDA software, and in the quantitative part, confirmatory factor analysis was used to verify the validity of the constructs and the PLS software. The findings in the qualitative part, based on the analysis performed in the open coding stage, identified 145 codes, and then 18 core codes and finally 3 selective codes (social media marketing, mobile marketing, and local marketing) were categorized. Also, in the quantitative part of the research, for the overall model fit, the GOF criterion was used, which resulted in a value of 0.830, indicating a strong overall fit of the model in the present research. The results of this research have provided the tourism sector with the opportunity to better access, interact with, and analyze customers, and have improved marketing management in this industry. Especially for the dissemination of attractive and valuable content, it can help attract and convert potential customers.

http://jist.ir/en/Article/Download/49419

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Explainable AI for Enhanced Anomaly Detection in Fraud DetectionExplainable AI for Enhanced Anomaly Detection in Fraud Detection4049enReza Amiriجهاد دانشگاهیMohammad HadiZahediدانشگاه صنعتی خواجه نصیرالدین طوسی MehdiAzadimotlaghدانشگاه خلیج فارس2025914Abstract: The application of machine learning has become indispensable in the critical domain of financial fraud detection. However, a major limitation of traditional models is their "black box" nature, which obscures the reasoning behind a flagged transaction. This lack of transparency often leads to many false positives, which can undermine customer trust and incur substantial operational expenses. To address this challenge, this paper proposes a novel framework for Explainable Anomaly Detection in financial fraud, using advanced Explainable AI (XAI) techniques to provide clear insights into the model's predictive processes. Our approach is designed to move beyond a simplistic binary output of "fraud/no fraud." Our framework combines advanced anomaly detection models, like Isolation Forests and Deep Auto-encoders with model-agnostic explanation methods such as SHAP and LIME, to clearly show which features contribute to a transaction’s anomaly score. The efficacy of our framework has been evaluated using a financial transaction benchmark dataset. The results show that integrating XAI not only makes the system more transparent and trustworthy, but also improves the efficiency of fraud investigations. Based on these results, our method reduces the time and resources needed for manual reviews, while still maintaining high accuracy in detecting fraudulent activities.

http://jist.ir/en/Article/Download/51448

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629KSDB: Improving cloud database security by using searchable encrypted dataKSDB: Improving cloud database security by using searchable encrypted data5058enDavudMohammadpurComputer Département, University of Zanjan Zanjan, Iranhttps://orcid.org/0000-0002-1207-5830MahmoodKhoeiniComputer Département, University of Zanjan Zanjan, Iran2025719Data encryption is a highly effective means of ensuring data security. It transforms readable data into a ciphertext format using cryptographic algorithms and keys. However, the challenge arises when performing query operations on encrypted data due to the alteration of the data structure. This article introduces an improved method that facilitates encryption and query operations on encrypted cloud data without requiring decryption. By leveraging reverse indexing, information mapping, and secret sharing across multiple servers, the proposed method KSDB guarantees data security and prevents data disclosure during both the encryption and query execution processes. The KSDB is an application-level encryption technique that the encrypted data is stored in the cloud storage. While existing methods primarily concentrate on numerical data, this study places emphasis on maintaining the confidentiality of string data, enabling search operations on partial strings without decryption. The results and evaluations demonstrate a significant reduction in memory consumption achieved by the proposed method. In KSDB all implementations have been migrated to a dedicated private server. This secure and reliable entity is responsible for managing critical data, including encryption keys. This strategic decision effectively resolves security issues present in pervious methods and facilitates encryption and decryption processes. Furthermore, it not only addresses concerns regarding information leakage but also enhances data confidentiality.

http://jist.ir/en/Article/Download/50894

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Distinguishing Human from Bot Texts: A Graph-Based and Few-Shot Learning Approach5966enOhood Al MinshidawiComputer engineering department, college of Alborz, University of Tehran, Tehran, IranAbdol-HosseinVahabieTehran Universityhttps://orcid.org/0000-0003-1603-88662025317Bots are automated programs created to carry out specific tasks on the internet. On social networks, bots frequently disseminate automated or misleading content, posing significant challenges to the integrity and reliability of these platforms. Detecting and mitigating bot activity is crucial for maintaining the trustworthiness of social media environments. This task becomes even more challenging in low-resource languages like Arabic, where intricate linguistic structures and limited annotated datasets complicate accurate classification. In this regard, a novel framework is introduced for distinguishing between human- and bot-generated Arabic text, using the AutoTweet-Dataset. The framework evaluates two categories of advanced models: graph neural networks and the SetFit model. The first category encompasses two distinct architectures: graph attention networks and graph convolutional networks. In contrast, the SetFit model leverages few-shot learning to facilitate efficient classification. Besides creating an advanced model for identifying bot-generated text, our primary objective is to compare graph neural network-based models with the SetFit model in addressing the complexities of Arabic text processing. The evaluation results determine that the SetFit model achieved the highest accuracy at 88.35%, demonstrating its effectiveness in differentiating between text generated by humans and bots. This research represents a significant advancement in bot detection techniques for low-resource languages. Introducing scalable and efficient methodologies enhances the accuracy of automated content detection, contributing to the security and authenticity of social media interactions in the face of increasingly advanced bot activity.

http://jist.ir/en/Article/Download/49750

مرکز منطقه ای اطلاع رسانی علوم و فناوریJournal of Information Systems and Telecommunication (JIST) 2322-143714532026629Image-Based Phishing URL Classification Using Convolutional Neural NetworksImage-Based Phishing URL Classification Using Convolutional Neural Networks6778en GholamrezaAhmadihttps://orcid.org/0000000243102053HamedMonkaresiدانشگاه رازیhttps://orcid.org/000000030747962720251122Phishing attacks continue to pose a significant threat to online security, with attackers increasingly leveraging deceptive URLs to steal sensitive information. Traditional phishing detection methods often rely on URL analysis or manual feature extraction, which can be time-consuming and less effective against evolving attack techniques. To address these limitations, more adaptive and intelligent detection mechanisms are increasingly required to keep pace with modern attack strategies. In this paper, we propose an image-based approach for phishing URL classification using Convolutional Neural Networks (CNNs). By transforming URLs into visual representations based on their features, we leverage the power of deep learning to automatically extract discriminative features for classification. We conduct a comprehensive comparison of various deep learning models, including different CNN architectures (both basic and pre-trained/fine-tuned), to evaluate their performance in terms of accuracy, computational efficiency, and training time. Our experiments demonstrate that image-based classification using CNNs achieves competitive accuracy while offering potential robustness against adversarial variations in phishing URLs. Additionally, we analyze the trade-offs between model complexity and inference time, providing insights into the practical deployment of such systems. The results highlight the potential of image-based deep learning models as an effective tool for phishing detection, paving the way for further research in this domain.

http://jist.ir/en/Article/Download/52165