Intellithings Journal

A Review and Comparative Analysis of Intrusion Detection Systems for Edge Networks in IoT

Ata Amrullah — Fri, 24 Jan 2025 00:00:00 +0700

This article presents a comprehensive review and comparative analysis of intrusion detection systems (IDS) designed for edge networks in Internet of Things (IoT) environments. The rapid growth of IoT has heightened security vulnerabilities in edge networks, which are the focus of this study. Various IDS approaches, including signature-based, anomaly-based, and hybrid methods, are explored, with an emphasis on the application of machine learning and deep learning techniques. The review includes an analysis of system architectures, algorithms (LSTM, CNN, Transformer), datasets, performance evaluation metrics, and experimental results from prior research. The literature study indicates that deep learning has significant potential to enhance intrusion detection accuracy; however, its effectiveness depends on dataset quality, appropriate data preprocessing, and handling class imbalances. Optimal feature selection, blockchain integration, and ensemble approaches are also critical. In conclusion, a multi-faceted approach combining advanced algorithms, suitable preprocessing techniques, and a deep understanding of IoT attacks is essential. Future research should focus on developing adaptive, efficient, and robust IDS with realistic datasets and comprehensive evaluation methods. This article provides a valuable resource for researchers and practitioners in the field of IoT edge network security.

Trends and Challenges in Anomaly Intrusion Detection at the Edge for IoT: A Review

Ata Amrullah, Dicka Yale Kardono, Mohammad Mansyur Abidin — Fri, 24 Jan 2025 00:00:00 +0700

The rapid proliferation of Internet of Things (IoT) devices has brought about new security challenges, particularly in the area of intrusion detection. This review article provides a comprehensive analysis of the trends and challenges in anomaly intrusion detection at the edge for IoT. By synthesizing findings from recent literature (2021-2023), we explore various approaches to anomaly detection, including those based on machine learning (ML), deep learning (DL), statistical methods, and rule-based techniques. We also examine network attacks relevant to IoT, such as man-in-the-middle (MitM), replay, and injection attacks. Our findings reveal a growing trend towards the use of ML and DL for anomaly detection, with many studies focusing on hybrid approaches to improve detection accuracy. While edge computing offers advantages in terms of reduced latency and enhanced privacy, significant challenges remain in implementing anomaly detection on resource-constrained edge devices. These include the heterogeneity of devices and protocols, the increasing sophistication of cyberattacks, the limited availability of labeled data, and privacy concerns. This review identifies unresolved research gaps, including the need for more efficient algorithms, more adaptive approaches, methods for generating synthetic anomaly data, and large-scale implementations. Furthermore, this work discusses the practical implications for enhancing IoT security and provides guidance for researchers and practitioners in the field. We conclude that future efforts should emphasize the development of adaptive and efficient methods, particularly for real-time detection, and consider ethical aspects like data privacy in the deployment of anomaly detection at the edge.

Advanced Sentiment Analysis Using Deep Learning: A Comprehensive Framework for High-Accuracy and Interpretable Models

Ata Amrullah — Fri, 24 Jan 2025 00:00:00 +0700

Sentiment analysis has become a critical tool for understanding public opinion, customer feedback, and social media trends. Despite significant advancements in deep learning, existing models often struggle with accuracy, generalizability, and interpretability, particularly when applied to complex and noisy datasets. In this paper, we propose a novel deep learning framework for sentiment analysis that addresses these limitations by combining the strengths of convolutional neural networks (CNNs) and transformer-based architectures. Our framework leverages verified and high-quality datasets, including Twitter Sentiment140, IMDb movie reviews, and Amazon product reviews, to ensure robustness and reliability. We introduce a hybrid model that integrates multi-head attention mechanisms with hierarchical feature extraction, enabling the model to capture both local and global contextual information effectively. Additionally, we employ state-of-the-art interpretability techniques, such as SHAP and LIME, to provide transparent and human-understandable explanations for model predictions. Experimental results demonstrate that our framework achieves superior performance compared to existing state-of-the-art methods, with an accuracy of 94.3%, an F1-score of 93.8%, and an AUC-ROC score of 97.2%. Furthermore, our model's interpretability features offer valuable insights into decision-making processes, making it highly applicable for real-world applications such as brand monitoring, market analysis, and political sentiment tracking. This study not only advances the field of sentiment analysis but also provides a scalable and interpretable solution for future research in natural language processing.

Internet Network Analysis on Local Provider X Using QoS Method

Mohammad Mansyur Abidin — Fri, 24 Jan 2025 00:00:00 +0700

Local Provider X is an internet service provider at the village level with a focus on supporting household internet needs with 2-5 device connections per home. There are At Least 55 home connections in this provider. With the increasing number of users, the service quality of the Local Provider X internet network needs to be tested to maintain service quality. The testing was carried out using the the Quality of Service (QoS) method with the Action Research approach. Four main parameters were tested, namely throughput, packet loss, delay, and jitter. The assessment used was the TIPHON standardization developed by the European Telecommunications Standards Institute (ETSI). The test results revealed that the throughput value was in the range of 395.60 - 755.42, which had an average index of 2 with the "Average" category. The packet loss value was in the range of 0.01 - 1.89 with an index of 4 and the "Very good" category. The delay value was in the range of 9.64 - 20.70, which had an index of 4 and the "Very Good" category. The jitter value is in the range of 12.04 - 21.94, which has an index of 3 and the category "Good.". Based on the evaluation of all parameters, the overall QoS Index for Local Provider X’s internet network was calculated at 81.25%.

Sentiment Analysis in the Age of Transformers and Large Language Models: A Comprehensive Review of Recent Advances and Future

Ata Amrullah — Fri, 24 Jan 2025 00:00:00 +0700

Sentiment analysis (SA) has undergone a significant transformation with the emergence of Transformer-based models and Large Language Models (LLMs). This review provides a comprehensive overview of recent advances in Transformer-based SA, highlighting the impact of LLMs on accuracy, nuance, and context-awareness. Architectural innovations, domain adaptation techniques, and methods for handling context and long-range dependencies are explored. The review also addresses the critical challenges and limitations associated with LLMs in SA, including bias and fairness, interpretability and explainability, data scarcity, computational cost, and robustness against adversarial attacks. Mitigation strategies and best practices are examined, focusing on data augmentation, adversarial training, bias-aware training objectives, attention visualization, and model distillation. Finally, the review outlines future research directions, emphasizing multimodal sentiment analysis, explainable AI, ethical considerations, low-resource languages, and domain-specific applications. This work concludes that while LLMs offer unprecedented opportunities for advancing SA, addressing the identified challenges is crucial for ensuring responsible and effective deployment in real-world scenarios.